Ben Callahan (00:00)
Hi everyone, I'm your host, Ben Callahan, founder of Sparkbox and Redwoods Design System Community. I wanna welcome you to part two of Episode 68 of The Question, where my co-host TJ Pitre and I facilitate a deep dive on the topic of design systems as AI context. If you have a moment, please get subscribed to the show out at bencallahan.com/thequestion Let's get into it.
ep068 p2 deep dive screenshare (00:25)
TJ, give us like 30 seconds on who you are. Tell us a little bit about your company too, because I feel like we didn't talk a lot about Southleft last time. Yeah, sure. So my name is TJ Petri. I own a company called Southleft. We are primarily a front end design development agency focusing on the intersection of AI intelligence and design systems. We've always traditionally been
Ben Callahan (00:25)
TJ, give us like 30 seconds on who you are. Tell us a little bit about your company too, because I feel like we didn't talk a lot about Southleft last time. Yeah, sure. So my name is TJ Petri. I own a company called Southleft. We are primarily a front end design development agency focusing on the intersection of AI intelligence and design systems. We've always traditionally been
ep068 p2 deep dive group (00:25)
TJ, give us like 30 seconds on who you are. Tell us a little bit about your company too, because I feel like we didn't talk a lot about Southleft last time. Yeah, sure. So my name is TJ Petri. I own a company called Southleft. We are primarily a front end design development agency focusing on the intersection of AI intelligence and design systems. We've always traditionally been
ep068 p2 deep dive screenshare (00:47)
design systems practitioners, both from just the consulting side, but also of like team augmentation to help design engineer design systems and then build out and then execute on the products that the design systems are helping. So we're also a team of product engineers. ⁓ We are somewhere between 20 and 24, depending on how many subcontractors we have at a time. And our primary work is
Ben Callahan (00:47)
design systems practitioners, both from just the consulting side, but also of like team augmentation to help design engineer design systems and then build out and then execute on the products that the design systems are helping. So we're also a team of product engineers. ⁓ We are somewhere between 20 and 24, depending on how many subcontractors we have at a time. And ⁓
ep068 p2 deep dive group (00:47)
design systems practitioners, both from just the consulting side, but also of like team augmentation to help design engineer design systems and then build out and then execute on the products that the design systems are helping. So we're also a team of product engineers. ⁓ We are somewhere between 20 and 24, depending on how many subcontractors we have at a time. And ⁓
Ben Callahan (01:13)
Our primary work is
ep068 p2 deep dive group (01:13)
Our primary work is
Ben Callahan (01:15)
senior level product engineering and design systems projects. I personally do a lot of consulting. We have a couple of ⁓ team members that do some consulting as well. And then we do a lot of team augmentation for usually enterprise level organizations that are looking for help with ⁓ infusing AI workflows into their existing design systems.
ep068 p2 deep dive group (01:15)
senior level product engineering and design systems projects. I personally do a lot of consulting. We have a couple of ⁓ team members that do some consulting as well. And then we do a lot of team augmentation for usually enterprise level organizations that are looking for help with ⁓ infusing AI workflows into their existing design systems.
ep068 p2 deep dive screenshare (01:15)
senior level product engineering and design systems projects. I personally do a lot of consulting. We have a couple of team members that do some consulting as well. And then we do a lot of team augmentation for usually enterprise level organizations that are looking for help with infusing AI workflows into their existing design systems,
⁓ teams and how they're working. ⁓
Ben Callahan (01:40)
⁓ teams and how they're working.
ep068 p2 deep dive group (01:40)
⁓ teams and how they're working.
But like I said, we also do a lot of ⁓ from the ground up working with teams to help build their design systems. It's been a fun ride. And TJ, you've been prolific in sharing a lot of your learning. This is the thing I love about what you're doing. It sort of aligns so perfectly with what we're trying to do on the question and over in Redwoods, just like being generous with our knowledge. And a lot of the folks on this call are the same, right?
Ben Callahan (01:44)
But like I said, we also do a lot of ⁓ from the ground up working with teams to help build their design systems. It's been a fun ride. And TJ, you've been prolific in sharing a lot of your learning. This is the thing I love about what you're doing. It sort of aligns so perfectly with what we're trying to do on the question and over in Redwoods, just like being generous with our knowledge. And a lot of the folks on this call are the same, right?
ep068 p2 deep dive screenshare (01:44)
But like I said, we also do a lot of ⁓ from the ground up, working with teams to help build their design systems. It's been a fun ride. And TJ, you've been prolific in sharing a lot of your learning. This is the thing I love about what you're doing. It sort of aligns so perfectly with what we're trying to do on the question and over in Redwoods, just like being generous with our knowledge. And a lot of the folks on this call are the same, right?
If you look here quickly, just along the bottom, I dropped in links of folks from Redwoods, this white little section over here. Look at all this stuff in the last couple of weeks that this community has created. So people are being so generous with their knowledge, and it is incredibly, incredibly helpful. ⁓ So please check all that stuff out. ⁓ It's really cool to see that happening. So thanks. And if you have other stuff to share, drop your community announcements down here.
Ben Callahan (02:08)
If you look here quickly, just along the bottom, I dropped in links of folks from Redwoods, this white little section over here. Look at all this stuff in the last couple of weeks that this community has created. So people are being so generous with their knowledge, and it is incredibly, incredibly helpful. ⁓ So please check all that stuff out. ⁓ It's really cool to see that happening. So thanks. And if you have other stuff to share, drop your community announcements down here.
ep068 p2 deep dive group (02:08)
If you look here quickly, just along the bottom, I dropped in links of folks from Redwoods, this white little section over here. Look at all this stuff in the last couple of weeks that this community has created. So people are being so generous with their knowledge, and it is incredibly, incredibly helpful. ⁓ So please check all that stuff out. It's really cool to see that happening. So thanks. And if you have other stuff to share, drop your community announcements down here.
ep068 p2 deep dive screenshare (02:36)
If you have something to succeed, that's a success that we want to hear the good news to tell us something you've succeeded in down here. If you have feedback for the format or if you have ideas for future questions, drop them. And just so you know, like this is a pretty open format. That means the zoom chat is wide open. You can all turn your cameras on. You can unmute. You can raise your hand. The fig jam is open. You can get in there. You can ⁓ take notes, help us take notes and all of that for that to work. We have to take our code of conduct very seriously.
Ben Callahan (02:36)
If you have something to succeed, that's a success that we want to hear the good news to tell us something you've succeeded in down here. If you have feedback for the format or if you have ideas for future questions, drop them. And just so you know, like this is a pretty open format. That means the zoom chat is wide open. You can all turn your cameras on. You can unmute. You can raise your hand. The fig jam is open. You can get in there. You can ⁓ take notes, help us take notes and all of that for that to work. We have to take our code of conduct very seriously.
ep068 p2 deep dive group (02:36)
If you have something to succeed, that's a success that we want to hear the good news to tell us something you've succeeded in down here. If you have feedback for the format or if you have ideas for future questions, drop them. And just so you know, like this is a pretty open format. That means the zoom chat is wide open. You can all turn your cameras on. You can unmute. You can raise your hand. The fig jam is open. You can get in there. You can ⁓ take notes, help us take notes and all of that for that to work. We have to take our code of conduct very seriously.
ep068 p2 deep dive screenshare (03:03)
⁓ And so that means we just need to be very careful with the space. ⁓ If you have any doubt about whether it's an appropriate comment, please opt on the side of caution there. We just want to be really careful with the space. I want everybody here to know they are very welcome. Everybody's experience here is really valuable. What we're trying to do is broaden each other's perspectives. That's the goal here today. So ⁓ thanks for being here. Thanks for that. The way I think about this is just
Ben Callahan (03:03)
⁓ And so that means we just need to be very careful with the space. ⁓ If you have any doubt about whether it's an appropriate comment, please opt on the side of caution there. We just want to be really careful with the space. I want everybody here to know they are very welcome. Everybody's experience here is really valuable. What we're trying to do is broaden each other's perspectives. That's the goal here today. So thanks for being here. Thanks for that. The way I think about this is just
ep068 p2 deep dive group (03:03)
⁓ And so that means we just need to be very careful with the space. ⁓ If you have any doubt about whether it's an appropriate comment, please opt on the side of caution there. We just want to be really careful with the space. I want everybody here to know they are very welcome. Everybody's experience here is really valuable. What we're trying to do is broaden each other's perspectives. That's the goal here today. So thanks for being here. Thanks for that. The way I think about this is just
ep068 p2 deep dive screenshare (03:31)
I want to be in learning mode. If I come to this conversation believing everybody here has something to teach me, we're going to be in a great place if we all do that. let's try to come with that attitude. Thank you so much. Every single Episode we do this, you all truly do it. And it just makes me so happy to know that we can have a space like this that is pretty open and people are respectful and treat it with care like you do. So thank you for that. ⁓ If you haven't yet, there's a space up at the top. I left the stuff from last week in here.
Ben Callahan (03:31)
I want to be in learning mode. If I come to this conversation believing everybody here has something to teach me, we're going to be in a great place if we all do that. So let's try to come with that attitude. Thank you so much. Every single Episode we do this, you all truly do it. And it just makes me so happy to know that we can have a space like this that is pretty open and people are respectful and treat it with care like you do. So thank you for that. ⁓ If you haven't yet, there's a space up at the top. I left the stuff from last week in here.
ep068 p2 deep dive group (03:31)
I want to be in learning mode. If I come to this conversation believing everybody here has something to teach me, we're going to be in a great place if we all do that. So let's try to come with that attitude. Thank you so much. Every single Episode we do this, you all truly do it. And it just makes me so happy to know that we can have a space like this that is pretty open and people are respectful and treat it with care like you do. So thank you for that. ⁓ If you haven't yet, there's a space up at the top. I left the stuff from last week in here.
If you came last week and you already added your info, no worries. But drop your bio up above. I'd love to just see who's here and get connected with you on LinkedIn and all that. ⁓ Goodness, let's dive in. ⁓ TJ, I thought, so we were talking a little bit before the session started. And since last week, there've been some interesting announcements in the space. Can you give us just like a high level, like what's happened in the space? Yeah, sure. ⁓ One of the big things that we've
ep068 p2 deep dive screenshare (04:00)
If you came last week and you already added your info, no worries. But drop your bio up above. I'd love to just see who's here and get connected with you on LinkedIn and all that. ⁓ Goodness, let's dive in. ⁓ TJ, I thought, so we were talking a little bit before the session started. And since last week, there've been some interesting announcements in the space. Can you give us just like a high level, like what's happened in the space? Yeah, sure. ⁓ One of the big things that we've
Ben Callahan (04:00)
If you came last week and you already added your info, no worries. But drop your bio up above. I'd love to just see who's here and get connected with you on LinkedIn and all that. ⁓ Goodness, let's dive in. ⁓ TJ, I thought, so we were talking a little bit before the session started. And since last week, there've been some interesting announcements in the space. Can you give us just like a high level, like what's happened in the space? Yeah, sure. ⁓ One of the big things that we've
ep068 p2 deep dive screenshare (04:29)
One of the main tools that we use internally, and I just posted the resources for him right there, is something that we've been using since probably July or August of last year, but then kind of turned it into something that's more public use, a tool that's a little bit more. Before, a lot of the tools that we create internally are usually specifically made for the projects and clients that we're working on. And a lot of those tools end up being kind of unique to certain code bases or certain.
Ben Callahan (04:29)
One of the main tools that we use internally, and I just posted the resources for him right there, is something that we've been using since probably July or August of last year, but then kind of turned it into something that's more public use, a tool that's a little bit more. Before, a lot of the tools that we create internally are usually specifically made for the projects and clients that we're working on. And a lot of those tools end up being kind of unique to certain code bases or certain.
ep068 p2 deep dive group (04:29)
One of the main tools that we use internally, and I just posted the resources for him right there, is something that we've been using since probably July or August of last year, but then kind of turned it into something that's more public use, a tool that's a little bit more, before, a lot of the tools that we create internally are usually specifically made for the projects and clients that we're working on. And a lot of those tools end up being kind of unique to certain code bases or certain.
ep068 p2 deep dive screenshare (04:58)
⁓ design systems. So we use it first and then we find the value or whether or not it's worth continuing to use in other projects. So we might kind of reverse engineer it to make it recognizable for other patterns that we see in other projects that we're working on. But then if it really sticks, feels like something that we can contribute back to the community. This is kind of what we saying earlier about like... ⁓
Ben Callahan (04:58)
⁓ design systems. So we use it first and then we find the value or whether or not it's worth continuing to use in other projects. So we might kind of reverse engineer it to make it recognizable for other patterns that we see in other projects that we're working on. But then if it really sticks, feels like something that we can contribute back to the community. This is kind of what we saying earlier about like...
ep068 p2 deep dive group (04:58)
⁓ design systems. So we use it first and then we find the value or whether or not it's worth continuing to use in other projects. So we might kind of reverse engineer it to make it recognizable for other patterns that we see in other projects that we're working on. But then if it really sticks, feels like something that we can contribute back to the community. This is kind of what we saying earlier about like...
ep068 p2 deep dive screenshare (05:24)
I wouldn't have gotten where I'm at if it wasn't for open source technology like years ago. And now it feels like I'm in this position to be able to give back. And so it feels like our responsibility to be able to give this kind of stuff back. so the Figma Console MCP was one that we had created that a lot of the other tools that we put out there as well as kind of based on. like Figma Lint has ⁓ some of the Figma Console MCP tools built into it.
Ben Callahan (05:24)
I wouldn't have gotten where I'm at if it wasn't for open source technology like years ago. And now it feels like I'm in this position to be able to give back. And so it feels like our responsibility to be able to give this kind of stuff back. And so the Figma Console MCP was one that we had created that a lot of the other tools that we put out there as well, it's kind of based on. like Figma Lint has ⁓ some of the Figma Console MCP tools built into it.
ep068 p2 deep dive group (05:24)
I wouldn't have gotten where I'm at if it wasn't for open source technology like years ago. And now it feels like I'm in this position to be able to give back. And so it feels like our responsibility to be able to give this kind of stuff back. And so the Figma Console MCP was one that we had created that a lot of the other tools that we put out there as well as kind of based on. like Figma Lint has ⁓ some of the Figma Console MCP tools built into it.
ep068 p2 deep dive screenshare (05:51)
our design systems assistant MCP, a lot of those kind of like stem from the idea of what the Figma console MCP is. And we announced it back in September of last year and it got mild traction. there was like a whole five part series that I put out on it and it was like, okay, people saw it they were like, all right, maybe I was appealing to a different crowd. But it wasn't until January of this year where I was demoing something and then as like a
Ben Callahan (05:51)
our design systems assistant MCP, a lot of those kind of like stem from the idea of what the Figma console MCP is. And we announced it back in September of last year and it got mild traction. there was like a whole five part series that I put out on it and it was like, okay, people saw it they were like, all right, maybe I was appealing to a different crowd. But it wasn't until January of this year where I was demoing something and then as like a
ep068 p2 deep dive group (05:51)
our design systems assistant MCP, a lot of those kind of like stem from the idea of what the Figma console MCP is. And we announced it back in September of last year and it got mild traction. there was like a whole five part series that I put out on it and it was like, okay, people saw it they were like, all right, maybe I was appealing to a different crowd. But it wasn't until January of this year where I was demoing something and then as like a
ep068 p2 deep dive screenshare (06:22)
like a small part of the demo, I figured what it was, but it was like, made a variable value update. Like I just kind of wrote back to Figma instead of showing like, cause the most of the tool that we were using it for is for analysis and documentation and doing some, code design parity. And then people really latched onto the right capabilities of the tool. So it was like, all right, well, so maybe I'm, maybe I'm talking about the wrong benefits of this. So I started to like lean more into
Ben Callahan (06:22)
like a small part of the demo, I figured what it was, but it was like, made a variable value update. Like I just kind of wrote back to Figma instead of showing like, cause the most of the tool that we were using it for is for analysis and documentation and ⁓ doing some, code design parity. And then people really latched onto the right capabilities of the tool. So it was like, all right, well, so maybe I'm, maybe I'm talking about the wrong benefits of this. So I started to like lean more into
ep068 p2 deep dive group (06:22)
like a small part of the demo, I figured what it was, but it was like, made a variable value update. Like I just kind of wrote back to Figma instead of showing like, cause the most of the tool that we were using it for is for analysis and documentation and ⁓ doing some, code design parity. And then people really latched onto the right capabilities of the tool. So it was like, all right, well, so maybe I'm talking about the wrong benefits of this. So I started to like lean more into
ep068 p2 deep dive screenshare (06:51)
⁓ discussing the benefits of actually being able to have code parity where you can actually make edits to the developed component because it's great for that because it always did like traditional official MCP, Figma MCP stuff where ⁓ like it would examine the component on the Figma side of things and then create a pretty high fidelity one-to-one match up on the code side of things. But going in the opposite direction was something that was always capable of doing too.
Ben Callahan (06:51)
⁓ discussing the benefits of actually being able to have code parity where you can actually make edits to the developed component because it's great for that because it always did like traditional official MCP, Figma MCP stuff where ⁓ like it would examine the component on the Figma side of things and then create a pretty high fidelity ⁓ one-to-one match up on the code side of things. But going in the opposite direction was something that was always capable of doing too.
ep068 p2 deep dive group (06:51)
⁓ discussing the benefits of actually being able to have code parity where you can actually make edits to the developed component because it's great for that because it always did like traditional official MCP, Figma MCP stuff where ⁓ like it would examine the component on the Figma side of things and then create a pretty high fidelity ⁓ one-to-one match up on the code side of things. But going in the opposite direction was something that was always capable of doing too.
ep068 p2 deep dive screenshare (07:19)
And then so recently, it was Mardi Gras Day. So I was off ⁓ enjoying the festivities and just my phone was blowing up and it was, ⁓ didn't, I ignored it as much as I could because I wanted to enjoy the day. But the next morning I had woke up and I read this announcement that the official Figma MCP added a feature called Claude Code to Figma.
Ben Callahan (07:19)
And then so recently, it was Mardi Gras Day. So I was off ⁓ enjoying the festivities and just my phone was blowing up and it was, ⁓ didn't, I ignored it as much as I could because I wanted to enjoy the day. But the next morning I had woke up and I read this announcement that the official Figma MCP added a feature called Claude Code to Figma.
ep068 p2 deep dive group (07:19)
And then so recently, it was Mardi Gras Day. So I was off ⁓ enjoying the festivities and just my phone was blowing up and it was, ⁓ didn't, I ignored it as much as I could because I wanted to enjoy the day. But the next morning I had woke up and I read this announcement that the official Figma MCP added a feature called Claude Code to Figma.
ep068 p2 deep dive screenshare (07:45)
where you can do the kind of the similar thing that we were doing where you could write back from a ⁓ coded component or coded what seems to be like an HTML structure back into Figma frames as layers within a Figma project. And it was close. ⁓ so like I was like, obviously let me carve out some time so I can jam on it for a little bit and try to understand the value in it. And it's cool.
Ben Callahan (07:45)
where you can do the kind of the similar thing that we were doing where you could write back from a coded component or coded what seems to be like an HTML structure back into Figma frames as layers within a Figma project. And it was close. ⁓ so like I was like, obviously let me carve out some time so I can jam on it for a little bit and try to understand the value in it. And it's cool.
ep068 p2 deep dive group (07:45)
where you can do the kind of the similar thing that we were doing where you could write back from a ⁓ coded component or coded what seems to be like an HTML structure back into Figma frames as layers within a Figma project. And it was close. ⁓ so I was like, obviously, let me carve out some time so I can jam on it for a little bit and try to understand the value in it. And it's cool.
ep068 p2 deep dive screenshare (08:13)
Like I spent about two hours on it. I had a storybook instance where I had some story UI generations that I just generated anything from like little signed up screens all the way to these like giant dashboards. And this was all React based. And I think I was doing it in a ShadCN or a Mantine based design system. And then sure enough, I was adding it back into a Figma project and it worked great. I mean, there was some like
Ben Callahan (08:13)
Like I spent about two hours on it. I had a storybook instance where I had some story UI generations that I just generated anything from like little signed up screens all the way to these like giant dashboards. And this was all React based. And I think I was doing it in a ShadCN or a Mantine based design system. And then sure enough, I was adding it back into a Figma project and it worked great. I mean, there was some like
ep068 p2 deep dive group (08:13)
Like, I spent about two hours on it. I had a storybook instance where I had some story UI generations that I just generated anything from like little signed up screens all the way to these like giant dashboards. And this was all React based. And I think I was doing it in a ShadCN or a Mantine based design system. And then sure enough, I was adding it back into a Figma project and it worked great. I mean, there was some like
ep068 p2 deep dive screenshare (08:42)
some of the larger complex structures, it missed things. But I saw the value in it and I thought it was pretty cool. And then I felt the need because the next morning when I had woke up, I had like over 200 messages throughout social media and emails and slacks and LinkedIn. And I just wrote my thoughts on it. And I think it's great. think there's definitely an audience for it. I think the Figma MCP, the Figma Console MCP,
Ben Callahan (08:42)
some of the larger complex structures, it missed things. But I saw the value in it and I thought it was pretty cool. And then I felt the need because the next morning when I had woke up, I had like over 200 messages throughout social media and emails and slacks and LinkedIn. And I just wrote my thoughts on it. And I think it's great. think there's definitely an audience for it. I think the Figma MCP, the Figma Console MCP,
ep068 p2 deep dive group (08:42)
some of the larger complex structures, it missed things. ⁓ But I saw the value in it and I thought it was pretty cool. And then I felt the need because the next morning when I had woke up, I had like over 200 messages throughout social media and emails and slacks and LinkedIn. And I just wrote my thoughts on it. And I think it's great. think there's definitely an audience for it. I think the Figma MCP, the Figma Console MCP,
ep068 p2 deep dive screenshare (09:07)
serves a different purpose because I feel like it's made more for design systems practitioners because of its ability to recreate components while leveraging the components from your design system, as well as the tokens and architecture and layer naming and all of those things. So while it does add frames, it's also aware of the system that it's rebuilding the coded component into, which is a little bit different from the way that the Claude code to Figma MCP
Ben Callahan (09:07)
serves a different purpose because I feel like it's made more for design systems practitioners because of its ability to recreate components while leveraging the components from your design system, as well as the tokens and architecture and layer naming and all of those things. So while it does add frames, it's also aware of the system that it's rebuilding the coded component into, which is a little bit different from the way that the Claude code to Figma MCP
ep068 p2 deep dive group (09:07)
serves a different purpose because I feel like it's made more for design systems practitioners because of its ability to recreate components while leveraging the components from your design system, as well as the tokens and architecture and layer naming and all of those things. So while it does add frames, it's also aware of the system that it's rebuilding the coded component into, which is a little bit different from the way that the Claude code to Figma MCP
ep068 p2 deep dive screenshare (09:36)
feature does, whereas it just takes the HTML and then creates static components in a frame in a new project on the Figma side of things. It feels like, and this is my opinion, the best value here in product design is here, keep me straight, because I'd love to hear anybody else's opinion on this too. I think as product designers have moved more towards being comfortable,
Ben Callahan (09:36)
feature does, whereas it just takes the HTML and then creates static components in a frame in a new project on the Figma side of things. It feels like, and this is my opinion, the best value here in product design is here, keep me straight, because I'd love to hear anybody else's opinion on this too. I think as product designers have moved more towards being comfortable,
ep068 p2 deep dive group (09:36)
feature does, whereas it just takes the HTML and then creates static components in a frame in a new project on the Figma side of things. It feels like, and this is my opinion, the best value here in product design is here, keep me straight, because I'd love to hear anybody else's opinion on this too. I think as product designers have moved more towards being comfortable,
ep068 p2 deep dive screenshare (10:02)
with getting ideas out into the world of vibe coding. So whether it's lovable, a cursor, or your weapon of choice to be able to quickly get these ideas out past design into functionality, it seems as though like, all right, I got this working app, but I'm not satisfied with this interface. So what I'd like to do is move this interface back into an environment that I'm more comfortable working within like Figma.
Ben Callahan (10:02)
with getting ideas out into the world of vibe coding. So whether it's lovable, a cursor, or your weapon of choice to be able to quickly get these ideas out past design into functionality, it seems as though like, all right, I got this working app, but I'm not satisfied with this interface. So what I'd like to do is move this interface back into an environment that I'm more comfortable working within like Figma.
ep068 p2 deep dive group (10:02)
with getting ideas out into the world of vibe coding. So whether it's lovable, a cursor, or your weapon of choice to be able to quickly get these ideas out past design into functionality, it seems as though like, all right, I got this working app, but I'm not satisfied with this interface. So what I'd like to do is move this interface back into an environment that I'm more comfortable working within like Figma.
ep068 p2 deep dive screenshare (10:30)
and then noodle on it there until you got it into a good spot and then make that round trip and then utilize the Figma MCP back again to push it into the interface where it had the working version of the application. That seems like to me the best example of the purpose of why something like this was done. In addition to like if you're a team of one and you've got somebody else develop the thing and then you want to bring it in and design, they give it back to the person who was developing it.
Ben Callahan (10:30)
and then noodle on it there until you got it into a good spot and then make that round trip and then utilize the Figma MCP back again to push it into the interface where it had the working version of the application. That seems like to me the best example of the purpose of why something like this was.
ep068 p2 deep dive group (10:30)
and then noodle on it there until you got it into a good spot and then make that round trip and then utilize the Figma MCP back again to push it into the interface where it had the working version of the application. That seems like to me the best example of the purpose of why something like this was ⁓ in addition to like if you're a team of one and you've got somebody else develop the thing and then you want to bring it in and design, they give it back to the person who was developing it.
Ben Callahan (10:50)
⁓ In addition to like if you're a team of one and you've got somebody else develop the thing and then you want to bring it in and design, they give it back to the person who was developing it.
That's kind of the obvious one that I think why it was there. But I could see some like latching onto the vibe coded part of this and why people might want to use it. So anyway, that's the whole thing. then the overwhelming ⁓ feedback that I've been getting, it seems like if anything, that announcement pushed
ep068 p2 deep dive screenshare (10:59)
That's kind of the obvious one that I think why it was there. But I could see some like latching onto the vibe coded part of this and why people might want to use it. So anyway, that's the whole thing. And then the overwhelming ⁓ feedback that I've been getting, it seems like if anything, that announcement pushed
ep068 p2 deep dive group (10:59)
That's kind the obvious one that I think why it was there. But I could see some like latching onto the vibe coded part of this and why people might want to use it. So anyway, that's the whole thing. And then the overwhelming ⁓ feedback that I've been getting, it seems like if anything, that announcement pushed
ep068 p2 deep dive screenshare (11:17)
the Figma console MCP into popularity. Because every day it's like these people are creating all these great experiments with them and tagging me in them. So it's been really fun to see unfold.
ep068 p2 deep dive group (11:17)
the Figma console MCP into popularity. Because every day it's like these people are creating all these great experiments with them and tagging me in them. So it's been really fun to see unfold.
Ben Callahan (11:17)
the Figma console MCP into popularity because every day it's like these people are creating all these great experiments with them and tagging me in them. So it's been really fun to see unfold.
Nice. When you experimented with the sort of newer stuff that Figma and Claude are doing, ⁓ the designs that were built, that it created for you in Figma, are those actually connected, like variables bound and all that kind of stuff, like MCPs, like your MCP will do?
ep068 p2 deep dive screenshare (11:28)
Nice. you experimented with the sort of newer stuff that Figma and Claude are doing, the designs that were built, that it created for you in Figma, are those actually connected, like variables bound and all that kind of stuff, like MCPs? No. Okay.
ep068 p2 deep dive group (11:28)
Nice. you experimented with the newer stuff that Figma and Claude are doing, ⁓ the designs that were built, that it created for you in Figma, are those actually connected, like variables bound and all that kind of stuff, like MCPs?
Ben Callahan (11:46)
It was hard-coded values. It was really accurate and it was fast, which is great too.
ep068 p2 deep dive screenshare (11:47)
No, it was hard to go to values. But it was really accurate and it was fast.
ep068 p2 deep dive group (11:47)
No. It was hard to go to values. But it was really accurate, and it was fast, which is great too.
ep068 p2 deep dive screenshare (11:52)
which is great too.
And it had some options for you. So like you could select a frame if you wanted so you can grab like specific parts of the markup, the specific parts of the page, the HTML page that you wanted to bring back in. So that was pretty cool too. But so I was truly impressed with like how well it worked. But it was more of just like a, it's like a prototype thing that it puts together for you. That's great. Yeah. So a lot's happening, right? And this stuff is like...
ep068 p2 deep dive group (11:54)
and it had some options for you. like you could select a frame if you wanted so you can grab like specific parts of the markup, the specific parts of the page, the HTML page that you wanted to bring back in. So that was pretty cool too. But so I was truly impressed with like how well it worked. But it was more of just like a, it's like a prototype thing that it puts together for you. That's great. Yeah. So a lot's happening, right? And this stuff is like, this
Ben Callahan (11:54)
It had some options for you. You could select a frame if you wanted so you can grab specific parts of the markup, the specific parts of the page, the HTML page that you wanted to bring back in. That was pretty cool too. I was truly impressed with how well it worked. But it was more of just like a.
It's like a prototype thing that it puts together for you. That's great. Yeah. So a lot's happening, right? And this stuff is like
this is what happened between last conversation that we had one week ago and today. Right. So as you know, the space is constantly just in flux. and that's, you know, that's part of the reason we're trying to do this. So that this whole idea for me and I want to get everybody here like open and into the conversation. So raise your hands if you want to jump in. I'd love us to talk for just a brief moment about
ep068 p2 deep dive screenshare (12:21)
This is what happened between last conversation that we had one week ago and today, right? So as you know, the space is constantly just in flux and ⁓ that's part of the reason we're trying to do this. this whole idea for me, and I want to get everybody here like open and into the conversation. So raise your hands if you want to jump in. I'd love us to talk for just a brief moment about
ep068 p2 deep dive group (12:21)
is what happened between last conversation
that we had one week ago and today, right? So as you know, this space is constantly just in flux and ⁓ that's part of the reason we're trying to do this. So this whole idea for me, and I wanna get everybody here like open and into the conversation. So raise your hands if you wanna jump in. I'd love us to talk for just a brief moment about
like where actually is the source of truth in this kind of world, right? Like we're talking about ease of flexibility of moving this stuff all around.
ep068 p2 deep dive screenshare (12:44)
like where actually is the source of truth in this kind of world, right? Like we're talking about.
Ben Callahan (12:44)
Like where actually is the source of truth in this kind of world? Right? Like we're talking about ease of flexibility of moving this stuff all around.
ep068 p2 deep dive screenshare (12:50)
ease of flexibility of moving this stuff all around.
you know, even if the quality of that move from code to design or design to code is not 100 % yet, it's definitely getting better with every iteration of these tools. So like in my mind, that means there's a future state where these things are very fluid. And so in that world, what is the source of truth? And so I just want to open that up for folks. Tell me where is your source of truth now? Do you have one? Do you have
Ben Callahan (12:53)
⁓ And, you know, even if the quality of that move from code to design or design to code is, is not a hundred percent yet, it's definitely getting better with every iteration of these tools. So like in my mind, that means there's a future state where these things are.
ep068 p2 deep dive group (12:53)
⁓ And, you know, even if the quality of that move from code to design or design to code is not 100 % yet, it's definitely getting better with every iteration of these tools. So like in my mind, that means there's a future state where these things are very fluid. And so in that world, what is the source of truth? And so I just want to open that up for folks. Tell me where is your source of truth now? Do you have one? Do you have
Ben Callahan (13:10)
very fluid. And so in that world, what is the source of truth? And so I just want to open that up for folks. Tell me, where is your source of truth now? Do you have one? Do you have
ep068 p2 deep dive group (13:19)
more than one?
Ben Callahan (13:19)
more than one? What is that like? ⁓ And how is that changing as you see these tools evolve? And I don't know, TJ, do you want to start while we let folks kind of get the courage to raise a hand and jump in? Yeah. Yeah. I'll say quick what I think and then I'll leave the floor to some people who are already raising their hands.
ep068 p2 deep dive screenshare (13:19)
more than one? What is that like? And how is that changing as you see these tools evolve? And I don't know, TJ, do you want to start while we let folks kind of get the courage to raise a hand and jump in? Yeah. Yeah. I'll say quick what I think, and then I'll leave the floor to some people who are already raising their hands.
ep068 p2 deep dive group (13:21)
What is that like? And how is that changing as you see these tools evolve? And I don't know, TJ, do you want to start while we let folks kind of get the courage to raise a hand and jump in? Yeah. I'll say quick what I think, and then I'll leave the floor to some people who are already raising their hands.
ep068 p2 deep dive screenshare (13:40)
I think the key part is the design to code parity on either end. I just recently put out two demos. It was later last week.
ep068 p2 deep dive group (13:40)
I think the key part is the design to code parity on either end. I just recently put out two demos. It was later last week.
Ben Callahan (13:40)
I think the key part is the design to code parity on either end. I just recently put out two demos ⁓ later, it was later last week,
where we talked about how, and also have one from a while ago where we were starting from two different points, one starting from like contracts and the other one starting from tokens. So meaning like there really isn't much of a Figma or code involved in it. It's kind of starting from this middle ground.
ep068 p2 deep dive screenshare (13:49)
where we talked about how, and also have one from a while ago where we were starting from two different points, one starting from like contracts and the other one starting from tokens. So meaning like there really isn't much of a Figma or code involved in it. It's kind of starting from this middle ground,
ep068 p2 deep dive group (13:49)
where we talked about how, and also have one from a while ago where we were starting from two different points, one starting from like contracts and the other one starting from tokens. So meaning like there really isn't much of a Figma or code involved in it. It's kind of starting from this middle ground,
ep068 p2 deep dive screenshare (14:07)
but whatever it is, it just feels like there needs to be some canonical source, whether it's a design or whether it's, or it's on the code side. So on a code side,
ep068 p2 deep dive group (14:07)
but whatever it is, it just feels like there needs to be some canonical source, whether it's a design or whether it's, or it's on the code side. So on a code side,
Ben Callahan (14:07)
But whatever it is, it just feels like there needs to be some canonical source, whether it's a design or whether it's on the code side. So on the code side,
if it's like a JSON contract, then it's basically establishing all the properties and values that make up your component. And then AI can work with that. then so you can generate something based off of those things. And I think the same thing happens when you start from more of a design leaning side, like if you're starting with a really advanced token architecture, that all of the variable modes or if we're not going to use
ep068 p2 deep dive screenshare (14:16)
If it's like a JSON contract, then it's basically establishing all the properties and values that make up your component. And then AI can work with that. then so you can generate something based off of those things. And I think the same thing happens when you start from more of a design leaning side. Like if you're starting with a really advanced token architecture, that all of the variable modes or, or if we're not going to use
ep068 p2 deep dive group (14:16)
If it's like a JSON contract, then it's basically establishing all the properties and values that make up your component. And then AI can work with that. then so you can generate something based off of those things. And I think the same thing happens when you start from more of a design leaning side. Like if you're starting with a really advanced token architecture, that all of the variable modes or, or if we're not going to use
ep068 p2 deep dive screenshare (14:36)
valuable modes, we'll say themes. If we're using themes, then, then it's enough for the AI to go off to ⁓ accurately put together these.
ep068 p2 deep dive group (14:36)
valuable modes, we'll say themes. If we're using themes, then, then it's enough for the AI to go off to ⁓ accurately put together these.
Ben Callahan (14:36)
ValveModules, we'll say themes. If we're using themes, then it's enough for the AI to go off to ⁓ accurately put together these.
⁓ It's going to have to use more of its general knowledge. Or what we'd like to do is use the Design Systems Assistant MCP to say, create this button based upon best practices. And here's the token architecture I'd like for you to reference. So that way we know when it generates that button.
ep068 p2 deep dive screenshare (14:45)
⁓ It's going to have to use more of its general knowledge or what we'd like to do is use the Design Systems Assistant MCP to say, create this button based upon best practices and here's the token architecture I'd like for you to reference. So that way we know when it generates that button,
ep068 p2 deep dive group (14:45)
⁓ It's going to have to use more of its general knowledge or what we'd like to do is use the Design Systems Assistant MCP to say, create this button based upon best practices and here's the token architecture I'd like for you to reference. So that way we know when it generates that button,
it's going to have all the properties and metadata associated with that button because of the best practices that it ⁓ just referenced. But with the contract, if you're using a JSON contract that'll have ⁓ all of the properties and values defined in it,
ep068 p2 deep dive screenshare (15:01)
it's going to have all the properties and metadata associated with that button because of the best practices that it ⁓ just referenced. But with the contract, if you're using a JSON contract that'll have all of the properties and values defined in it,
Ben Callahan (15:01)
it's going to have all the properties and metadata associated with that button because of the best practices that it just referenced. But with a contract, if you're using a JSON contract that'll have all of the properties and values defined in it,
ep068 p2 deep dive screenshare (15:14)
then you kind of have to give it some aesthetic helpers. So then we don't have an MZP that helps with those things. So it would have to use general knowledge, but you're going to have to goose it a little bit to say like, we want it to look like, I don't know,
ep068 p2 deep dive group (15:14)
then you kind of have to give it some aesthetic helpers. So then we don't have an MZP that helps with those things. So it would have to use general knowledge, but you're going to have to goose it a little bit to say like, we want it to look like, I don't know,
Ben Callahan (15:14)
then you have to give it some aesthetic helpers. So then we don't have an MCP that helps with those things, so it would have to use general knowledge, but you're going to have to goose it a little bit to say, oh, we want it to look like, I don't know, the...
you know, material or, you know, carbon or something to give it some sort of aesthetic feel, but the functionality will be there. And those are just two ways where it feels like even without the world of Figma, that you can have a starting point that doesn't necessarily rely on a
ep068 p2 deep dive screenshare (15:27)
material or carbon or something to give it some sort of aesthetic feel, but the functionality will be there. And those are just two ways where it feels like even without the world of Figma, that you can have a starting point that doesn't necessarily rely on.
ep068 p2 deep dive group (15:28)
the material or carbon or something to give it some sort of aesthetic feel, but the functionality will be there. And those are just two ways where it feels like even without the world of Figma, that you can have a starting point that doesn't necessarily rely on.
Ben Callahan (15:42)
predefined code base or a predefined design system, like, like aesthetic UI design system. Yeah. So anybody throw that out there, open for debate. Love it. Yeah. Let's hear some responses. Eric, jump right in. Yeah.
ep068 p2 deep dive screenshare (15:42)
a predefined code base or a predefined design system, like a static UI design system. So anybody throw that out there, open for debate. Love it. Yeah. Let's hear some responses. Eric, jump right in. Yeah.
ep068 p2 deep dive group (15:42)
a predefined code base or a predefined design system, like, like aesthetic UI design system. Yeah. So anybody throw that out there open for debate. Love it. Yeah. Let's hear some responses. Eric, jump right in. Yeah.
Ben Callahan (15:56)
It's a tricky question for us. Our documentation, our source of truth is actually spread across four different things right now. And we're in the process of consolidating that down.
ep068 p2 deep dive group (15:56)
So, um, it's a, it's a tricky question for us. Our, our documentation, our source of truth is actually spread across four different things right now. And we're in the process of consolidating that down.
ep068 p2 deep dive screenshare (15:58)
It's a tricky question for us. Our documentation, our source of truth is actually spread across four different things right now. And we're in the process of consolidating that down.
We so the things that we have it stored in our JS doc or for a code Jason for our tokens,
ep068 p2 deep dive group (16:11)
We so the things that we have it stored in our JS doc or for a code Jason for our tokens,
Ben Callahan (16:12)
So the things that we have it stored in are JS doc for our code, JSON for our tokens.
ep068 p2 deep dive group (16:20)
Figma for design decisions, and then our reference website, which has all of the examples and documentation and I'm not even going to mention our storybook. So we're in the process of consolidating those all together into one. What we have really been been focusing on is our JS doc documentation.
ep068 p2 deep dive screenshare (16:20)
Figma for design decisions, and then our reference website, which has all of the examples and documentation and I'm not even going to mention our storybook. So we're in the process of consolidating those all together into one. What we have really been focusing on is our JS doc documentation.
Ben Callahan (16:20)
Figma for design decisions, and then our reference website, which has all of the examples and documentation. And I'm not even gonna mention our storybook. So we're in the process of consolidating those all together into one. What we have really been focusing on is our JS doc documentation,
because we're generating a custom elements manifest for our custom HTML elements. And we're using that to help populate our MCP server to get more information about what we're,
ep068 p2 deep dive screenshare (16:39)
because we're generating a custom elements manifest for our custom HTML elements. And we're using that to help populate our MCP server to get more information about what
ep068 p2 deep dive group (16:39)
because we're generating a custom elements manifest for our custom HTML elements. And we're using that to help populate our MCP server to get more information about what
Ben Callahan (16:50)
what our design system is capable of. Jesse can talk a little bit more about that because he's been leading the charge on that. ⁓ our goal is to unify into a single place. So I'm excited to hear what other people are doing for that. And maybe we can take some ideas.
ep068 p2 deep dive screenshare (16:50)
our design system is capable of. Jesse can talk a little bit more about that because he's been leading the charge on that. But our goal is to unify into a single place. So I'm excited to hear what other people are doing for that. And maybe we can take some ideas.
ep068 p2 deep dive group (16:50)
our design system is capable of. Jesse can talk a little bit more about that because he's been leading the charge on that. But ⁓ our goal is to unify into a single place. So I'm excited to hear what other people are doing for that. And maybe we can take some ideas.
Ben Callahan (17:09)
Yeah, sounds like you've got things everywhere, which I think is a common, that's like a common thing, right? Like when I talk to teams.
ep068 p2 deep dive group (17:09)
Yeah, sounds like you've got things everywhere, which I think is a common, that's like a common thing, right? Like when I talk to teams,
ep068 p2 deep dive screenshare (17:09)
Yeah, sounds like you've got things everywhere, which I think is a common, that's like a common thing, right? Like when I talk to teams,
ep068 p2 deep dive group (17:15)
is everywhere. It's spread around. And part of our mandate, I think, is to try and figure out where it makes sense to consolidate that stuff. so thanks for the context there. We'll get Jesse on later, I'm sure. Peter, what's up? Yeah, so I found as I've been kind of developing my own kind of custom design system that the source of truth has evolved.
ep068 p2 deep dive screenshare (17:15)
is everywhere. It's spread around. And part of our mandate, I think, is to try and figure out where it makes sense to consolidate that stuff. so thanks for the context there. We'll get Jesse on later, I'm sure. Peter, what's up? Yeah, so I found as I've been kind of developing my own kind of custom design system that the source of truth has evolved.
Ben Callahan (17:15)
It is everywhere. It's spread around and part of our mandate, I think, is to try and figure out where it makes sense to consolidate that stuff. So thanks for the context there. We'll get Jesse on later, I'm sure. Peter, what's up? Yeah, so ⁓ I found as I've been kind of developing my own kind of custom design system that the source of truth has evolved.
So like it began in design just to understand the framing of how we kind of wanted to structure things. But that
ep068 p2 deep dive screenshare (17:35)
So like it began in design just to understand the framing of how we kind of wanted to structure things. ⁓
ep068 p2 deep dive group (17:35)
So like it began in design just to understand the framing of how we kind of wanted to structure things. But once
ep068 p2 deep dive screenshare (17:42)
But once that kind of got converted into code, the code became the source of truth. ⁓ But to manage all that, I've developed kind of our documentation MCP off of like 65, 70 some documents that we have, and that has become the source of truth. So that when we are kind of working through, I say we because I'm talking about me and.
Ben Callahan (17:42)
But once that kind of got converted into code, the code became the source of truth. ⁓ But to manage all that, I've developed kind of our documentation MCP off of like 65, 70 some documents that we have, and that has become the source of truth. So that when we are kind of working through, ⁓ I say we because I'm talking about me and
ep068 p2 deep dive group (17:42)
that kind of got converted into code, the code became the source of truth. But to manage all that, I've developed kind of our documentation MCP off of like 65, 70 some documents that we have, and that has become the source of truth. So that when we are kind of working through, I say we because I'm talking about me and.
Ben Callahan (18:01)
three AI agents. as we kind of... ⁓
ep068 p2 deep dive screenshare (18:01)
three AI agents. as we kind of,
ep068 p2 deep dive group (18:01)
three AI agents. as we kind of,
Ben Callahan (18:07)
In development, what we always go back to is like, all right, what does the MCP, what does our documentation and MCP say? And that becomes kind of the guiding point to kind of keep those things on rails. So it's just kind of evolved eventually to where documentation is the source of truth more than anything else, because it informs the design and the engineering. Yeah. I love that. more of a spectrum of an approach.
ep068 p2 deep dive screenshare (18:07)
in development, what we always go back to is like, what does the MCP, what does our documentation and MCP say? And that becomes kind of the guiding point to kind of keep those things on rails. So it's just kind of evolved eventually to where documentation is the source of truth more than anything else, because it informs the design and the engineering. Yeah, I love that. like more of a spectrum of an approach.
ep068 p2 deep dive group (18:07)
in development, what we always go back to is like, what does the MCP, what does our documentation and MCP say? And that becomes kind of the guiding point to kind of keep those things on rails. So it's just kind of evolved eventually to where documentation is the source of truth more than anything else, because it informs the design and the engineering. Yeah, I love that. like more of a spectrum of an approach.
ep068 p2 deep dive screenshare (18:29)
Yeah, it is. Yeah. So the idea for me is that I don't want to go from design to code and code to design.
ep068 p2 deep dive group (18:29)
Yeah, it is. Yeah. So the idea for me is that I don't want to go from design to code and code to design.
Ben Callahan (18:29)
Yeah. It is. Yeah. So the idea for me is that I don't want to go from design to code and code to design.
ep068 p2 deep dive group (18:36)
I want to create more of kind of this word flywheel where the ⁓ code informs the tools in Figma. So basically I'm using right now a console to kind of port everything from the code into Figma. And then I want the Figma edits that I make to go into and form a spec.
ep068 p2 deep dive screenshare (18:36)
I want to create more of kind of this word flywheel where the ⁓ code informs the the tools in Figma. So basically I'm using right now console console to kind of port everything from the code into Figma. And then I want the Figma edits that I make to go into and form a spec.
Ben Callahan (18:36)
I want to create more of a kind of, I this word, flywheel, where the ⁓ code informs the ⁓ tools in Figma. So basically I'm using right now a console to kind of port everything from the code into Figma. And then I want the Figma edits that I make to go into and form a spec
that I develop. And the spec then goes back to inform the code. So it's not necessarily a binary kind of kind of
ep068 p2 deep dive screenshare (18:58)
that I develop and the spec then goes back to inform the code. So it's not necessarily a binary kind of kind of
ep068 p2 deep dive group (18:58)
that I develop and the spec then goes back to inform the code. So it's not necessarily a binary kind of kind of
Ben Callahan (19:04)
a dual direction thing. It's more of a, exactly. It's all kind of one direction, but it all kind of informs each other, but it has those opportunities to kind of adjust for drift ⁓ and to correct little details as we kind of iterate and go along. That's one of the cool things about what I've loved early on that felt
ep068 p2 deep dive screenshare (19:04)
a dual direction thing. It's more of a, exactly. It's all kind of one direction, but it all kind of informs each other, but it has those opportunities to kind of adjust for drift and to correct little details as we kind of iterate and go along. That's one of the cool things about what I've loved early on that felt.
ep068 p2 deep dive group (19:04)
a direction thing. It's more of a, exactly. It's all kind of one direction, but it all kind of informs each other, but it has those opportunities to kind of adjust for drift and to correct little details as we kind of iterate and go along. That's one of the cool things about what I've loved early on that felt.
Ben Callahan (19:23)
great about the Figma Consult MCP was the ability to do that.
ep068 p2 deep dive screenshare (19:23)
great about the Figma console MCP was the ability to do that.
ep068 p2 deep dive group (19:23)
great about the Figma console MCP was the ability to do that.
This kind of that feedback loop, how you can establish like whatever your canonical source was that the AI was going to use to build whatever it was, whether you're going to go from design to code or code to design, whichever, usually we're going from design to code or something. So if you say that documentation or that spec was what you're using. So what we would do is say,
ep068 p2 deep dive screenshare (19:28)
This kind of that feedback loop, how you can establish like whatever your canonical source was that the AI was going to use to build whatever it was, whether you're going to go from design to code or code to design, whichever, usually we're going from design to code or something. So if you say that documentation or that spec was what you're using. So what we would do is say,
Ben Callahan (19:28)
this kind of that feedback loop, how you can establish like whatever your canonical source was that the AI was gonna use to build whatever it was, whether you're gonna go from design to code or code to design, whichever way, usually we're going from design to code or something. So if you say that documentation or that spec was what you're using, so what we would do is say,
this is what we're getting all of our use from to define that spec, we would use the design systems assistant to ensure that everything was done according to best practices.
ep068 p2 deep dive screenshare (19:49)
this is what we're getting all of our use from to define that spec. We would use the design systems assistant to ensure that everything was done according to best practices.
ep068 p2 deep dive group (19:49)
this is what we're getting all of our use from to define that spec. We would use the design systems assistant to ensure that everything was done according to best practices.
ep068 p2 deep dive screenshare (19:57)
And then it's got accessibility rules and self-builds into that too. And then we'd use the console MCP to create the code. And then it has this loop where it takes the screenshot of what it created and then feeds it back into the documentation to say, this abiding by the rules of this documentation? And then if it says no or yes, kind of a pass fail, and then it makes a little checklist.
ep068 p2 deep dive group (19:57)
And then it's got accessibility rules and stuff built into that too. And then we'd use the console MCP to create the code. And then it has this loop where it takes the screenshot of what it created and then feeds it back into the documentation to say, this abiding by the rules of this documentation? And then if it says no or yes, kind of a pass fail, and then it makes a little checklist.
Ben Callahan (19:57)
And then it's got accessibility rules and stuff built into that too. And then we'd use the console MCP to create the code. And then it has this loop where it takes the screenshot of what it created and then feeds it back into the documentation to say, this abiding by the rules of this documentation? And then if it says no or yes, kind of a pass fail, and then it makes a little checklist
of what actually needs to get addressed with it and then continues that loop over and over again. So it's one of those moments where you can just go get a Diet Coke and hang out and watch it.
ep068 p2 deep dive screenshare (20:18)
of what actually needs to get addressed with it, and then continues that loop over and over again. So it's one of those moments where you can just go get a Diet Coke and hang out and watch it.
ep068 p2 deep dive group (20:18)
of what actually needs to get addressed with it, and then continues that loop over and over again. So it's one of those moments where you can just go get a Diet Coke and hang out and watch it.
ep068 p2 deep dive screenshare (20:26)
And then it'll keep doing its thing for like, you know, I don't know, 20 minutes or so if it's like a more composed structure. But to me, it was like a glimpse into the future of like, it's not perfect, but I could see like a lot of this like really beautiful automation working out.
ep068 p2 deep dive group (20:26)
And then it'll keep doing its thing for like, I don't know, 20 minutes or so if it's like a more composed structure. to me, it was like a glimpse into the future of like, it's not perfect, but I could see like a lot of this like really beautiful automation working out.
Ben Callahan (20:26)
and then it'll keep doing its thing for like, I don't know, 20 minutes or so if it's like a more composed structure. ⁓ to me, it was like a glimpse into the future of like, it's not perfect, but I could see like a lot of this like really beautiful automation working out
no matter what the source of truth is, whether it's a design or it's what you're saying, something like a spec. There's like ⁓ a weird analogy here. I was talking with somebody ⁓
ep068 p2 deep dive screenshare (20:43)
no matter what the source of truth is, whether it's a design or it's what you're saying, something like a spec. There's like a weird analogy here. I was talking with somebody a
ep068 p2 deep dive group (20:43)
no matter what the source of truth is, whether it's a design or it's what you're saying, something like a spec. There's like a weird analogy here. I was talking with somebody a
Ben Callahan (20:55)
couple days ago about source of truth and we were using language like human language as an example. it's it's kind of like, part of me is trying to figure out like, why do we need something other than the artifacts themselves? In other words, like,
ep068 p2 deep dive screenshare (20:55)
couple of days ago about source of truth and we were using language, like human language as an example. it's kind of like, part of me is trying to figure out like, why do we need something other than the artifacts themselves? In other words, like,
ep068 p2 deep dive group (20:55)
couple of days ago about source of truth and we were using language, like human language as an example. it's kind of like, part of me is trying to figure out like, why do we need something other than the artifacts themselves? In other words, like,
Ben Callahan (21:11)
if I'm an English speaker, and you're a Spanish speaker, my source of truth is English, that's what I know, because I live in that world, yours is Spanish.
ep068 p2 deep dive screenshare (21:11)
If I'm an English speaker and you're a Spanish speaker, my source of truth is English. That's what I know because I live in that world. Yours is Spanish.
ep068 p2 deep dive group (21:11)
If I'm an English speaker and you're a Spanish speaker, my source of truth is English. That's what I know because I live in that world. Yours is Spanish.
ep068 p2 deep dive screenshare (21:19)
We don't have some sort of generic language that serves as an intermediary, right? We learn how to speak another language. We don't learn to speak the generic so that we can talk to each other. And in some sense, I feel like we're trying to do that here. And I don't know if the analogy holds, but like, where does this sit? You know, like, do we need to have a standard like we do? And I see Kaligs on the call who's helped us.
ep068 p2 deep dive group (21:19)
We don't have some sort of generic language that serves as an intermediary, right? We learn how to speak another language. We don't learn to speak the generic so that we can talk to each other. And in some sense, I feel like we're trying to do that here. And I don't know if the analogy holds, but like, where does this sit? You know, like, do we need to have a standard like we do? And I see Kaligs on the call who's helped us.
Ben Callahan (21:19)
We don't have some sort of generic language that serves as an intermediary, right? We learn how to speak another language. We don't learn to speak the generic so that we can talk to each other. And in some sense, I feel like we're trying to do that here. And I don't know if the analogy holds, but like, where does this sit? know, like, do we need to have a standard like we do? And I see Kaligs on the call who's helped us
build an actual standard around design decisions in the form of tokens, right?
ep068 p2 deep dive screenshare (21:41)
build an actual standard around design decisions in the form of tokens, right?
ep068 p2 deep dive group (21:41)
build an actual standard around design decisions in the form of tokens, right?
ep068 p2 deep dive screenshare (21:47)
And so that to me is like this sort of intermediary, but do we need that, I guess, is what I'm asking. I'm not trying to call into question all that hard work, Kayleigh, because I know it's been a long journey, but I'm just wondering, does that resonate in this conversation in some way as well? And then also, ⁓ Jenny, I want to call you out a little bit, if I can. If you're willing to jump on, I'd love to have you explain the note you added while some of this was going on, because I think it's a
ep068 p2 deep dive group (21:47)
And so that to me is like this sort of intermediary, but do we need that? guess is what I'm asking. I'm not trying to call into question all that hard work, Kayleigh, because I know it's been a long journey, but I'm just wondering, does that resonate in this conversation in some way as well? And then also, Jenny, I want to call you out a little bit, if I can, if you're willing to jump on, I'd love to have you explain the note you added while some of this was going on, because I think it's a
Ben Callahan (21:47)
And so that to me is like this sort of intermediary, but do we need that, I guess, is what I'm asking. I'm not trying to call into question all that hard work, Kayleigh, because I know it's been a long journey, but I'm just wondering, does that resonate in this conversation in some way as well? And then also, Jenny, I want to call you out a little bit, if I can. If you're willing to jump on, I'd love to have you explain the note you added while some of this was going on, because I think
it's a really nice sort of clarifying ⁓ explanation.
ep068 p2 deep dive screenshare (22:12)
It's a really nice sort of clarifying ⁓ explanation.
ep068 p2 deep dive group (22:12)
It's a really nice sort of clarifying ⁓ explanation.
ep068 p2 deep dive screenshare (22:15)
So raise your hand if you're willing and we'll get you to come up as well. I'd love to hear that. ⁓ Christian, jump in.
ep068 p2 deep dive group (22:15)
So raise your hand if you're willing and we'll get you to come up as well. I'd love to hear that. ⁓ Christian, jump in.
Ben Callahan (22:15)
So raise your hand if you're willing and we'll get you to come up as well. I'd love to hear that. ⁓ Christian, jump in.
ep068 p2 deep dive screenshare (22:24)
Hey, everyone. Yeah, rolling rolling back. I I actually show you all the stuff I have been doing doing but yeah for me it was super easy to To put the source of truth on the go directly because we're a small team only designer working with ten depths So the source of truth decision read there was super easy like okay most of the team is dead moving there and
ep068 p2 deep dive group (22:24)
Hey, everyone. rolling back, I actually show you all the stuff I have been doing. But yeah, for me, it was super easy to put the source of truth on the go directly because we're a small team, only designer working with 10 devs. So the source of truth, decision read, there was super easy. Like, okay, most of the team is devs, moving there.
Ben Callahan (22:25)
Hey, Rolling back, I actually show you all the stuff I have been doing. But yeah, for me, it was super easy to put the source of truth on the go directly because we're a small team, only designer working with 10 devs. So the source of truth, decision read, there was super easy. Like, okay, most of the team is devs, moving there. And
also because
ep068 p2 deep dive screenshare (22:53)
Also because
ep068 p2 deep dive group (22:53)
Also because
ep068 p2 deep dive screenshare (22:55)
we
ep068 p2 deep dive group (22:55)
we can choose very agnostic formats like the JSON for the docks or maybe the JS docs, right? For the metadata and design decisions. Maybe it's a markdown with the skill instructions. And I have something that I work after. I show you this. And we started creating a skill that audits but also understanding the...
Ben Callahan (22:56)
we can choose very agnostic formats like the JSON for the docks or maybe the JS docs, right? For the metadata and design decisions. Maybe it's a markdown with the skill instructions. And I have something that I work after. I show you this. And we started creating a skill that audits but also understanding the
ep068 p2 deep dive screenshare (22:56)
can choose very agnostic formats like the JSON for the docks or maybe the JS docs, right? For the metadata and the sign decisions. Maybe it's a markdown with the skill instructions. And I have something that I work after. I show you this. And we started creating a skill that audits but also understanding the...
the same decision as intent
ep068 p2 deep dive group (23:23)
the same decision as the intent
Ben Callahan (23:23)
design decision and the intent.
ep068 p2 deep dive screenshare (23:25)
behind the tokens. So we integrated that. And now we are creating reports, ⁓ auditing the components we already have and checking, hey, it's not only a linter on, are we using the variables correctly? But is the correct variable the one that we are using? So ⁓ I dropped somewhere around there a Fig.Jam link.
ep068 p2 deep dive group (23:25)
behind the tokens. So we integrated that. And now we are creating reports, auditing the components we already have and checking, hey, it's not only a linter on, are we using the variables correctly? But is the correct variable the one that we are using? So ⁓ I dropped somewhere around there a Fig.Jam link.
Ben Callahan (23:25)
behind the tokens. So we integrated that. And now we are creating reports, ⁓ auditing the components we already have and checking, hey, it's not only a linter on, are we using the variables correctly? But is the correct variable the one that we are using? So I dropped somewhere around there a Fig.Jam link.
with the full AI structure I'm working. So there is a lot because as you mentioned, the source of truth depends on the buff. I'm an IE signer, the source of truth is in Figma maybe. I'm a dev, the source of truth is in the Go. Probably I'm a PM or the CEO, the source of truth is in an Excel sheet. So that's something.
ep068 p2 deep dive screenshare (23:51)
with the full AI structure I'm working. So there is a lot because as you mentioned, the source of truth depends on the buff. I'm an IE signer, the source of truth is in Figma maybe. I'm a dev, the source of truth is in the Go. Probably I'm a PM or the CEO, the source of truth is in an Excel sheet. So that's something.
ep068 p2 deep dive group (23:51)
with the full AI structure I'm working. So there is a lot because as you mentioned, the source of truth depends on the buff. I'm an IE signer, the source of truth is in Figma maybe. I'm a dev, the source of truth is in the Go. Probably I'm a PM or the CEO, the source of truth is in an Excel sheet. So that's something.
Ben Callahan (24:21)
And I have been finding a way to introduce the design decisions into the system so it can auto self-heal itself or self-document or self-report the stuff. it's been working, not great, great, but it's working and making this stuff easy because it's mostly about dumping our design rules into the environment so the AIs can use it.
ep068 p2 deep dive screenshare (24:21)
And I have been finding a way to introduce the design decisions into the system so it can auto self-heal itself or self-document or self-report the stuff. it's been working, not great, great, but it's working and making this stuff easy because it's mostly about dumping our design rules into the environment so the AIs can use it.
ep068 p2 deep dive group (24:21)
And I have been finding a way to introduce the design decisions into the system so it can auto self-heal itself or self-document or self-report the stuff. it's been working, not great, great, but it's working and making this stuff easy because it's mostly about dumping our design rules into the environment so the AIs can use it.
Ben Callahan (24:51)
and understanding. Yeah, you're getting an intent, right? Like the design intent. And how do we capture that, which I think is critical to this. If you haven't read Christian's work, there's a bunch of his articles down here as well. So please check out some of the stuff he's doing. It's pretty incredible stuff. And drop more links if I haven't gotten them all, Christian, because it's awesome. Thank you. ⁓ Greg, what's up? Yeah, so I think a lot of this
ep068 p2 deep dive screenshare (24:51)
and understanding. Yeah, you're getting an intent, right? Like the design intent. And how do we capture that, which I think is critical to this. If you haven't read Christian's work, there's a bunch of his articles down here as well. So please check out some of the stuff he's doing. It's pretty incredible stuff. And drop more links if I haven't gotten them all, Christian, because it's awesome. Thank you. ⁓ Greg, what's up? Yeah, so I think a lot of this
ep068 p2 deep dive group (24:51)
and understanding. Yeah, you're getting an intent, right? Like the design intent. And how do we capture that, which I think is critical to this. If you haven't read Christian's work, there's a bunch of his articles down here as well. So please check out some of the stuff he's doing. It's pretty incredible stuff and drop more links if I haven't gotten them all, Christian, because it's awesome. Thank you. ⁓ Greg, what's up? Yeah, so I think a lot of this.
Ben Callahan (25:16)
depends on the point of view. think a lot of the folks on this call, because we're systems practitioners and we're looking at it from that lens, I think a lot of us would probably agree that the source of truth is in the code or the documentation. You start with intent maybe in the design and then you codify those decisions into the code. And then there's a protective layer around it because there's
ep068 p2 deep dive screenshare (25:16)
depends on the point of view. think a lot of the folks on this call, because we're systems practitioners and we're looking at it from that lens, I think a lot of us would probably agree that the source of truth is in the code or the documentation. You start with intent maybe in the design and then you codify those decisions into the code. And then there's a protective layer around it because there's
ep068 p2 deep dive group (25:16)
depends on the point of view. think a lot of the folks on this call, because we're systems practitioners and we're looking at it from that lens, I think a lot of us would probably agree that the source of truth is in the code or the documentation. You start with intent maybe in the design and then you codify those decisions into the code. And then there's a protective layer around it because there's
all sorts of version control and all this stuff. it's, you know, we're very careful about in development of how we, how and when we make changes to things, right? Whereas in design, because typically in the life cycle, the process, the design stuff is more loose. And sometimes, you know, we're looking at things like, you know, when I think of like using AI for things like VIDE coding,
Ben Callahan (25:45)
all sorts of version control and all this stuff. it's, you know, we're very careful about in development of how we, how and when we make changes to things, right? Whereas in design, because typically in the life cycle, the process, the design stuff is more loose. And sometimes, you know, we're looking at things like, you know, when I think of like using AI for things like byte coding,
ep068 p2 deep dive screenshare (25:45)
all sorts of version control and all this stuff. it's, you know, we're very careful about in development of how we, how and when we make changes to things, right? Whereas in design, because typically in the life cycle, the process, the design stuff is more loose. And sometimes, you know, we're looking at things like, you know, when I think of like using AI for things like byte coding,
Ben Callahan (26:15)
That to me feels like it's always early process stuff, right? Like you're trying to figure out new ideas and try new things. there, in some ways, system source of truth actually matters less because you're like, I don't want to think about my limitations. I want to think about possibilities. And then you have to reconcile while you're going through the process, OK, now I need to actually use as much of the system as I can.
ep068 p2 deep dive screenshare (26:15)
That to me feels like it's always early process stuff, right? Like you're trying to figure out new ideas and try new things. there, in some ways, system source of truth actually matters less because you're like, I don't want to think about my limitations. I want to think about possibilities. And then you have to reconcile while you're going through the process, OK, now I need to actually use as much of the system as I can.
ep068 p2 deep dive group (26:15)
That to me feels like it's always early process stuff, right? Like you're trying to figure out new ideas and try new things. And there, in some ways, system source of truth actually matters less because you're like, I don't want to think about my limitations. I want to think about possibilities. And then you have to reconcile while you're going through the process, OK, now I need to actually use as much of the system as I can.
so that I'm not causing unnecessary work, we need to actually start to think about reality a little bit more. And then you get further into the process where you get into trying to use AI to actually make code that uses the system appropriately. And that's like a whole different thing. And I think a lot of stuff gets lost in just kind of
Ben Callahan (26:45)
so that I'm not causing unnecessary work. We need to actually start to think about reality a little bit more. And then you get further into the process where you get into trying to use AI to actually make code that uses the system appropriately. And that's like a whole different thing. And I think a lot of stuff gets lost in just kind of
ep068 p2 deep dive screenshare (26:45)
so that I'm not causing unnecessary work. We need to actually start to think about reality a little bit more. And then you get further into the process where you get into trying to use AI to actually make code that uses the system appropriately. And that's like a whole different thing. And I think a lot of stuff gets lost in just kind of
talking broadly about, I'm using AI, I'm going to use Figma Make or Lovable or something, and I'm going to make stuff. And yeah, when you're doing that stuff and it is more conceptual or loose and creative or iterative or innovative even, if that's how you're thinking about it, ⁓ and a lot of what I'm describing, I think, is thinking about more of like,
Ben Callahan (27:15)
talking broadly about, I'm using AI, I'm going to use Figma Make or Lovable or something, and I'm going to make stuff. And yeah, when you're doing that stuff and it is more conceptual or loose and creative or iterative or innovative even, if that's how you're thinking about it, and a lot of what I'm describing, I think, is thinking about more of like,
ep068 p2 deep dive group (27:15)
talking broadly about, I'm using AI, I'm going to use Figma Make or Lovable or something, and I'm going to make stuff. yeah, when you're doing that stuff and it is more conceptual or loose and creative or iterative or innovative even, if that's how you're thinking about it, a lot of what I'm describing, think, is thinking about more of like,
Ben Callahan (27:42)
the consumers of the system, the product designers that are using the system. Because there's a lot of that conversation going on in my world right now. We're trying to figure out how do we help build things to help the product designers do their work and use the system when it's appropriate and as much as they can. But where's that line? If they're going to use Figma Make, we've to make templates that
ep068 p2 deep dive screenshare (27:42)
the consumers of the system, the product designers that are using the system. Because there's a lot of that conversation going on in my world right now. We're trying to figure out how do we help build things to help the product designers do their work and use the system when it's appropriate and as much as they can. But where's that line? If they're going to use Figma Make, we've to make templates that
ep068 p2 deep dive group (27:42)
the consumers of the system, the product designers that are using the system. ⁓ Because there's a lot of that conversation going on in my world right now. We're trying to figure out how do we help build things to help the product designers do their work ⁓ and use the system when it's appropriate and as much as they can. But where's that line? If they're going to use Figma Make, we've to make templates that
Ben Callahan (28:12)
maybe use the components from the system? Or do we not do that? It just needs to kind of be a wire framing tool to do more loose and quick work and prove out ideas. So you can have those collaborative discussions in your triads with the product and development people. So. haven't thought on that. Yeah. Remember Figma prototypes?
ep068 p2 deep dive screenshare (28:12)
maybe use the components from the system? Or do we not do that? It just needs to kind of be a wire framing tool to do more loose and quick work and prove out ideas. So you can have those collaborative discussions in your triads with the product and development people. So. haven't thought on that. Yeah. Remember Figma prototypes?
ep068 p2 deep dive group (28:12)
maybe use the components from the system? Or do we not do that? It just needs to kind of be a wire framing tool to do more loose and quick work and prove out ideas. So you can have those collaborative discussions in your triads with the product and development people. ⁓ So. haven't thought on that. Yeah. Remember Figma prototypes?
Ben Callahan (28:42)
I want them to go away. But it seems like we've quickly advanced so far into what we used to accept sometimes when we would get our Figma comps would be like, all right, here's the interaction or the expected interaction. And we would deal with prototypes. so quickly have you forgotten about prototypes? Because now it's like we're advancing directly into the actual experience.
ep068 p2 deep dive screenshare (28:42)
Oh, I want them to go away. But it seems like we've quickly advanced so far into what we used to accept sometimes when we would get our Figma comps would be like, all right, here's the interaction or the expected interaction. And we would deal with prototypes. so quickly have you forgotten about prototypes? Because now it's like we're advancing directly into the actual experience.
ep068 p2 deep dive group (28:42)
I want them to go away. But it seems like we've quickly advanced so far into what we used to accept sometimes when we would get our Figma comps would be like, all right, here's the interaction or the expected interaction and we would deal with prototypes. But so quickly have you forgotten about prototypes because now it's like we're advancing directly into the actual experience.
Ben Callahan (29:08)
And then people are so quick to push product designers into a world where they're actually starting to create and contribute to the product. When I think it's OK to use tools like Figma Make to do a more advanced version of what the original intent of what the Figma prototypes were doing, even if it's not something that's going to eventually make it to the product, I think it's OK. Because at least it's one step further into mimicking the actual interaction that the users will.
ep068 p2 deep dive screenshare (29:08)
And then people are so quick to push product designers into a world where they're actually starting to create and contribute to the product. When I think it's OK to use tools like Figma Make to do a more advanced version of what the original intent of what the Figma prototypes were doing, even if it's not something that's going to eventually make it to the product, I think it's OK. Because at least it's one step further into mimicking the actual interaction that the users will.
ep068 p2 deep dive group (29:08)
And then people are so quick to push product designers into a world where they're actually starting to create and contribute to the product. When I think it's OK to use tools like Figma Make to do a more advanced version of what the original intent of what the Figma prototypes were doing, even if it's not something that's going to eventually make it to the product, I think it's OK. Because at least it's one step further into mimicking the actual interaction that the users will.
Ben Callahan (29:37)
And then also looking more like the tool, looking more like the design system. Also to the fact that you can like generate it all and it takes way less time to create something like that using thing we make than it does using the old prototyping tools. Yeah, that's a good point. And I'm just thinking too, like part of what I heard from you, Greg, was just this idea that like, there's different use cases for this stuff and the source of truth might be different depending on the use case, you know, like how close to production ready, like
ep068 p2 deep dive screenshare (29:37)
And then also looking more like the tool, looking more like the design system. Also to the fact that you can like generate it all and it takes way less time to create something like that using thing we make than it does using the old prototyping tools. Yeah, that's a good point. And I'm just thinking too, like part of what I heard from you, Greg, was just this idea that like, there's different use cases for this stuff and the source of truth might be different depending on the use case, you know, like how close to production ready, like
ep068 p2 deep dive group (29:37)
And then also looking more like the tool, looking more like the design system. Also to the fact that you can like generate it all and it takes way less time to create something like that using thing we make than it does using the old prototyping tools. Yeah, that's a good point. And I'm just thinking too, like part of what I heard from you, Greg, was just this idea that like, there's different use cases for this stuff and the source of truth might be different depending on the use case, you know, like how close to production ready, like
Ben Callahan (30:07)
perfect, we need to be with the artifact we're trying to generate? ⁓ that's maybe an important thing for us to keep in mind too. Gannon, what's up?
ep068 p2 deep dive screenshare (30:07)
perfect, we need to be with the artifact we're trying to generate? ⁓ that's maybe an important thing for us to keep in mind too. Gannon, what's up?
ep068 p2 deep dive group (30:07)
perfect do we need to be with the artifact we're trying to generate. So ⁓ that's maybe an important thing for us to keep in mind too. Gannon, what's up?
ep068 p2 deep dive screenshare (30:19)
I just will jump in and say on the source of truth question, ⁓ I'm somebody from an org, but like, we absolutely don't have one. And I think what we're kind of trying to figure out right now is this, ⁓ idea of, okay. If different sort of sections of the company, different parts of the product development, engineering space, all have different levels of comfort with AI in general, different strategies for sort of.
Ben Callahan (30:19)
I just will jump in and say on the source of truth question, ⁓ I'm somebody from an org, but like, we absolutely don't have one. And I think what we're kind of trying to figure out right now is this, ⁓ idea of, okay. If different sort of sections of the company, different parts of the product development, engineering space, all have different levels of comfort with AI in general, different strategies for sort of.
ep068 p2 deep dive group (30:19)
I just will jump in and say on the source of truth question, ⁓ I'm somebody from an org, but like, we absolutely don't have one. And I think what we're kind of trying to figure out right now is this, ⁓ idea of, okay. If different sort of sections of the company, different parts of the product development, engineering space, all have different levels of comfort with AI in general, different strategies for sort of.
Ben Callahan (30:48)
keeping track of the source of truth they feel like is in their domain and they own. As opposed to trying to find a single source of truth, like can we start layering, you know, AI powered systems kind of over the top of everything to do things like figure out where is their drift? How can we, you know, maybe create, almost I'm starting to think of it like an orchestration layer. And I'm curious if anybody else has started to experiment with that kind of stuff. I got thrown into a project working on
ep068 p2 deep dive screenshare (30:48)
keeping track of the source of truth they feel like is in their domain and they own. As opposed to trying to find a single source of truth, like can we start layering, you know, AI powered systems kind of over the top of everything to do things like figure out where is their drift? How can we, you know, maybe create, almost I'm starting to think of it like an orchestration layer. And I'm curious if anybody else has started to experiment with that kind of stuff. I got thrown into a project working on
ep068 p2 deep dive group (30:48)
keeping track of the source of truth they feel like is in their domain and they own. As opposed to trying to find a single source of truth, like can we start layering, you know, AI powered systems kind of over the top of everything to do things like figure out where is their drift? How can we, you know, maybe create, almost I'm starting to think of it like an orchestration layer. And I'm curious if anybody else has started to experiment with that kind of stuff. I got thrown into a project working on
Ben Callahan (31:18)
chat bot stuff, basically AI powered where you're trying to divert people to a chat bot as opposed to a customer service agent. And that was sort of my first introduction to retrieval augmented generation and vector databases and all of that kind of stuff. I'm sort of curious if people are starting to look at that and like an orchestration layer model over the top of potentially a lot of different systems. Yeah, jump in Ismael.
ep068 p2 deep dive screenshare (31:18)
chat bot stuff, basically AI powered where you're trying to divert people to a chat bot as opposed to a customer service agent. And that was sort of my first introduction to retrieval augmented generation and vector databases and all of that kind of stuff. I'm sort of curious if people are starting to look at that and like an orchestration layer model over the top of potentially a lot of different systems. Yeah, jump in Ismael.
ep068 p2 deep dive group (31:18)
chat bot stuff, basically AI powered where you're trying to divert people to a chat bot as opposed to a customer service agent. And that was sort of my first introduction to retrieval augmented generation and vector databases and all of that kind of stuff. I'm sort of curious that people are starting to look at that and like an orchestration layer model over the top of potentially a lot of different systems. Yeah, jump in Ismael.
Ben Callahan (31:45)
So personally, I'm working on an agent ⁓ to automatize UX audit. And I did it a bit the rough way asking Claude to do the thing for me. And on the first run that I did, I just burned something like $10 of tokens because it was not well architected.
ep068 p2 deep dive group (31:45)
⁓ So personally, I'm working on an agent ⁓ to automatize UX audit. And I did it a bit the rough way asking Claude to do the thing for me. And on the first ⁓ run that I did, I just burned something like $10 of tokens because it was not well architected.
ep068 p2 deep dive screenshare (31:45)
⁓ so personally, I'm working on an agent, ⁓ to automatize, UX audit. And I did it a bit the rough way asking Claude to do the thing for me. And on the first run that I did, I just burned something like $10 of tokens, ⁓ because it was not well architected.
Ben Callahan (32:11)
Yeah. Sorry. And I thought that it was the right way to do it because Claude told me to do it that way. But that's how I always, also a good salesman. Yeah. Your idea is the best idea and you're so wonderful. Yeah. And so I talked with an AI expert in my, the new company that I'm working for.
ep068 p2 deep dive group (32:11)
Yeah. Sorry. And I thought that it was the right way to do it because Claude told me to do it that way. But that's how it always is. Your idea is the best idea and you're so wonderful. Yeah. And so I talked with an AI expert in the new company that I'm working for.
ep068 p2 deep dive screenshare (32:11)
Yeah. Sorry. And I thought that it was the right way to do it because Claude told me to do it that way. But that's how I always also good salesman. Yeah. Your idea is the best idea and you're so wonderful. Yeah. And so I talked with an AI expert in my, the new company that I'm working for.
Ben Callahan (32:40)
And we had a really super interesting conversation around like thinking ⁓ the different modules as ⁓ like the atomic way. Like we think of, you know, atomic research on atomic design. So that's how I am still working on it, but thinking it as like little modules.
ep068 p2 deep dive screenshare (32:40)
And we had a really super interesting conversation around like thinking ⁓ the different modules as ⁓ like the atomic way. Like we think of, you know, atomic research on atomic design. So that's how I am still working on it, but thinking it as like little modules.
ep068 p2 deep dive group (32:40)
And we had a really super interesting conversation around like thinking ⁓ the different modules as ⁓ like the atomic way. Like we think of, you know, atomic research on atomic design. So that's how I am still working on it, but thinking it as like little modules.
Ben Callahan (33:08)
like each sub-agent with his own skills, his own MCP, his own everything and thinking that way. So to answer what you are working on, that's how personally I am working on at the moment, like thinking it as each theme, each pieces of the theme and what skills, what tools and what knowledge each piece needs and not...
ep068 p2 deep dive screenshare (33:08)
like each sub-agent with his own skills, his own MCP, his own everything and thinking that way. So to answer what you are working on, that's how personally I am working on at the moment, like thinking it as each theme, each pieces of the theme and what skills, what tools and what knowledge each piece needs and not...
ep068 p2 deep dive group (33:08)
like each sub-agent with his own skills, his own MCP, his own everything and thinking that way. So to answer what you are working on, that's how personally I am working on at the moment, like thinking it as each theme, each pieces of the theme and what skills, what tools and what knowledge each piece needs and not...
Ben Callahan (33:37)
all the thing as once because you'll load the whole context of the whole project each time and you will burn a lot, a lot of tokens and even giving too much context loses the context. Yeah, interesting. Yeah. I'm seeing like a couple of things happening in the FigJam that I want to get to. Thank you Ismail for sharing.
ep068 p2 deep dive screenshare (33:37)
all the thing as once because you'll load the whole context of the whole project each time and you will burn a lot, a lot of tokens and even giving too much context loses the context. Yeah, interesting. Yeah. I'm seeing like a couple of things happening in the FigJam that I want to get to. Thank you Ismail for sharing.
ep068 p2 deep dive group (33:37)
all the thing as once because you'll load the whole context of the whole project each time and you will burn a lot, a lot of tokens and even giving too much context loses the context. Yeah, interesting. Yeah. I'm seeing like a couple of things happening in the FigJam that I want to get to. Thank you Ismail for sharing.
Ben Callahan (34:04)
⁓ especially the sales ability of Claude, something we should keep out of the look out for. But I see, I know the New York State crews here, the pattern engine image over here, I think somebody dropped in and then I see Jenny's ⁓ AI infrastructure, system as AI infrastructure. I'd love to hear from those teams if you're willing to share a little bit more to just give us the 60 to 90 second unpack of these images. So if so, raise your hands, we'll get you up here. ⁓
ep068 p2 deep dive screenshare (34:04)
⁓ especially the sales ability of Claude, something we should keep out of the look out for. But I see, I know the New York State crews here, the pattern engine image over here, I think somebody dropped in and then I see Jenny's ⁓ AI infrastructure, system as AI infrastructure. I'd love to hear from those teams if you're willing to share a little bit more to just give us the 60 to 90 second unpack of these images. So if so, raise your hands, we'll get you up here. ⁓
ep068 p2 deep dive group (34:04)
⁓ especially the sales ability of Claude, something we should keep out a look for. But I see, I know the New York State crews here, the pattern engine image over here, I think somebody dropped in and then I see Jenny's ⁓ AI infrastructure, system as AI infrastructure. I'd love to hear from those teams if you're willing to share a little bit more to just give us the 60 to 90 second unpack of these images. So if so, raise your hands, we'll get you up here. ⁓
Ben Callahan (34:34)
And I know somebody else raised a hand and I missed it. if you want to raise your hand again, too, we'll get you into the combo. either of you want to... Okay, Jesse, jump in,
ep068 p2 deep dive group (34:34)
And I know somebody else raised a hand and I missed it. if you want to raise your hand again, too, we'll get you into the combo. either of you want to... Okay, Jesse, jump in,
ep068 p2 deep dive screenshare (34:34)
And I know somebody else raised a hand and I missed it. if you want to raise your hand again too, we'll get you into the combo. either of you want to, okay, Jesse jump in buddy.
Ben Callahan (34:46)
Yeah, so I put in the chat, I think one of the things that I'm trying to figure out now is we have like two MCP servers. We've been using Code Connect to connect our code components to parity with our Figma components. And that's great because our teams, a lot of the folks on our teams are product designers in Figma. So that was sort of, we initially built that out as a way to make it easy for developers to understand if this is here,
ep068 p2 deep dive screenshare (34:46)
Yeah, so I put in the chat, I think one of the things that I'm trying to figure out now is we have like two MCP servers. We've been using Code Connect to connect our code components to parity with our Figma components. And that's great because our teams, a lot of the folks on our teams are product designers in Figma. So that was sort of, we initially built that out as a way to make it easy for developers to understand if this is here,
ep068 p2 deep dive group (34:46)
Yeah, so I put in the chat, think one of the things that I'm trying to figure out now is we have like two MCP servers. We've been using Code Connect to connect our code components to parity with our Figma components. And that's great because our teams, a lot of the folks on our teams are product designers in Figma. So that was sort of, we initially built that out as a way to make it easy for developers to understand if this is here,
Ben Callahan (35:16)
you can then use our code components. So we initially built it through that, but then lo and behold, Figma MCP server makes it so that, you you could pass those code connect snippets in and it had to do less generative work. It was pulling in code snippets. Great. So that was sort of the pitch behind that. But because we use web components there, ⁓ and we were doing JS doc, it was really easy. There's tools out there that make it really easy to pull in the custom element manifest.
ep068 p2 deep dive screenshare (35:16)
you can then use our code components. So we initially built it through that, but then lo and behold, Figma MCP server makes it so that, you you could pass those code connect snippets in and it had to do less generative work. It was pulling in code snippets. Great. So that was sort of the pitch behind that. But because we use web components there, ⁓ and we were doing JS doc, it was really easy. There's tools out there that make it really easy to pull in the custom element manifest
ep068 p2 deep dive group (35:16)
you can then use our code components. So we initially built it through that, but then lo and behold, Figma MCP server makes it so that, you you could pass those code connect snippets in and it had to do less generative work. It was pulling in code snippets. Great. So that was sort of the pitch behind that. But because we use web components there, ⁓ and we were doing JS doc, it was really easy. There's tools out there that make it really easy to pull in the custom element manifest.
ep068 p2 deep dive screenshare (35:43)
and serve that up as an MCP server.
Ben Callahan (35:43)
and serve that up as an MCP server.
ep068 p2 deep dive group (35:43)
and serve that up as an MCP server.
So that unlocked a ton and we converted, we were just using CSS variables, we switched over to using the actual latest spec and using Torazzo to export everything. So we got that all in order and started feeding that into the MCP server. And that has been a big unlock because it kind of helped me understand better, it's the context thing. You're essentially giving,
Ben Callahan (35:45)
So that unlocked a ton and we converted, we were just using CSS variables, we switched over to using the actual latest spec and using Torazzo to export everything. So we got that all in order and started feeding that into the MCP server. And that has been a big unlock because it kind of helped me understand better, it's the context thing. You're essentially giving,
ep068 p2 deep dive screenshare (35:46)
So that unlocked a ton ⁓ and we converted we were just using CSS variables we switched over to using the actual latest spec and using terrazzo to to you know export everything so we got that all in order and started feeding that into the MCP server and That has been a big unlock because it kind of helped me understand better You're saying it's the context thing. You're essentially giving
really
Ben Callahan (36:13)
really
ep068 p2 deep dive group (36:13)
really
Ben Callahan (36:14)
targeted context to the AI. And so I did a thing, I wrote about it on LinkedIn, but I basically said, what are the top 40 interaction patterns? I described that like small, singular, but made up of more than one component. ⁓ What are the top 40 of those on New York state sites? And it's like, if you're registering household members for an application or if you're logging in or whatever. So I did that.
ep068 p2 deep dive group (36:14)
targeted context to the AI. And so I did a thing, I wrote about it on LinkedIn, but ⁓ I basically said, what are the top 40 interaction patterns? I described that like small, singular, but made up of more than one component. What are the top 40 of those on New York state sites? And it's like, if you're registering household members for an application or if you're logging in or whatever. So I did that.
ep068 p2 deep dive screenshare (36:14)
targeted context to the AI. And so I did a thing, I wrote about it on LinkedIn, but ⁓ I basically said, what are the top 40 interaction patterns? I described that like small, singular, but made up of more than one component. ⁓ What are the top 40 of those on New York State sites? And it's like, if you're registering household members for an application or if you're logging in or whatever. So I did that.
And then I said, build those out as individual pages.
Ben Callahan (36:41)
And then I said, build those out as individual pages
ep068 p2 deep dive group (36:41)
And then I said, build those out as individual pages
Ben Callahan (36:44)
using the design system MCP server. If you can't find a component that works, use our tokens. And then at the end of that, if you have to build something, run a gap analysis and identify what components you built and then sort of document that for us and explain that. And that was really helpful because then I exported those all to FigJam and then I gave it to our designers and they said, take a look at this, what works, what doesn't. And it was a way for us to audition.
ep068 p2 deep dive group (36:44)
using the design system MCP server. If you can't find a component that works, use our tokens. And then at the end of that, if you have to build something, run a gap analysis and identify what components you built and then sort of document that for us and explain that. And that was really helpful because then I exported those all to FigJam and then I gave it to our designers and they said, take a look at this, what works, what doesn't. And it was a way for us to audition.
ep068 p2 deep dive screenshare (36:44)
using the design system MCP server. If you can't find a component that works, use our tokens. And then at the end of that, if you have to build something, run a gap analysis and identify what components you built and then sort of document that for us and explain that. And that was really helpful because then I exported those all to FigJam and then I gave it to our designers and they said, take a look at this, what works, what doesn't. And it was a way for us to audition.
Ben Callahan (37:10)
So that's cool and that's what kind of brought me here, but it's kind of a manual process because the reviewing and refining is people going, well, that thing looks really good. Maybe we should turn this into a component, which is probably gonna require some Figma work. So I'm really curious TJ about the, you know, going from like what's in the code MCP server up to Figma. And I'm not sure because we have kind of both existing in manually synced ways.
ep068 p2 deep dive screenshare (37:10)
So that's cool and that's what kind of brought me here, but it's kind of a manual process because the reviewing and refining is people going, well, that thing looks really good. Maybe we should turn this into a component, which is probably gonna require some Figma work. So I'm really curious TJ about the, you know, going from like what's in the code MCP server up to Figma. And I'm not sure because we have kind of both existing in manually synced ways.
ep068 p2 deep dive group (37:10)
So that's cool and that's what kind of brought me here, but it's kind of a manual process because the reviewing and refining is people going, well, that thing looks really good. Maybe we should turn this into a component, which is probably gonna require some Figma work. So I'm really curious TJ about the, you know, going from like what's in the code MCP server up to Figma. And I'm not sure because we have kind of both existing in manually synced ways.
Ben Callahan (37:39)
And we are much more on the side where we have not that many designers, tons of engineers at the state. So anyway, that's a little bit of what we've learned. the process of like having it go out and build a bunch of stuff, even before we have ⁓ like, it's not going to production, it's just building so that we can audition and see, we just noticed we didn't catch that the select is shorter than the input. When you put them side by side, you see that. Stuff like that, where you get to identify that.
ep068 p2 deep dive screenshare (37:39)
And we are much more on the side where we have not that many designers, tons of engineers at the state. anyway, that's a little bit of what we've learned. the process of like having it go out and build a bunch of stuff, even before we have ⁓ like, it's not going to production, it's just building so that we can audition and see, we just noticed we didn't catch that the select is shorter than the input. When you put them side by side, you see that. Stuff like that, where you get to identify that.
ep068 p2 deep dive group (37:39)
And we are much more on the side where we have not that many designers, tons of engineers at the state. anyway, that's a little bit of what we've learned. the process of like having it go out and build a bunch of stuff, even before we have ⁓ like, it's not going to production, it's just building so that we can audition and see, we just noticed we didn't catch that the select is shorter than the input. When you put them side by side, you see that. Stuff like that, where you get to identify that.
Ben Callahan (38:09)
And it allows then our team to be able to see that and make much more informed decisions. Sometimes when you're working on the design system team, it's very abstract. And so being able to see all of this stuff built out in very realistic ways has been helpful. But in the review and refine part, like that's the, that's the human part, right? Yeah. And it is refuel. It's reviewing and refining coded components.
ep068 p2 deep dive screenshare (38:09)
And it allows then our team to be able to see that and make much more informed decisions. Sometimes when you're working on the design system team, it's very abstract. And so being able to see all of this stuff built out in very realistic ways has been helpful. But in the review and refine part, like that's the, that's the human part, right? Yeah. And it is refuel, it's reviewing and refining coded components.
ep068 p2 deep dive group (38:09)
And it allows then our team to be able to see that and make much more informed decisions. Sometimes when you're working on the design system team, it's very abstract. And so being able to see all of this stuff built out in very realistic ways ⁓ has been helpful. in the review and refine part, like that's the, that's the human part, right? Yeah. And it is refuel. It's reviewing and refining coded components.
Ben Callahan (38:38)
It's reviewing and refining what AI generated and saying, so for example, it's like this component sucks in this way. Or it was, for example, our stepper was not clear about the layout because you have to use layout classes in order to achieve the side that collapses. So that's something we're like, we need to refine how that, what's there. But sometimes it's not components. For example, we don't have a specific guidance around errors. How should errors be reported on forms? There's no specific component around that.
ep068 p2 deep dive screenshare (38:38)
It's reviewing and refining what AI generated and saying, so for example, it's like this component sucks in this way. Or it was, for example, our stepper was not clear about the layout because you have to use layout classes in order to achieve the side that collapses. So that's something we're like, we need to refine how that, what's there. But sometimes it's not components. For example, we don't have a specific guidance around errors. How should errors be reported on forms? There's no specific component around that.
ep068 p2 deep dive group (38:38)
It's reviewing and refining what AI generated and saying, so for example, it's like this component sucks in this way. Or it was, for example, our stepper was not clear about the layout because you have to use layout classes in order to achieve the side that collapses. So that's something we're like, we need to refine how that, what's there. But sometimes it's not components. For example, we don't have a specific guidance around errors. How should errors be reported on forms? There's no specific component around that.
Ben Callahan (39:07)
So then was like, I should probably create like an endpoint that says when you're doing error validation or when you're reporting errors, put it at the top and list them out and they should link you to directly where the error is and then jump back up. So that's a piece of design guidance really. So I've been trying to help. Right, I'm trying to help our team, design team better understand what it means to go from something where you might.
ep068 p2 deep dive screenshare (39:07)
So then was like, I should probably create like an endpoint that says when you're doing error validation or when you're reporting errors, put it at the top and list them out and they should link you to directly where the error is and then jump back up. So that's a piece of design guidance really. So I've been trying to help. Right, I'm trying to help our team, design team better understand what it means to go from something where you might.
ep068 p2 deep dive group (39:07)
So then was like, I should probably create like an endpoint that says when you're doing error validation or when you're reporting errors, put it at the top and list them out and they should link you to directly where the error is and then jump back up. So that's a piece of design guidance really. So I've been trying to Right, I'm trying to help our team, design team better understand what it means to go from something where you might.
Ben Callahan (39:34)
sketch it out to how do we articulate that in a like a rule sort of documented way. ⁓ and we should talk after this too, because like we similar things, like this is where we kind of get into like the intent and purpose of a lot of these, like the compositions of structures are depending whatever it is that you're trying to tackle. This was something that we did with story UI a while ago when we were trying to get the story UI MCP to generate things using those components.
ep068 p2 deep dive screenshare (39:34)
sketch it out to how do we articulate that in a like a rule sort of documented way. ⁓ and we should talk after this too, because like we similar things, like this is where we kind of get into like the intent and purpose of a lot of these, like the compositions of structures are depending whatever it is that you're trying to tackle. This was something that we did with story UI a while ago when we were trying to get the story UI MCP to generate things using those components.
ep068 p2 deep dive group (39:34)
sketch it out to how do we articulate that in a like a rule sort of documented way. ⁓ and we should talk after this too, because like we similar things, like this is where we kind of get into like the intent and purpose of a lot of these like the compositions of structures or depending whatever it is that you're trying to tackle. This was something that we did with story UI a while ago when we were trying to get the story UI MCP to generate things using those components.
Ben Callahan (40:02)
But the AI just knew it was a bunch of components. And then unless that component library had layout components, which a lot don't, some of them just will use Tailwind, Grid, and Flex or whatever to put these things together, the AI doesn't know that. It knows it exists, but it doesn't know what your preference is. So the Story UI MCP has two things that it does. It's one that it looks for a docs directory that usually would contain all your markdown files that are not AI specific, but just
ep068 p2 deep dive screenshare (40:02)
But the AI just knew it was a bunch of components. And then unless that component library had layout components, which a lot don't, some of them just will use Tailwind, Grid, and Flex or whatever to put these things together, the AI doesn't know that. It knows it exists, but it doesn't know what your preference is. So the Story UI MCP has two things that it does. It's one that it looks for a docs directory that usually would contain all your markdown files that are not AI specific, but just
ep068 p2 deep dive group (40:02)
But the AI just knew it was a bunch of components. And then unless that component library had layout components, which a lot don't, some of them just will use Tailwind, Grid, and Flex or whatever to put these things together, the AI doesn't know that. It knows it exists, but it doesn't know what your preference is. So the Story UI MCP has two things that it does. It's one that it looks for a docs directory that usually would contain all your markdown files that are not AI specific, but just
Ben Callahan (40:29)
documentation for a traditional design system. So it'll have guidelines and usage and governance and all those things, which is good to know. It definitely helps when it comes to like creating layouts and other component structures and inventories. But then there's another file that we have called story UI considerations. And then that might be something similar that would help you there because it tells the AI.
ep068 p2 deep dive screenshare (40:29)
documentation for a traditional design system. So it'll have guidelines and usage and governance and all those things, which is good to know. It definitely helps when it comes to like creating layouts and other component structures and inventories. But then there's another file that we have called story UI considerations. And then that might be something similar that would help you there because it tells the AI.
ep068 p2 deep dive group (40:29)
documentation for a traditional design system. So it'll have guidelines and usage and governance and all those things, which is good to know. It definitely helps when it comes to like creating layouts and other component structures and inventories. But then there's another file that we have called story UI considerations. And then that might be something similar that would help you there because it tells the AI.
things that it should know when thinking about how to assemble components, whether it's like tiny little primitive component structures or whether it's like, you know, inventory dashboard type things. So it'll say preference is grid over, you know, like it'll in the component inventory manifest, because we'll do a similar thing where that's kind of how it does this enhanced discovery on the design system components within that component library. And it'll understand that's where it gathers all the props from the TypeScript files. And so
Ben Callahan (40:52)
things that it should know when thinking about how to assemble components, whether it's like tiny little primitive component structures or whether it's like, you know, inventory dashboard type things. So it'll say preference is grid over, ⁓ you know, like it'll in the component inventory manifest, because we'll do a similar thing where that's kind of how it does this enhanced discovery on the design system components within that component library. And it'll understand that's where it gathers all the props from the TypeScript files. And so
ep068 p2 deep dive screenshare (40:52)
things that it should know when thinking about how to assemble components, whether it's like tiny little primitive component structures or whether it's like, you know, inventory dashboard type things. So it'll say preference is grid over, ⁓ you know, like it'll in the component inventory manifest, because we'll do a similar thing where that's kind of how it does this enhanced discovery on the design system components within that component library. And it'll understand that's where it gathers all the props from the TypeScript files. And so
Ben Callahan (41:22)
if it doesn't see a layout component, then it's going to default to whatever it thinks is best unless it reads in that considerations file that I prefer flex over grid or I prefer left, float left or whatever it might be. Do you run into context bloat with that docs approach? Cause that's another thing that we've been thinking about. I've been watching the context countering in cloud code and trying to like evaluate, Oh, I'm going to say no percent. Yeah. It's a, so definitely using cloud code.
ep068 p2 deep dive screenshare (41:22)
if it doesn't see a layout component, then it's going to default to whatever it thinks is best unless it reads in that considerations file that I prefer flex over grid or I prefer left, float left or whatever it might be. Do you run into context bloat with that docs approach? Cause that's another thing that we've been thinking about. I've been watching the context countering in cloud code and trying to like evaluate, Oh, I'm going to say no percent. Yeah. It's a, so definitely using cloud code.
ep068 p2 deep dive group (41:22)
If it doesn't see a layout component, then it's going to default to whatever it thinks is best unless it reads in that considerations file that I prefer flex over grid or I prefer left, float left or whatever it might be. Do you run into context bloat with that docs approach? Because that's another thing that we've been thinking about. I've been watching the context counter in in cloud code and trying to like evaluate. ⁓ I'm going to say no percent. Yeah, it's ⁓ so definitely using cloud code.
Ben Callahan (41:51)
helps with that for layout things. This is before a son of 4.6 came out because it seems to be much better. But these are task oriented things. It already has the information that it needs from the discovery. It has discovery that's built up in cache and memory. It has that so it doesn't have to do as much searching and finding and understanding upfront before you ask it anything. That initializes once you start it up. When you prompt it,
ep068 p2 deep dive screenshare (41:51)
helps with that for layout things. This is before a son of 4.6 came out because it seems to be much better. But these are task oriented things. It already has the information that it needs from the discovery. It has discovery that's built up in cache and memory. It has that so it doesn't have to do as much searching and finding and understanding upfront before you ask it anything. That initializes once you start it up. When you prompt it,
ep068 p2 deep dive group (41:51)
helps with that for layout things. And this is before a son of 4.6 came out because it seems to be much better. But these are task oriented things. It already has the information that it needs from the discovery. So it has discovery that's built up in cache and memory. So it has that so it doesn't have to do as much searching and finding and understanding upfront before you ask it anything. So that initializes once you start it up. So when you prompt it,
Ben Callahan (42:20)
It'll run through that consideration sets already been cached and then no. So usually it'll spit out whatever the prompt is, depending on how complex it is. We have a couple of instances where we're like recreate the LinkedIn homepage and it took about 45 seconds for it to generate. But for the most part, it gets it right. But as long what we'll do is create the context and then ask it to refine the context to make it as ⁓ concise, but as impactful as you can. And then I found making allowing it to
ep068 p2 deep dive screenshare (42:20)
It'll run through that consideration sets already been cached and then no. So usually it'll spit out whatever the prompt is, depending on how complex it is. We have a couple of instances where we're like recreate the LinkedIn homepage and it took about 45 seconds for it to generate. But for the most part, it gets it right. But as long what we'll do is create the context and then ask it to refine the context to make it ⁓ as ⁓ concise, but as impactful as you can. And then I found making allowing it to
ep068 p2 deep dive group (42:20)
It'll run through that consideration sets already been cached and then no. So usually it'll spit out whatever the prompt is, depending on how complex it is. We have a couple of instances where we're like recreate the LinkedIn homepage and it took about 45 seconds for it to generate. But for the most part, it gets it right. But as long what we'll do is create the context and then ask it to refine the context to make it as ⁓ concise, but as impactful as you can. And then I found making allowing it to
Ben Callahan (42:50)
it to tell me what it thinks is best for it to be able to read, consume, and execute on seems to help better than me. usually it starts out with me dictating. I'll use Whisper Flow and I'll say, we prefer to do this. This is the way we want to use icons. This is the way we want to import things. And then I'll say, all right, now review the code base and then put this in a very concise, usually a little like format it need to mark down to XML.
ep068 p2 deep dive screenshare (42:50)
it to tell me what it thinks is best for it to be able to read, consume, and execute on seems to help better than me. usually it starts out with me dictating. I'll use Whisper Flow and I'll say, we prefer to do this. This is the way we want to use icons. This is the way we want to import things. And then I'll say, all right, now review the code base and then put this in a very concise, usually a little like format it need to mark down to XML.
ep068 p2 deep dive group (42:50)
it to tell me what it thinks is best for it to be able to read, consume, and execute on seems to help better than me. usually it starts out with me dictating. I'll use Whisper Flow and I'll say, we prefer to do this. This is the way we want to use icons. This is the way we want to import things. And then I'll say, all right, now review the code base and then put this in a very concise, usually a little like format it need to mark down to XML.
Ben Callahan (43:15)
And then, and then that's what it'll say, all right, I've got I've condensed, you know, the 7500 word characters into 2500 characters, and then it'll consume that those considerations to memory and then know that every prompt, it'll automatically use those rules. It's skills might help to in a situation like this, if it's like a repetitive thing. It feels like we're just like, finding ways to do the work we should have been doing all along in terms of articulating.
ep068 p2 deep dive screenshare (43:15)
And then, and then that's what it'll say, all right, I've got I've condensed, you know, the 7500 word characters into 2500 characters, and then it'll consume that those considerations to memory and then know that every prompt, it'll automatically use those rules. It's skills might help to in a situation like this, if it's like a repetitive thing. It feels like we're just like, finding ways to do the work we should have been doing all along in terms of articulating.
ep068 p2 deep dive group (43:15)
And then and then that's what it'll say. All right. I've got I've condensed, you know, the 7500 word characters into 2500 characters, and then it'll consume that those considerations to memory and then know that every prompt, it'll automatically use those rules. It's skills might help to in a situation like this, if it's like a repetitive thing. It feels like we're just like, finding ways to do the work we should have been doing all along in terms of articulating.
Ben Callahan (43:44)
Yeah. Our design, know, like our intent. It's the unsexy stuff that should have been there. That's right. That's right. All right. This is awesome. We've only got a few minutes left, so I want to get Doug and Jenny both in. Doug, what's up? I just have a really quick question. Is Storybook help at all in this process or does it just get in the way? I'm hearing most people mentioning TypeScript in the source directly. Can it help or is it just in the way and it should be bypassed for this type of stuff?
ep068 p2 deep dive screenshare (43:44)
Yeah. Our design, know, like our intent. It's the unsexy stuff that should have been there. That's right. That's right. All right. This is awesome. We've only got a few minutes left, so I want to get Doug and Jenny both in. Doug, what's up? I just have a really quick question. Is Storybook help at all in this process or does it just get in the way? I'm hearing most people mentioning TypeScript in the source directly. Can it help or is it just in the way and it should be bypassed for this type of stuff?
ep068 p2 deep dive group (43:44)
Yeah. Our design, know, like our intent. It's the unsexy stuff that should have been there. That's right. That's right. All right. This is awesome. We've only got a few minutes left, so I want to get Doug and Jenny both in. Doug, what's up? I just have a really quick question. Is Storybook help at all in this process or does it just get in the way? I'm hearing most people mentioning TypeScript in the source directly. Can it help or is it just in the way and it should be bypassed for this type of stuff?
Ben Callahan (44:12)
Question for me, a storybook is a key integral part of nearly every design system project I'm on, only because we've built tools that we leverage to help accelerate our efficiencies and optimize client communication and the way to do QA and check accessibility. And then we have AI tools built in-story UIs, like one of the critical things that we use for our projects. So storybook.
ep068 p2 deep dive screenshare (44:12)
Question for me, a storybook is a key integral part of nearly every design system project I'm on, only because we've built tools that we leverage to help accelerate our efficiencies and optimize client communication and the way to do QA and check accessibility. And then we have AI tools built in-story UIs, like one of the critical things that we use for our projects. So storybook.
ep068 p2 deep dive group (44:12)
Question for me, a storybook is a key integral part of nearly every design system project I'm on, only because we've built tools that we leverage to help accelerate our efficiencies and optimize client communication and the way to do QA and check accessibility. And then we have AI tools built in-story UIs, like one of the critical things that we use for our projects. So storybook.
Ben Callahan (44:36)
is a big part of what we do. And recently, we've started to use the Storybook MCP with a lot of projects too, like completely bypassing Story UI, depending on the team that we're working with. If it's engineers, then usually we'll just go with Storybook MCP and then we'll work with them. But if it's mostly product designers or product managers who want to be able to quickly iterate on things, then we'll install Story UI and let them play with it so that way they don't have to mess with installing IDE or anything.
ep068 p2 deep dive screenshare (44:36)
is a big part of what we do. And recently, we've started to use the Storybook MCP with a lot of projects too, like completely bypassing Story UI, depending on the team that we're working with. If it's engineers, then usually we'll just go with Storybook MCP and then we'll work with them. But if it's mostly product designers or product managers who want to be able to quickly iterate on things, then we'll install Story UI and let them play with it so that way they don't have to mess with installing IDE or anything.
ep068 p2 deep dive group (44:36)
is a big part of what we do. And recently, we've started to use the Storybook MCP with a lot of projects too, like completely bypassing Story UI, depending on the team that we're working with. If it's engineers, then usually we'll just go with Storybook MCP and then we'll work with them. But if it's mostly product designers or product managers who want to be able to quickly iterate on things, then we'll install Story UI and let them play with it so that way they don't have to mess with installing IDE or anything.
Ben Callahan (45:05)
Is that the official one that's in preview that you're talking about, or is this the other, like the community Storybook MCP? There's two different things. One is the community, the one that was recently released through Storybook and Chromatic, the Storybook MCP. look for it. Yeah, it's great. Highly recommend checking it out. Cool. Thank you. Yeah. Awesome. Jenny, what's going on? I see you dropping all kinds of cool stuff in the Fig Jam for us. Hey. So happy to share. We just pitched.
ep068 p2 deep dive screenshare (45:05)
Is that the official one that's in preview that you're talking about, or is this the other, like the community Storybook MCP? There's two different things. One is Storybook. It's the community, the one that was recently released through Storybook and Chromatic, the Storybook MCP. look for it. Yeah, it's great. Highly recommend checking it out. Cool. Thank you. Yeah. Awesome. Jenny, what's going on? I see you dropping all kinds of cool stuff in the Fig Jam for us. Hey. So happy to share. We just pitched.
ep068 p2 deep dive group (45:05)
Is that the official one that's in preview that you're talking about, or is this the other, like the community Storybook MCP? There's two different things. One is Storybook. It's the community, the one that was recently released through Storybook and Chromatic, the Storybook MCP. a look for it. Yeah, it's great. Highly recommend checking it out. Cool. Thank you. Yeah. Awesome. Jenny, what's going on? I see you dropping all kinds of cool stuff in the Fig Jam for us. Hey. So happy to share. We just pitched.
ep068 p2 deep dive screenshare (45:35)
design systems as AI infrastructure two weeks ago because we've been piloting our MCP server and it's working a lot. So I would say we before just put like we're in a monorail, like a massive one and people were just accessing and asking, you know, how do tokens work? How do all the typical questions you're asking while trying to generate UI.
ep068 p2 deep dive group (45:35)
design systems as AI infrastructure two weeks ago because we've been piloting our MCP server and it's working a lot. So I would say we before just put like we're in a monorail, like a massive one and people were just accessing and asking, you know, how do tokens work? How do all the typical questions you're asking while trying to generate UI.
Ben Callahan (45:35)
design ⁓ systems as AI infrastructure two weeks ago because we've been piloting our MCP server and it's working a lot. So I would say we before just put like we're in a monorail, like a massive one and people were just accessing and asking, you know, how do tokens work? How do all the typical questions you're asking while trying to generate UI.
And then it was, we would find that it would just hallucinate and make up stuff. So we have been doing some research and then realized when you package all the things you have into an MCP server, it becomes the like source of truth, like the actual context. it's not guessing because it's, you're, creating the like actions to query and get your sources of truth here on the left side. So.
ep068 p2 deep dive screenshare (46:05)
And then it was, we would find that it would just hallucinate and make up stuff. So we have been doing some research and then realized when you package all the things you have into an MCP server, it becomes the like source of truth, like the actual context. it's not guessing because it's, you're, creating the like actions to query and get your sources of truth here on the left side. So.
ep068 p2 deep dive group (46:05)
And then it was, we would find that it would just hallucinate and make up stuff. So we have been doing some research and then realized when you package all the things you have into an MCP server, it becomes the like source of truth, like the actual context. it's not guessing because it's, you're, creating the like actions to query and get your sources of truth here on the left side. So.
You've codified everything, your docs, examples, code, styles, everything. And then we're basically creating structured actions and reliable ways to get those answers. So the LLM is not guessing any time. So it's proven out to be very accurate to help us. And then if you go to the next one, we have a whole bunch of teams trying all kinds of like agents, AI tools, right? So.
Ben Callahan (46:35)
You've codified everything, your docs, examples, code, styles, everything. And then we're basically creating structured actions and reliable ways to get those answers. So the LLM is not guessing any time. So it's proven out to be very accurate to help us. And then if you go to the next one, we have a whole bunch of teams trying all kinds of like agents, AI tools, right? So.
ep068 p2 deep dive screenshare (46:35)
You've codified everything, your docs, examples, code, styles, everything. And then we're basically creating structured actions and reliable ways to get those answers. So the LLM is not guessing any time. So it's proven out to be very accurate to help us. And then if you go to the next one, we have a whole bunch of teams trying all kinds of like agents, AI tools, right? So.
Ben Callahan (47:04)
how do we meet them all in their workflows is to like, this is like the agnostic thing that they can connect into to go through all these workflows. And now we're on this rampage of like creating all these like skills. So it's kind of crazy. have design linting skills, accessibility, visual design, interaction design, coding.
ep068 p2 deep dive screenshare (47:04)
how do we meet them all in their workflows is to like, this is like the agnostic thing that they can connect into to go through all these workflows. And now we're on this rampage of like creating all these like skills. So it's kind of crazy. have design linting skills, accessibility, visual design, interaction design, coding.
ep068 p2 deep dive group (47:04)
How do we meet them all in their workflows is to like, this is like the agnostic thing that they can connect into to go through all these workflows. And now we're on this rampage of like creating all these like skills. So it's kind of crazy. have design linting skills, accessibility, visual design, interaction design, coding,
like all these.
Ben Callahan (47:31)
Like all these,
ep068 p2 deep dive screenshare (47:31)
Like all these,
ep068 p2 deep dive group (47:32)
Anything you can think of, like something that you can do that you want to multiply to others, how can you help them bundle that up into do those things is how we're kind of packaging up the skills into these, I guess, bundles to power these workflows. So we're just investing in proving out these four workflows right now end to end
Ben Callahan (47:32)
anything you can think of, like something that you can do that you want to multiply to others. How can you help them bundle that up and to do those things is how we're kind of packaging up the skills into these, I guess, bundles to power these workflows. So we're just investing in proving out these four workflows right now end to end.
ep068 p2 deep dive screenshare (47:32)
anything you can think of, like something that you can do that you want to multiply to others. How can you help them bundle that up and to do those things is how we're kind of packaging up the skills into these, I guess, bundles to power these workflows. So we're just investing in proving out these four workflows right now end to end.
ep068 p2 deep dive group (48:01)
and
Ben Callahan (48:01)
And
ep068 p2 deep dive screenshare (48:01)
And
ep068 p2 deep dive group (48:02)
It's surprisingly working for us. This is really cool. This is awesome. ⁓ I'm guessing that, well, maybe you can answer this question as a follow-up, Jenny. What work did you have to do in the Figma, the code, the docs to get this to be serving as the context well? Was there a lot of cleanup there? What was that process for you?
Ben Callahan (48:03)
it's surprisingly working for us. This is really cool. This is awesome. ⁓ I'm guessing that, well, maybe you can answer this question as a follow-up, Jenny. What work did you have to do in the Figma, the code, the docs to get this to be serving as the context well? Was there a lot of cleanup there? What was that process for you?
ep068 p2 deep dive screenshare (48:03)
it's surprisingly working for us. This is really cool. This is awesome. ⁓ I'm guessing that, well, maybe you can answer this question as a follow-up, Jenny. What work did you have to do in the Figma, the code, the docs to get this to be serving as the context well? Was there a lot of cleanup there? What was that process for you?
ep068 p2 deep dive group (48:30)
Yeah. I mean, if you remember when I was on last year, I basically said everything here was a mess. So we've been rebuilding everything, writing all the docs, redoing all the components, tokens, everything. It's like rebuild everything from scratch and is super structured. So I've been saying structure is your strategy. That becomes your infrastructure in your context. So it can understand the patterns built from there. love that.
Ben Callahan (48:30)
Yeah. I mean, if you remember when I was on last year, I basically said everything here was a mess. So we've been rebuilding everything, writing all the docs, redoing all the components, tokens, everything. It's like rebuild everything from scratch and is super structured. So I've been saying structure is your strategy. That becomes your infrastructure in your context. So it can understand the patterns. Yeah. And build from there. I love that.
ep068 p2 deep dive screenshare (48:30)
Yeah. I mean, if you remember when I was on last year, I basically said everything here was a mess. So we've been rebuilding everything, writing all the docs, redoing all the components, tokens, everything. It's like rebuild everything from scratch and is super structured. So I've been saying structure is your strategy. That becomes your infrastructure in your context. So it can understand the patterns built from there. love that.
ep068 p2 deep dive group (49:01)
I'm going to put structure is your strategy. This is our t-shirt for the day. TJ, what do you think of this? How's this sitting with you?
Ben Callahan (49:01)
I'm gonna put structure is your strategy. This is our t-shirt for the day. TJ, what do you think of this? How's this sitting with you?
ep068 p2 deep dive screenshare (49:01)
I'm going to put structure is your strategy. This is our t-shirt for the day. TJ, what do you think of this? How's this sitting with you?
ep068 p2 deep dive group (49:13)
you're muted, buddy.
Ben Callahan (49:13)
you're muted, buddy.
ep068 p2 deep dive screenshare (49:13)
you're muted, buddy.
ep068 p2 deep dive group (49:19)
I was going to say the use of skills as something that I feel like we need to dig further into the, the, that we feel like we felt like we needed a skill for, we'd find some other like, like reason why it's hard to determine like what is skill because skill is kind of deterministic. Like it's like, you need to find repeatable patterns in order to create this skill. The hard part was finding those repeatable patterns for
ep068 p2 deep dive screenshare (49:19)
I was going to say the use of skills as something that I feel like we need to dig further into the, the, that we feel like we felt like we needed a skill for, we'd find some other like, like reason why it's hard to determine like what is skill because skill is kind of deterministic. Like it's like, you need to find repeatable patterns in order to create this skill. The hard part was finding those repeatable patterns for
Ben Callahan (49:19)
I was going to say the use of skills as something that I feel like we need to dig further into the, the, that we feel like we felt like we needed a skill for, we'd find some other like, like reason why it's hard to determine like what is skill because skill is kind of deterministic. Like it's like, you need to find repeatable patterns in order to create this skill. The hard part was finding those repeatable patterns for
ep068 p2 deep dive group (49:48)
but we do agency client work. it was difficult for us to kind of pinpoint one specific suite of issues that we can identify and create skills for. I use them for personal things, but what we've usually deferred to in these where you have skills would be custom agents. And so if there's things where we need to allow AI to be autonomous, and so in that sense, the AI could actually do things like create
ep068 p2 deep dive screenshare (49:48)
but we do agency client work. it was difficult for us to kind of pinpoint one specific suite of issues that we can identify and create skills for. I use them for personal things, but what we've usually deferred to in these where you have skills would be custom agents. And so if there's things where we need to allow AI to be autonomous, and so in that sense, the AI could actually do things like create
Ben Callahan (49:48)
but we do agency client work. it was difficult for us to kind of pinpoint one specific suite of issues that we can identify and create skills for. I use them for personal things, but what we've usually deferred to in these where you have skills would be custom agents. And so if there's things where we need to allow AI to be autonomous, and so in that sense, the AI could actually do things like create
ep068 p2 deep dive group (50:17)
scripts that a skill would be, but on its own. So we would allow those agents to be able to do those types of things. And it would still be called depending on similar to how skills are called, depending on the need. I do, if you have a use for these types of skills, then I think that's great. And I think we actually need to dabble in a little bit more. We use skills internally for a lot of operational things too, that are repeatable, like lot of proposals and RFP and operations type things.
ep068 p2 deep dive screenshare (50:17)
scripts that a skill would be, but on its own. So we would allow those agents to be able to do those types of things and it would still be called depending on similar to how skills are called ⁓ depending on the need. But I do, if you have a use for these types of skills, then I think that's great. And I think we actually need to dabble in a little bit more. We use skills internally for a lot of operational things too that are repeatable, like lot of proposals and RFP and operations type things.
Ben Callahan (50:17)
scripts that a skill would be, but on its own. So we would allow those agents to be able to do those types of things and it would still be called depending on similar to how skills are called ⁓ depending on the need. But I do, if you have a use for these types of skills, then I think that's great. And I think we actually need to dabble in a little bit more. We use skills internally for a lot of operational things too that are repeatable, like lot of proposals and RFP and operations type things.
ep068 p2 deep dive group (50:46)
But I like this. I like I like this thinking. Yeah. Incredible. ⁓ This is so y'all. This is so good. I feel like we did have a chance to get a little deeper here. It still feels like we could go deeper, of course, as always. Thank you for being so active in the Fig Jam. My goodness, this is great. I have to I'm going to have to go read all the notes that you all have dropped. ⁓ Please, if you have feedback or Ben, you could just use a eye. That's right. That's right. I'll do that. I'll do that. ⁓
ep068 p2 deep dive screenshare (50:46)
But I like this. I like I like this thinking. Yeah. Incredible. ⁓ This is so y'all. This is so good. I feel like we did have a chance to get a little deeper here. It still feels like we could go deeper, of course, as always. Thank you for being so active in the Fig Jam. My goodness, this is great. I have to I'm going to have to go read all the notes that you all have dropped. ⁓ Please, if you have feedback or Ben, you could just use a eye. That's right. That's right. I'll do that. I'll do that. ⁓
Ben Callahan (50:46)
But I like this. I like I like this thinking. Yeah. Incredible. ⁓ This is so y'all. This is so good. I feel like we did have a chance to get a little deeper here. It still feels like we could go deeper, of course, as always. Thank you for being so active in the Fig Jam. My goodness, this is great. I have to I'm going to have to go read all the notes that you all have dropped. ⁓ Please, if you have feedback or Ben, you could just use a eye. That's right. That's right. I'll do that. I'll do that. ⁓
ep068 p2 deep dive group (51:16)
If you have feedback on this whole thing, please, please let me know. Like where it's only February. I've got a year of these things. I'm booking out these calls. So I'd love to know how we can make these more valuable for you. Um, let me know, drop that over here. Um, ideas for future questions, digging deeper, drop it over here. And like I mentioned earlier, you know, Redwoods is open. This is the kind of thing we try to do every week in Redwood. So if this is your thing, this kind of collaborative learning, check it out.
ep068 p2 deep dive screenshare (51:16)
If you have feedback on this whole thing, please, please let me know. Like where it's only February. I've got a year of these things. I'm booking out these calls. So I'd love to know how we can make these more valuable for you. Um, let me know, drop that over here. Um, ideas for future questions, digging deeper, drop it over here. And like I mentioned earlier, you know, Redwoods is open. This is the kind of thing we try to do every week in Redwood. So if this is your thing, this kind of collaborative learning, check it out.
Ben Callahan (51:16)
If you have feedback on this whole thing, please, please let me know. Like where it's only February. I've got a year of these things. I'm booking out these calls. So I'd love to know how we can make these more valuable for you. Um, let me know, drop that over here. Um, ideas for future questions, digging deeper, drop it over here. And like I mentioned earlier, you know, Redwoods is open. This is the kind of thing we try to do every week in Redwood. So if this is your thing, this kind of collaborative learning, check it out.
ep068 p2 deep dive group (51:41)
⁓ Lots of stuff from the community has been posted here and then things from folks as well who are in the larger systems community. ⁓ And again, if you're looking for a new gig or if you have something you've shared recently you want to get the word out on, please do so. ⁓ I'll take a look at these celebrations. It's so cool to see folks are willing to share that stuff too. You all are awesome. Thank you. TJ, any final word for folks? ⁓ This is all fun stuff. I love.
ep068 p2 deep dive screenshare (51:41)
⁓ Lots of stuff from the community has been posted here and then things from folks as well who are in the larger systems community. ⁓ And again, if you're looking for a new gig or if you have something you've shared recently you want to get the word out on, please do so. ⁓ I'll take a look at these celebrations. It's so cool to see folks are willing to share that stuff too. You all are awesome. Thank you. TJ, any final word for folks? ⁓ This is all fun stuff. I love.
Ben Callahan (51:41)
⁓ Lots of stuff from the community has been posted here and then things from folks as well who are in the larger systems community. ⁓ And again, if you're looking for a new gig or if you have something you've shared recently you want to get the word out on, please do so. ⁓ I'll take a look at these celebrations. It's so cool to see folks are willing to share that stuff too. Y'all are awesome. Thank you. TJ, any final word for folks? ⁓ This is all fun stuff. I love.
ep068 p2 deep dive group (52:10)
I love it when a group comes up with really good questions and I love hearing about everybody else's problems and then how they arrived at the same or different solutions. I think that's awesome that we're all in this together. yeah, I just want to keep these conversations going. I think it's good for everybody. Yeah, I'll just say to remember that it's never too late to start trying to start experimenting, right? Like, it can definitely feel like
ep068 p2 deep dive screenshare (52:10)
I love it when a group comes up with really good questions and I love hearing about everybody else's problems and then how they arrived at the same or different solutions. I think that's awesome that we're all in this together. yeah, I just want to keep these conversations going. I think it's good for everybody. Yeah, I'll just say to remember that it's never too late to start trying to start experimenting, right? Like, it can definitely feel like
Ben Callahan (52:10)
I love it when a group comes up with really good questions and I love hearing about everybody else's problems and then how they arrived at the same or different solutions. I think that's awesome that we're all in this together. yeah, I just want to keep these conversations going. I think it's good for everybody. Yeah, I'll just say to remember that it's never too late to start trying to start experimenting, right? Like, it can definitely feel like
ep068 p2 deep dive group (52:36)
you're behind. Like I feel that every day, just to be honest. And ⁓ everybody probably does. And the reality is you're probably further, much further ahead than, many folks in who are not working in this space. Right. So don't let that keep you from just installing something and spending a half an hour experimenting, right? That experimentation is what will lead you down the path of discovering the next thing to learn. And that's what, that's what we need to be doing. So, ⁓ just being here puts you in the 1%.
ep068 p2 deep dive screenshare (52:36)
you're behind. Like I feel that every day, just to be honest. And ⁓ everybody probably does. And the reality is you're probably further, much further ahead than, many folks in who are not working in this space. Right. So don't let that keep you from just installing something and spending a half an hour experimenting, right? That experimentation is what will lead you down the path of discovering the next thing to learn. And that's what, that's what we need to be doing. So, ⁓ just being here puts you in the 1%.
Ben Callahan (52:36)
you're behind. Like I feel that every day, just to be honest. And ⁓ everybody probably does. And the reality is you're probably further, much further ahead than, many folks in who are not working in this space. Right. So don't let that keep you from just installing something and spending a half an hour experimenting, right? That experimentation is what will lead you down the path of discovering the next thing to learn. And that's what, that's what we need to be doing. So, ⁓ just being here puts you in the 1%.
ep068 p2 deep dive group (53:05)
It's true. It's true. Consider this your first step into trying to be curious, experiment, find things to... Somebody just put pencil that... Jess, you put that in there. Check it out. I think it's one of those cool, like, whoa, it did that moments. Yeah. Super easy to install. Yeah, that's it. And I just want to say, too, get connected with us, please, please, please. This is one of my favorite parts of every week is when somebody I don't know...
ep068 p2 deep dive screenshare (53:05)
It's true. It's true. Consider this your first step into trying to be curious, experiment, find things to... Somebody just put pencil that... Jess, you put that in there. Check it out. I think it's one of those cool, like, whoa, it did that moments. Yeah. Super easy to install. Yeah, that's it. And I just want to say, too, get connected with us, please, please, please. This is one of my favorite parts of every week is when somebody I don't know...
Ben Callahan (53:05)
It's true. It's true. Consider this your first step into trying to be curious, experiment, find things to... Somebody just put pencil that... Jess, you put that in there. Check it out. I think it's one of those cool, like, whoa, it did that moments. Yeah. Super easy to install. Yeah, that's it. And I just want to say, too, get connected with us, please, please, please. This is one of my favorite parts of every week is when somebody I don't know...
ep068 p2 deep dive group (53:34)
messages me on LinkedIn and is just like, hey, I watched this thing. I learned this. Thanks for doing it. And I can get a chance to sort of interact and hear a little bit about their journey. So I know TJ would feel the same. Just get connected. Don't be shy. Reach out, say hello on LinkedIn or wherever you're at, and we'll keep the convo going. You all rock. Thank you so much for being here. TJ, thanks for doing a part two so quickly, Really appreciate it. Thanks for having me. All right, everybody. Thank you all. We'll see you next week. Yep. Bye-bye.
ep068 p2 deep dive screenshare (53:34)
messages me on LinkedIn and is just like, hey, I watched this thing. I learned this. Thanks for doing it. And I can get a chance to sort of interact and hear a little bit about their journey. So I know TJ would feel the same. Just get connected. Don't be shy. Reach out, say hello on LinkedIn or wherever you're at, and we'll keep the convo going. You all rock. Thank you so much for being here. TJ, thanks for doing a part two so quickly, Really appreciate it. Thanks for having me. All right, everybody. Thank you all. We'll see you next week. Yep. Bye-bye.
Ben Callahan (53:34)
messages me on LinkedIn and is just like, hey, I watched this thing. I learned this. Thanks for doing it. And I can get a chance to sort of interact and hear a little bit about their journey. So I know TJ would feel the same. Just get connected. Don't be shy. Reach out, say hello on LinkedIn or wherever you're at, and ⁓ we'll keep the convo going. You all rock. Thank you so much for being here. TJ, thanks for doing a part two so quickly, Really appreciate it. Thanks for having me. All right, everybody. Thank you all. We'll see you next week. Yep. Bye-bye.
ep068 p2 deep dive screenshare (54:03)
Bye everybody.
ep068 p2 deep dive group (54:03)
Bye everybody.
Ben Callahan (54:03)
Bye everybody.
Ben Callahan (54:04)
Thank you so much for joining us on part two of episode 68 of The Question. This format doesn't work unless folks like you are willing to share some of your experience and then show up here in learning mode. Remember, you can get access to the raw data, the collaborative FigJam, and all of the recordings for this episode on my website, bencalahan.com. If you or your team could use an outside perspective on your design system program, I'd be honored to support you in that way.
There's so much more information about my individual and team coaching program over on the website. Thanks for being here and remember, stay in learning mode.
Transcript source: Provided by creator in RSS feed: download file