What's up, Colin? How is it going in your world?
Pretty good. Welcome back to Build Learn. I hear you were in my neck of the woods
I was like right down the street and I'm sorry for not hooking up. We had like just an insane amount of work to do. And yeah, we, our town in North Lake Tahoe They're like doing a multi year project to move how the municipal water line is hooked up from the back of the property to the front of the property. So they can like probably do easier maintenance from the road. You can just cut into the road, fix the water lines instead of having to go on to everybody's property and schedule when you're.
interrupting homeowners to go in and make changes and stuff. It was also weird too, because the meter was at the back of the property. So every once in a while someone from the PUD would like just go into your backyard and be walking around in your yard, which was weird. So now they'll have all the meters at the front. but that meant that either we paid, we would pay like a contractor, like between eight and 12, 000 to come and install it for us or install it ourselves and do the work.
And my dad is super familiar with all this stuff. He went and just downloaded all the permitting and inspections and requirements and documentation. And then, rented the equipment. And so I just showed up and was the work, like the manual labor for digging. so we dug a couple of trenches. They had the, I think the frost line in Tahoe is like 40 inches below grade. So they were super deep trenches. and we used an excavator.
So we got like a mini excavator that we went and dug like 20 plus foot, lines, down like four feet basically. And. For my house, it had to go underneath a sewer line. So that was gnarly, like just realizing how much you have to dig by hand when you're going around existing services or rocks or tree branches or things like that. so yeah, it was very much get up at six in the morning, dig all day, even if it was pouring rain.
And then we got to the point where, we got, my house and his house completely dug, ready to go for inspection. And then you have to wait 72 hours for the first inspection. Then they do the inspection and tell you like, okay, yeah, it looks fine. It's not going to freeze. And we approve this connection or whatever to the municipal thing. And then. You have to basically bury everything. And then you have another inspection that you have to call in. That takes another 72 hours to wait. So
Oh boy. Nice.
like dig a ditch, punch a hole underneath the foundation, pull up the waterline from underneath the house and hook it up to the new, like where the old main came in. And then. That also involves like removing an old deck and like taking apart a fence and changing all like a whole bunch of landscaping. And so it was quite a project. thankfully my brother came up and, my sister's boyfriend, Chris came up and just, they were, they were ready to work. So we got it all done in about seven days.
And then we had like our last inspection and the mains just after I left. a big lift, but we ended up saving probably like 20 grand. I think it was worth it.
Nice. You get to travel and see some family and roll up your sleeves and get away from the computer and
Yes. So I spent a little, like in the morning I went down to Java hut and sat at the beach for a bit just to enjoy a few minutes here and there. And one night we did go to Garwoods for dinner. Incidentally, we bumped into Eric Holder, former U S attorney general at Garwoods. It was like, Whoa, Hey,
Garwoods. That's, I haven't thought about that place. We have a trip coming up next weekend, like a guy's trip to Tahoe just to we usually plan a trip every year with some college friends and, and we try to go somewhere else, but in a pinch, we're like, let's just look at what's at Tahoe. we've been thinking about where to book a dinner, like a nicer dinner. And, we usually do like Lone Eagle Grill for near incline, but. I don't know.
I think our weights might be a good way to switch it up this year. It's a little bit of a drive from incline, but worth it.
Yeah, I think King's Beach also has Caliente, which is another restaurant by the same folks, and they have I don't know, they have a An interesting kind of like, theme, each of them has, these like really funny drink menus that are based on, like slurpee machine, alcoholic beverages. So Garwoods has goodies and then, Caliente, I forget what theirs is, but I know Garwoods is all rum based. And then Caliente, I think is all tequila based, super fun. But
bar of America as part of that. And then there's one in Southlake as well. I can't remember the name of that one.
yeah, I think there's actually I don't know. The last time I checked there were seven or eight and I think one just opened in Legends recently in Sparks.
The water bar.
Yeah, the other I guess like the other recommendations I'd make to on in North Lake and even in West Shore is Jake's in the lake, which is like killer. You sit out the marina. and then on West Shore, if you go to Sunnyside, that one is amazing. Amazing. yeah. So do you usually go, is it like a weekend thing or you go for just a couple of days?
Just the weekends. some of us are still here. Some of us are elsewhere. So it's a little bit easier to, Paul, I run to Tahoe and, we've got some aspirations for a little bit of a longer trip. Like we, like right now is literally Oktoberfest. I think it's a big lift to get us to all go to Germany, but maybe next year. We'll see.
so yeah, while I was gone, there was a whole bunch of drama that happened. Apparently, I don't know if I was like living under a rock and I bumped into someone and they were like, have you heard about what's going on with WordPress? I'm like, what WordPress drama? Like
Yeah. so I will try to explain this really quickly. It is definitely drama. It's hard to watch because, it's I think you and I are both very familiar with the Rails community and the Laravel community and WordPress has its own community and I very much owe a lot of my early career to WordPress and to Rails. I kind of bounce between the two. and And the community is super important.
And what, I think that the tricky thing with all of these is that they're like open source platforms with some commercialization of you have hosts, you have plugins, you have theming, you have all this stuff that people could spend money on. You have agencies who are building and, billing out time and materials for building someone a website on WordPress. You have WooCommerce, trying to go up against Shopify. Like it's just this whole ecosystem.
And. It took me a minute to fully understand because like there's a lot of complexity in how WordPress works. And we maybe can tease this out into another episode of you have the WordPress projects, org, you have the WordPress foundation, you have automatic, which is fully run by Matt. And they've raised VC and all sorts of other things. You have woo, which I think is owned by automatic and automatic has been buying up some other things as well. and like, where do all those lines cross?
Where, what are the boundaries? So this all started with Matt basically saying that. WP Engine misuses WordPress and they like disabled, like version history on posts just to save money because like they don't want to have all this versioning of all the posts on WP Engine and he basically wrote a blog post calling WP Engine a cancer to WordPress.
Which, we'll post some links, I'll post a link to the WordPress Drama Explained article from TechCrunch but they basically then did a cease and desist to Mullenweg and to automatic to withdraw their comments. And they basically were like, I guess in his article, they also called out, Matt calls out that the WP engine is misusing the WP brand and that WP engine is not distinguishable from WordPress. And that, they're not allowed.
A lot of legal online legal scholars have dug through and like it specifically calls out in the WordPress foundation that WP is not covered by the
Oof.
framework. and you have a million plugins that are called WP this WP that, you have plugins, you have themes, you have theme marketplaces that use the WP brand, like as instead of WordPress, And this one's an interesting one, because I've been a WP Engine customer for a really long time. We used it at Strava. We were very high paying customers at Strava, and they were one of the best. They still, I would still say they're the best WordPress hosts.
So a lot of people are like, we don't know which side to choose in this. but it only ramped up from there. So Matt basically took it. Said that he was going to go nuclear scorched earth, against them unless they agreed to pay, essentially like a license, for the trademark and a percentage of revenue.
fast forward, they basically said no, I guess some other drama has come out in that, in WooCommerce, when things go through Stripe, there's some sort of Stripe partner ID that Woo uses to actually make a percentage of revenue off of all the Stripes. Purchases that happen on WooCommerce. it turns out, WP Engine has their own partner ID and they swapped it out for theirs instead of Woo. that means Automattic isn't making any money on any of the purchases on WooCommerce through WP Engine.
Which no one really knew, I think, that was happening, even on the Automattic side. but then WordPress blocked, All access to plugins on wordpress. com to WP Engine customers. so it's gotten pretty silly. It makes Matt and automatic look really bad. In my opinion, I think some people think WP Engine is the bad guy here. But it's fully in their right to take this open source software and offer very specific bespoke hosting.
And I think in some of the language that Matt uses, it gets pretty brutal. Like he calls it like, I'll have to find the actual link, but he does call it like some sort of like bastardization fork of WordPress, which You're in your right to fork WordPress. WordPress itself existed and came off of a fork. the WordPress does follow GPL too, I believe. And it's not MIT licensed. So there are, limitations and there are things you need to do. it's you're supposed to contribute back.
You can't, I'm not super familiar with all the legalities of it, but it is not like a full car launch, like MIT license. So there are some rules around it, but when you're running like 40 percent of the internet, this is a bad look to your community
it's wild to see how many sites are running WordPress. my, my personal blog was on WordPress for many years and it's just such an easy, like CRM, if all you want as a blog or like a news situation, it's so easy to get up and running, but the, this whole story actually also reminds me of the rails world talk that I did. David just gave. And so we'll talk about that next time.
But like this whole, there's definitely like this trend of framework or open source, system maintainers that now want to try to make a viable path forward that doesn't necessarily depend on VC. And so how do you commercialize? Or have some sort of like open source thing that if people want, they can run it on their own. But also, you can make a business out of it. And it's really interesting watching how everyone is like taking different approaches to this from, 37 signals doing their thing.
Trying to build products and ship them and do their once, once, situation versus Laravel launching their own, like Laravel hosting versus
and raising money.
yeah. And raising tons of funding and you look at Vercel and what's going on with the next JS ecosystem and then just gobbling up a whole bunch of JavaScript, Like giant JavaScript maintainers and frameworks and becoming the host and provider for those packages. It's really interesting. and it's unfortunate that they're fighting and that there isn't like a, smooth way to go about this. It's interesting too, because.
The open source maintainer, I think probably feels like, Hey, I gave you a gift, I gave you this giant gift.
Why aren't you being more friendly or whatever, if you're a business that's making so much money off of this, I think I heard too, that one of the, one of the points of contention was that the investors in WP engine, like one of the investors, like never gave any money to the WordPress foundation or something like that, like there was no money flowing back into the open source, I don't know how I feel about that.
I don't think it should be required in any regard, but I also think, Hey, if your entire business is built on this open source thing, like maybe it would be a good idea to help support and sponsor the maintainer so that they can have a sustainable future. It is, yeah, it is a messy for sure.
and I think that was the criticisms was, Oh, is this mean the automatics not in a good financial spot? Matt's of course we are. It's not about that. And I do feel it's similar where you have DHH and Rails and Matt and WordPress, both of them. Rails is not, is not David. And WordPress is not Matt. Do Matt and DHH work on their respective projects? Absolutely.
But, they are not monoliths unto themselves, and I think, when we get into it in the Railsworld keynote, you can see, just because Dave doesn't like something, he's gonna go scorch earth and destroy it, and that's what Matt is doing here, too. The language is clearly didn't go by a PR team. Take your CEO off of, off the blog post publishing button. We get it. You made the publishing button. So it's a really weird thing.
It's just like someone woke up on the wrong side of the bed, had a little too much coffee. And I get it. there are times when people will get under my skin and I'm like, okay, this is the snarky version of the response back. Now let's write the less snarky version. and, But you're a representative of all of that community. And I think some people have been grappling with that in the DHH side. I look at Laravel and I'm like, I wish like I was part of the Laravel community.
Like it's pretty sweet over there. do they have issues? Sure. Like whatever. But, I think Silverlake is actually a private equity firm that I think now owns WP Engine. And so the, that criticism of, they like maybe. contribute 40 hours of time back to WordPress a week, like one, one person's time a week, which I think is, it's not nothing. but then the idea that you're going to block access to, so anyone who was a WP engine customer could not install W wordpress.
org plugins, which is like where you go get plugins from. someone then wrote a Cloudflare worker to proxy traffic around. And Matt starts attacking this person, this developer, like, why are you helping the private equity firms? And then the same guy goes back, like, why are you doing this at all? And then I think WP Engine went and just wrote their own cloud player proxy around it too. you're just, it's a shell game, you're moving the goal posts.
If it really does come down to this, Stripe thing, I have a feeling it has more to do with that. And they're just trying to, pound the table a little bit and make it sound scary, but WP Engine can go toe to toe, I think, with them, so it'll be interesting to watch. I also, I'm like, I could never go back to WordPress, personally, just because some of the things that they've done, it feels like it's time for a new WordPress. it does.
Unfortunately, you have 40 percent of the internet running on it, and so what does a new WordPress look like? we have things like Statimatic, Satamic? Yeah.
statin, yeah, statin, yeah. Yeah,
On Laravel, and there's been a lot of talk about how there isn't like an equivalent in Rails. and so that's been piquing my interest. I haven't used Statimix, so I want to pull it up. I know Justin Jackson's a fan of it and others. yeah, that's the drama and the update in the CMS world.
I guess one final thought is that it reminds me of all the drama that happened with Reddit too. It's in a similar vein where like Reddit shut off developer access. And I think that was them like setting up a moat and protecting themselves from, like LLMs coming and just like hammering them to pull content out.
But it definitely feels in a similar vein where There are certain companies are making moves that are, yeah, changing the landscape and how, yeah, how major pieces of the internet are working, which is interesting.
and I think the thing that we'll talk about more in the Rails world keynote is it's almost oh, anyone can make money with this, but don't make too much money because that's our job. And what 37 signals wants to make lots of money. But if you're out there trying to be a pass for rails, can't do that. Cause you should just be running. You should be running rails on a computer under your desk. And you have to do one man framework. Like everything has to be under your desk and no, I don't want that.
I want Heroku of 2012 is what I really want. I don't want Heroku of today, but same with WordPress. Oh. How dare they make more money or who knows if they make more money, but how dare they be successful, and use WP. If it was called WordPress engine,
Yeah.
go to town. But, yeah, it's, it starts to become a power struggle for sure. And, I'm going to follow along because I used to be a part of the community. So I'll report back.
Nice. Yeah. Okay. That's our tease for next episode. We're going to talk about Rails 8 and all of our takeaways from Rails world next time we'll save that for you. But,
I would love to go to rails world next year. We'll have to find out where the next one is.
but before then, what's, what's going on over at discord? What are you working on building?
Yeah. So I think the funny thing about all the things we just talked about is this episode is going to come out much time delayed, like after words. So maybe the WordPress thing will have been solved by then.
but yesterday, which was probably a few weeks ago yesterday, We now made it so we talked about on the show where you can build activities in discord so you can bring like Web based games web based experiences to discord and they run inside of discord So we yesterday made it available so that anyone can play Like publish. We took everything out of developer preview.
we made it so that anyone could take their activity public and make it discoverable and make it so other servers can install it and play it.
so if people are building games and they want to get eyeballs, there's a really good posts that I think I'm a lot like we can share in the show notes, where you can see the kinds of traffic that some of these games are getting on launch to the point where like we, if you really want to test your infrastructure, Running your game and be like, Oh, I have this like small delay in my queue, my queuing system. Like you're, it's going to be a big delay when you hit discord scale traffic.
there's a really good blog posts from battle tabs that we'll put in the show notes, when they launched their activity and the traffic that they saw. And we had to take them out of discovery so that they could. Slow down the traffic and recover. And, those we're just a game developer and we want to work for ourselves. And you're like, you're checking your cues in the middle of the night to make sure nothing's falling over. yeah.
that's awesome. I think that's, it's got to feel good to see so many things launched so successfully to, That scale of audience and like the audience is actually using it. so that's got to feel exciting. Very cool.
Yeah. And I think as game devs, they often will build a thing in the, building anything. You want people to see it, use it, play it. And it's if you have to go to my website and download my game or play my game on my, on itch or, watch it on Twitch, or whatever these things are. It's not quite the same as there's lots of people who want to play things, or have things to do. on my end, it was a lot of docs, a lot of okay, what does it mean when we push this button? what can people do?
What can't people do? There's, thankfully we're not shipping CDs, so anything that we got wrong, or is still in flux, we can go update all of that, but we, we made some new markdown components for this one, so I've been getting some, doing some MDX stuff, and yeah, just trying to figure out, I guess the biggest call out there is that, We have been trying to like iteratively upgrade our developer docs instead of doing one big redesign, rebuilds, launch this big thing.
It's Hey, all the headings have weird padding. Let's just change it in one PR. And like tomorrow the docs are gonna look slightly different. And then the next day we'll ship another thing rather than doing this really big, huge project that almost becomes like too big to fail. yeah.
Like when you launched, what would you say is like the main. incentive that devs had to get their game up.
So I think we heard a lot of people like we have everything under developer preview And I think a lot of people just didn't know like we with that you couldn't add like in app purchases You couldn't you didn't know where people are gonna find your game because we didn't have docs around What is discovery and distribution and monetization look like because if you're a game producer and you're like do we want to make this bet? You want to know what you can do.
And so if you're a partner with us and you're a studio and you've been able to talk to us, then you get that info.
But if you are an indie game developer or just someone who hasn't talked to us yet, it's a, another level of just yes, when you and then we have those case studies of people who have, built games with us during developer preview and, they've had to go up and down with us on breaking changes and all the things during developer preview and they were willing to do that with us, whereas with the open platform, we hope for a lot less breaking changes now that we're public with it and, and just being
able to say, you can bring your game to where we have Players and eyeballs and all of that.
It sounds like you've got killer discovery too. And that this, because you just launched it, it might be like a very awesome time to go get something up in the store just because like it's early days for that sort of app store. Almost if you can get something up and launch it right now, then, you probably have a short window before. It's flooded with flashlight apps or something. Actually. Yeah. is there like a review process?
do you have to look at all of the games that are being submitted to see does this meet discords,
We don't. So you can, you do have to do identity verification and then you have to fill out all of your description and your images and all that stuff. but once you get through all of that, it's technically discoverable. It doesn't mean it's going to be at the top. It means it's now able to be played in those places.
And if you wanted to do your own marketing campaign and get it to the, as it rises through the algorithm and stuff, and then we do collections similar to Apple does on their front of their app store. So we can highlight certain things that we think are like rising stars or, as we see them, things like that, but we don't do an actual review process, so we don't actually know all the things that. people might necessarily be building.
Whereas before this moment, we did know what all the things were because you had to come talk to us.
do you, are there games that if someone's listening right now and they're like, Oh yeah, I really want to make a game. and put it up just to see, is there a game that you have in mind that's like entry level, probably like simple game mechanics that you think would be a good one to start with?
Yeah, it's a little tricky because the platform is really geared towards multiplayer, but we have a lot of really good like solo plus type things. Like the one that I think a lot of people really love right now is it's called Farm Merge Valley and it's like Farmville. and so you mostly play by yourself, but then you visit other people's farms. So it's, there's some Animal Crossing type things there.
I would say any of the, party games that you love, if you can what I learned in this process is that game mechanics are not Like, trademarkable, copyrightable, enforceable, you just, the art is, the name, the words, all that stuff is. you can go out and, if you really love a certain board game mechanic, make it uniquely yours, but technically the mechanics itself are not something, you can copy those. party games, Would be a good one to look at if there's something you really like.
definitely having that, we give you the SDK to pull in okay, another person joined, what does it look like to play chess with two people, but then a third person joins? so anything that you can think of a game, what experience would you want for that? I think it doesn't have to be a game. what does pairing look as an activity in discord? I don't think we know yet. I think. I don't know, like it'd be really interesting to see, multiple cursors running around like tuple style, for editing code.
We have one called whiteboard that is similar to a fig jam type thing. So it doesn't have to be a game. It could be any shared canvas, that you can render. maybe something around music would be really like creating music together. We have a few music activities. Music's hard because you gotta get rights and all that stuff, but what does it look like to have, a sequencer and a bunch of beat machines and, drum machines and stuff like that?
In the like environment that you actually build the game in is like the browser or yeah. are you building it with something like unity or
yeah. you can.
Okay.
so anything that's web targetable, so if you can bring Unity 6 is coming out, next month and it's very web focused with Unity 6, whereas with Unity 5, you had to do a lot of hoops, make it smaller, make it run faster, be more performing, because you are running in an iframe inside of Discord, so you also, it's, you don't have as many resources as if you were running your own app, because you're running inside of Discord. And inside of an iframe, which is then being served up by your server.
so you're not, shuffling all of your, assets to us. We're not, compiling them into the app or anything like that. it is very much if you can build a web, game or web experience, you do have to then hook into all the right hooks to launch it inside of discord. but then it's cool because then you see CJ is playing, chess in the park. and then you can join right from that.
So that, I think that's the other one is that like you see your friends playing stuff and you literally just click and jump in, which is nice.
I don't know the right term for it, but in discord, there's like different organizations or whatever on the left hand side. Is that I, when I joined them, I call them a discord, like a discord. Like I'm joining the rails discord or whatever. Is that like the use the term used internally or no
so this has some funny history to it. They're called servers, publicly, to users, which comes with a lot of baggage because people think of servers as, a computer, or, right? In the API, they're called guilds. And so that is also a distinction where it's I'll often use the word guilds, but then publicly in the docs, technically we go back and forth between when to use server and when to use guilds. I think it was, the attempt was to get away from gaming only.
And now we actually have a lot more like guilds specific features and things too, like for gaming guilds too. So there's a lot of I think, D H H is video that we'll talk about next week. naming things is hard.
So when you publish a game, do people, does it like, is it tied to a guild or is there a sec separate section of discord where you like install the activities or how do you like find the activities?
Yeah, so before, like probably about six months ago when you built an app, which was like a bot inside of us, you could, it was only tied to a server. we have since launched user installable apps. So now you can have a to do list or a reminder app or a calendar app, or honestly, even more like MidJourney now is a user installable app. So you used to have to go to MidJourney server to use MidJourney. Now you can use MidJourney anywhere.
If you install it on your user account, you can take it with you to any server. You can also install it into your own server. So you can do either one. And activities, Can also be either one, but they're mostly user installable, so you can launch them from any server, so I could, unless the server says no activities are allowed here. there's a permission set there. so if I go play, Chef Showdown in your server, and you see me playing it, you can jump into it.
And then now that you know about it, you might go to the Rails Discord, and if they allow activities, you can launch it there as well. Now that you know about it. No, there's a little bit of a viral loop there.
Yeah, that's cool. I like that. Yeah. I think because you can monetize that makes it so attractive. And I think a lot of people will want to go out and build stuff. and because it's in an iframe, it seems super flexible. Like you could just build whatever you want on your own site, dump it in an iframe and then make sure you got the game mechanics, And the payment stuff, It's
yeah, I mean, the DevRel side of this is that we're trying to think like how do we make people successful like quickly and it's not the same as these like time to first hello world or time to sending an email in SendGrid because you like I have to draw the rest of the owl. You have to integrate and then you have to build a whole game. So and the more that I look at how even simple games are built, like it's just, it is a marvel.
I've been playing, Star Wars Unlimited, which is Star Wars flavor of Magic the Gathering. And I was starting to think through, what it would take to code that, and it's pretty intense. And then you also have to animate, and have cards, and graphics, and all this stuff, but some cards can do certain things on certain turns, and it's just like a whole, Wow, to build, and that's not to build like an FPS shooter or a top down adventure game.
game devs, it's amazing to me that they're always having such a hard time in the game dev industry. Because, it is impressive to pull off building a game. I don't care who you are.
Especially those like really high fidelity games that are animated and whatever, I remember thinking about like 2048 or, like Sudoku or these really simple ones, that don't have tons of graphics and sounds and like, all like very tricky, interactive, Multiplayer fight scenes or anything like, no sort of war, no yeah, no really complex game mechanic.
Sudoku would be a good one to build. I don't know if we have a Sudoku one yet.
those are the games that I actually love the most. The ones that are just like very simple game mechanic that lets you shut off that part of your brain. And the reason I like them is because I can listen to a book, like an audio book or a podcast or something while I'm playing it. and yeah, I don't know. it might be interesting to, to do, but trying to figure out how to make it multiplayer
what I was gonna say, yeah. You could just do leaderboards, but I don't know if it's, Sudoku's not really that kind of game. but yeah, like that's a great example. What does a, the game mechanics of Sudoku plus multiplayer look like? But also, how do you also make it fun for one person, still? So is it Sudoku? Is it Sudoku Clash? you and another person are trying to beat, fill it out as fast as possible. Problem is, it's like a thing you come back to. it's like a crossword puzzle.
So, do you do it all in one sitting? Do you save state? All that stuff.
another one that comes to mind is like type racer. I really want my kids to be able to type fast. And so I've tried so hard to get them to type. That was like the first multiplayer game I think I ever built. It was like an experiment in WebSockets, like back in when WebSockets were like a brand new thing, I was like, Oh, it would be so cool to build a multiplayer type racer game. And now they just use, there's a million apps and websites for them to learn that.
But that might be another cool one that's embedded. and then you can get you could probably do in app purchases for customize my vehicle or, like I want, custom skins or give me a turbo boost button where I can like at the end of the game, just hold control delete or something that like just leads me to the end or whatever, like
Oh, man. Yeah. this reminds me of like math blasters to some math games.
totally. Yeah, that's another one. We just started last. one of my sons is doing, they're doing these quizzes in class now, where they have to do multiplication, like how fast can you do this giant sheet of multiplication things? And so we're, we were like looking on iPhone for just some simple, like little apps that let you practice multiplication and, There are so many that have in app purchases. I was like, what is this? This is just a flashcard like thing. Like, I don't know.
Yeah, that, that might be another fun one. Like race multiplication or something.
everyone's trying to, it's back to the making money. Everyone's trying to be sustainable with their thing.
Yes. Yes, totally. cool. congrats on the launch. That's pretty sweet. Very cool.
What is, what are you working on over there?
we want to nurture customers who are, basically not responding when they reach out and, or they've fallen into, I don't know, some sort of state where we just wanna kinda reach out to them without very much effort. 'cause there's a lot of customers that will reach out, get an estimate, and then just ghost us. But we wanna make sure that we're following up and such. And a lot of that can be automated. There's tons and tons of platforms that do this.
we've looked at, things like braise and segment. Lindsay actually works at a company that does this to enter iterable. And all of these are like great for customer messaging, but, we even set up a Zapier and we found that at the end of the day, we want to be able to.
Build the sequence of messaging, but then also take actions on our own database or be able to automate different things like, okay, now create a task for the sales executive or transition the project from this status to this status, or, okay, now it's time to send this email to request colors or place this order or generate this review or whatever. So we wanted a little bit deeper integration than what we were finding from these adaptations. Out of the box solution.
So now we're starting to plan out basically our own Zapier, like a workflow automation engine where we'll build our own like library of, actions that you can take. And then a suite of triggers, just a small, like group of triggers that can happen when customers take certain actions or we take certain actions.
And then, yeah, just it's been fun kind of thinking through what does that look like and how do you build it in a way that's not too generic, that's like confusing, but also not, too constrained so that we can build out whatever workflows we might, encounter. So that's been fun to start thinking about, but, yeah, I don't know. Have you used any workflow automation things that you're super into?
so I've used iterable that we use that at Strava. What else I've used Zapier, obviously. I've also tried to build my own. That was cloud snap. I think I understand the issue you're running into, That you want like Zapier can't reach into your database. I guess you. You could make an endpoint for it. I think the same thing happens. I've been listening to bootstrap web and, they've Jordan's been working on Rosie and it's answering.
The phone for you, AI as like service providers, mostly, painters, plumbers, whatever. Right. Um, but they constantly, it's Oh, I just want you to book this appointment. they're like, I want you to check on the status of our order. It's we don't. People want that reach into our database and either move a thing, check on the status of a thing, send a thing, create tasks, like what you're saying. and I don't know if I've seen anyone really do that.
all of these do this, like even the open AI and all these different companies are like rabbit and all these, it's Oh, watch this cool demo of this one use case. And you're like, what about this other use case? Oh no, we're not going to show you that. So it's as soon as we get into the real world, a lot of these, Don't pan out. but I think that was also a downfall of CloudSnap. we thought it was too simple and didn't do enough.
And then, we went towards the developer approach of, what does it look like to give, like for this tool that you guys are thinking about, Is it for, Developers to build out these workflows or is it, do you want an end user to build out these workflows?
The idea would be an end user can build them out. Like initially I think we'll probably build them just cause the UI will be, clunky I expect because it'll be so like customized, but I think ultimately the idea is to make it like Zapier or make. com where you can drag and drop nodes and then those can, flow into one or many other nodes and then you can have, conditional logic and repeating and delays and. but yeah, there's those two layers, right?
as a developer who wants to add a new action or like a new activity that you can do within the framework, we'll have to figure out how we want to expose that and then as the end user, what is your experience for Defining what data is flowing from the previous node into your node. And then how do you decide whether or not you're actually going to take action? it's tricky.
Yeah. That sounds like a whole company right there, CJ.
Yeah, we keep rebuilding giant pieces of other companies. but yeah, hopefully the longterm is We might have a person who's full job, full time jobs, sit down and go through every single customer journey and workflow and optimize it and AB test it. And like, is this messaging performing well or should we change it? So it's more clear or
Yeah. It's your last mile problem, right? Every, all the work has been done. They got this quote, they did all the things to get the quote and then they just disappear. And so if one person can be their whole job to like, how do we close more? it's like a weird. It's I'm sure there's a word for it, but it's like a follow up salesperson.
yeah, it's like an, almost like an automated, yeah. Like, sales development representative
success management, engagement specialists, yeah.
But even like after the project is done, like following up to ask for a review and or, like trying to sell again later down the road.
If that person can then also go in and optimize the path, they're like, okay, we learned that this doesn't work. Versus, in companies I've had to be in, I've had to go interview all the people doing the work to figure out what they're doing, and I'm like, why do you do this every time by hand? you're, this person is doing, It's working over here, but they're also sending a handwritten text and email, and this other person's not.
And this person's getting ghosted more often, this person's closing more things. Does that mean, was it the customers in particular, was it because they were doing that hand, Which also means, if you automate it, is it going to have the same effect as if it came from them? Can you make it look like it came from them? Is it obvious that it's generated? All of that stuff.
right. Yeah, it's interesting. and I think we're gonna we're getting pretty close to just letting AI do some of this. But then yeah, how do you expose them? As tools or whatever, like what's the way that you actually give this to people? I don't know. We'll see.
Yeah. I think Jordan was coming up against this, do you tell them up front that I am AI? Because it's going to get to this point where it's going to be so good that I've been on calls where I'm like, I'm not sure if this is a human or not.
Yeah
so do you be up front with it? Does it change the outcome? Does it matter? Do you want to talk to a human? sometimes I don't want to talk to a human, so I'm happy that it's a bot. As long as it's, responsive and like you can interrupt it and move to the next thing and stuff like that.
Have you ever have you heard of replicate this site replicate. com?
Pull it up. I have not.
Okay, so the first time I heard about this, was a while ago and I played around with it a little bit and it didn't work how I expected, but I heard about it again when I was listening to the Peter Levels interview on Lex Friedman. And this is like the thing that, like the tool that he uses behind photo AI. And so they let you host and run models. And then they also have a bunch that you can run.
And so something I was trying to figure out is if we spun up our own models for things, can we run them somewhere more affordably than if we just keep using the open AI and anthropic APIs? but I also based on something that we talked about before, it was like, Hey, let's figure out how to like actually train stuff. so I'm going to send you this photo in Slack. so this is so funny. so I trained a model. On replicate with my face. And I was like, show a picture of CJ sitting next to Lake Tahoe.
so maybe we'll put this in the notes, but, it was, surprisingly straightforward. They actually have a web based thing where you can upload a zip file too. but I thought it was going to be like way harder to do, but you basically just get a bunch of images, make sure they're roughly the right size. And then. you have to come up with a, some sort of special token or word that represents the thing that you're trying to teach the AI.
I think part of the issue that I ran, like there's gotta be like a million optimizations. I basically just took 10 photos of myself and dumped them in there and said, this is a picture of CJ doing this is a picture of CJ doing that or whatever, and then this is what it spit out when I said, give me a new picture of CJ,
did you train it on what sitting next to Lake Tahoe looks like
no, it knew that.
Okay. Because it's clearly, I think it's been trained on lots of women sitting on the edge of Lake Tahoe. With, I had to zoom in, but with an engagement ring, potentially?
maybe,
that's just like a sunspot, but like the, like you see your ring finger?
definitely didn't properly. It, I think my ring finger has four knuckles maybe,
Yeah, and we know AI is not as good at hands, and it did hands, you have five fingers, so that's good.
it definitely didn't train on how hairy I am. That's for sure. This person is like, is a hairless younger version of
Yeah, you have no pores in this photo.
no. Yeah, it's so good. It's so funny.
Amazing. this out.
yeah, play around with it. they've got a couple of different models that you can throw images at. And then, I also took a bunch of our like craftwork professional photography that we did and all of the ones that were like, these are all rooms that are pink. And then I took my pink, I got a picture of my room that was gray. And I was like, Paint the walls pink and funny, like it just only did the trim. It only got like the window frames pink. So there's definitely a lot to be learned.
I think there's like stuff around masking and like trying to figure out, if you upload a picture of a face, like actually just make everything that's not the face black, remove the background and do all this pre processing. So lots more to learn, but, a little bit of progress towards learning how to train our own models, which is like super fun.
yeah. I haven't gotten into this yet. We'll tease this. We have, thankfully we have so many future episodes to talk about, but I've been working with the Gmail API to pull down my email, and the more I look at it, I don't necessarily think I need, AI models. I just need, random tree forests, math stuff, right? And yes, I could throw all of it. at an AI.
But what it feels like is that like AI is almost I think people are really obsessed with it because you don't have to do as much to get the answers that you want. Whereas with these ML models that we've been using for tens of years now, you have to do a lot of math and a lot of work to get the answers that you want and a lot of training and tagging and data labeling and all this stuff. But it sounds like you, To get really good answers, you need to do that too in the AI models.
And then they're just really expensive to run. Whereas once I tune around emails, like there's a set, there's a lot of parameters in an email, but once I get the full set, there's not really like new. special things in email that we're going to be surprised by. so I'm curious to see how much I can get away with without having to use a model, but I want to look at replicate a little bit to see what else I could do around, like, you know, what was the Japanese restaurant I went to last year?
Things like that. Like how much of that Do I, can I use, do I need to train every single email on that? There's going to be levels of some people are not going to want their emails trained. And so it'd be, it'll have to be a local model. I don't want to send someone's emails or mine to open AI. That's not going to happen. yeah. So we'll have to chat more AI stuff in the future, but,
Totally.
yeah,
Yeah. Fun stuff. Fun stuff. Let's leave it there. Thanks for listening. yeah, if you want to head over to buildandlearn. dev, that's where you can find links to all the resources and stuff we talked about in the show notes. And, yeah, I guess we'll see you next time.
All right. Bye friends.