Search Is In Everything You Build (And You Probably Don't Know It)

Hi everyone, My name is Patrick Akeel and joining me today is Prateem Bhosili. She's senior developer, experienced engineer over at Treble. And I love this conversation because we get into the magic of search. Search is actually in more places than you would think initially. And next to that, we discuss how AI has impacted content generation, all the content we see on LinkedIn and on Twitter nowadays or X. And lastly, we discuss the pros and cons of being in developer experience.

So enjoy. Search is always shaped, at least in my opinion, the user experience. I go to an e-commerce website and we didn't have vector search, similarity search. I search on text, I'm trying to hit a title match basically. And then I can filter for keywords. Those are like attributes of a certain product, but I can't say give me a product and then just talk and then the product pops up.

It's always like artifacts, or that's how it has been and now with new technology we can evolve search as well. Yeah, I, I actually feel like I think a lot about search, which is a bit weird. Like sometimes I even dream about it or like just because I feel search is is such a deep topic or a problem that we're now realising can be solved in more depth. And let's take like from extremely early days, we would search for CDs in our CD box.

We would search for files. And then we had file systems come up. And then we started building queries on file systems. So there was no concept of the tool understanding what you mean. It was always giving what you literally ask for or like literally we, we search based on I DS, right? Like we have I DS with which we connect, we do joins.

And then we had this whole layer of like web search comments, search engines, where we started doing the same thing but for Internet. We started putting the exact things that we want and then we would get results around those things. I don't know if you remember when I was in school and I would not get some things that I'm looking like I'm searching online. My seniors and my friends would tell me you should add this plus between 2:00 between two keywords.

Like if you're looking for AAI think I shouldn't say coffee shop specifically when I'm, but like if I'm looking for a sandwich shop, then you should say sandwich shop plus like the area that you're in, like for example, Mumbai or like Amsterdam. And then it will show you more results because that's how the search engine actually connected keywords. And now we are at a point where we can literally put in things like the best sandwich near my house. You don't even have to put your

house address anymore. Like it's so much into, Yeah, we've come far, far, far long ahead from where we were in terms of search, I feel. Yeah, I've never thought about

that, that before the Internet, the way we used to search it was like alphabetical sorting. And then probably in there you have a certain another sorting to get to certain documents that you would store and you would store them meticulously to be able to find them. And then also later on, and now you can ask and the content kind of, yeah, contextualizes and knows based on your ask what the right or at least predicts what the answer should be. That has never been there

before. Yeah. And even before, like everybody's still had their own system for search and strategy. Like some people love to keep things chronologically. Some people like to keep things colour coded. And like, if you talk to like, I remember my uncle, my husband's uncle, actually, he has his books, like, he has a big collection of Tintin, Tintin's books. And he has his own way of

keeping them. And that's not in terms of the year when they were published, but it's based on Tintin's like, journey from one country to another. Like, wow. And it's, yeah, that's his way. He's like, no, no, no, you, you can't look for the one where he, when he was in France. You have to look when he was on the ship trip. And it's like, OK, so everyone has their search strategies, no matter if they know the technicalities or no, I think. Yeah, I feel like search is in a

lot more places. But if I I put my software engineering hat on the intricacies of search and how it worked, I never really dove into that unless I had to implement it specifically. And then it would either be on a

database level or it will be software as a service and then I would really dive into it. Do you think or would you advise people to dive into search just because it's becoming kind of more ubiquitous also with AI technology nowadays? Definitely, I was myself not interested in search like at

all. Like my experience with search was if I'm searching for something in a file or like on APDF and if that tool had a good search system, I would get the answer and I was happy if I didn't get something. I was just frustrated like that was it with me and search like I'm searching for things I either find that I don't, but it was really when I was working with Saria Libi and I got I got a long time that I spent with the search engineer Emmanuel.

His name is Emmanuel. And he was really, he really opened the doors for me and I was working very closely with him. And that's when I got to see how search looks on a database query level to what is the output of it. Like how does that actually look like when someone's searching for a word or even for a phrase and they have a big data set or a database of a collection of things? And when you're adding a layer of search on it, how does it technically work?

And it for me, it was a lot of people think AI is magical, but for me, search was magical because I was like, OK, like this is how like you actually get like search and this is how you actually find that one word that you've been looking in your whole PDF. And I'm actually sad that I didn't get into search earlier. But yeah, I did.

I would definitely ask anyone and everyone to just read more about it maybe and understand what happens when LLMS are searching or like when we're using Search Insider I DS now like with all the I DS like, how is that tool giving you what you want? Yeah, I. Mean for me it's, it's super fascinating that search, I always looked at it from a buying perspective, e-commerce perspective, attributes to a certain product. And that's what I'm trying to manipulate.

That's what I'm trying to find. And then with websites, it's kind of the same. You have a document, it's a page somewhere, it has keywords, and then those keywords you're trying to hit and you're trying to match on that. And now we have something completely different, right? I can just speak in my language and then it's still very much search because it just tries to contextualize based on language. It tries to similarity match. It goes through a whole

prediction tree. I mean, LLMS are still kind of prediction algorithms and that's basically it. But it is in essence, I feel like it's still a search problem. Same when I look on Google Maps and I'm trying to do something that's nearby, like that's geospatial searches, just search on a different axis. It is kind of almost everywhere. Yeah, yeah. Which? Is pretty cool.

Yeah, I agree. Yeah. Again, I don't want to get into that depth of like, I just think also like when we when you spoke about geospatial search and I think it's the combinations that we are getting good at now like hybrid search, the concept of like combining two different types of searches is what's really unlocking a lot of things

for us now initially with and I think Amazon was a big pioneer in terms of like this whole recommendation systems when we would buy something and then it would recommend a similar product. And I feel that also really pushed the boundaries for semantic search because it could make the connection between what would be the next, what would be something that a person would like to buy. If they've bought basically a packet of popcorn, then they would probably also want some ice cream.

Or if they actually bought movies, then they would probably also want popcorn. There's this very famous research, I'm not sure by which e-commerce company, but about like what people buy in supermarkets. And there was this research that around bottles of beers or popcorn, there would be diapers because there was always like if dads are having like a, a, a movie night by themselves and like the mum is the mum is like out for a, a night for herself. And then the kid is with the dad.

Then he would want to pick up a a beer and and diapers for the kid because I was like, oh, interesting. Yeah. So this this concept of hybrid search is what is really got us to another level. We've had full text search, which has helped us search through textual things and now we combine that with any other search that we want like geospatial plus full text, semantic plus full text. So yeah, I think there will be a lot more with how we improve on image.

I feel understanding images, converting images into vectors, so and that's going to also be big, I feel for sure. Interesting. I I thought of hybrid search, but I I can't think of a good way to have a hybrid search where there's semantic search and then keyword search. Like I don't have a good use case for those yet. If I have semantic search, I can just speak a natural language and then there's a similarity match and there's a certain

level of accuracy. Maybe having keyword search or full text search next to it UPS that accuracy, but I don't know when I would use that. Yeah, you're right. Like semantic search will still is still works on some seven level of prediction, right? Like you have you, you'll convert them into vectors and then you'll match the vectors to the closest vector from your data set. And then there's keyword search, which is literally matching your keyword and searching for that keyword.

So if you combine them, then you make sure that you're not missing out on any, any possible match from your data set. And that's even if it's not, so it helps with accuracy when it's built on top of each other, but even when they're, when they're used together and then an average of these two searches are made. The possibility is that you don't lose out on similar matches or potential matches.

So based on like what use case you're going for, like if if the use case is like you're looking for a house, then it's not about one exact house. It's about how many houses fit into your search category and you want to maximize on that. That's when you would want to have more hybrid search power. Whereas where there is semantic as well to say that I want a good vibe. Now, what is good vibe like?

How is someone going to say, but if the vector data has more similar objectives or similar, similar things defining houses being like, I feel positive in this house and this had good light. And then the the meting model could connect these two things and that could come into your search category. Then like in your search results, whereas no full text search is going to give you a a

good wipe house. So depends on what use case you're using for. That's how I look at it. Got you. Yeah, you and I spoke before the show specifically about the role of developer advocacy and you share something which to me was like a really interesting insight that developer advocates have this kind of ebb and flow with regards to the tools that they develop.

But really when you're internal somewhere, when you're a product engineer, you're part of a team and you you're there for years, you can really go deep with regards to technology. How do you combine this kind of going deep with regards to your current role now? This is actually something that a lot of developer advocates struggle with. I have struggled with that as well. You have to figure that out with the team you're working with. First of all, it needs to be something that you want.

I feel we are at a point where it is OK if you are specifically not chasing depth. Like if you want to have more broader and more generic information for more things and you want to be in this role where you can quickly pick up different things, test them out, try them out, and if once they feel more logical for your team to pick up, you can pass them on

to your team. This, this would be more towards a community side of role where you just want to, you want to know enough that you can pass things to the right team. But then for people who still want depth for them, it's something that they need to carefully sit down with your team or look for roles where they offer you depth. Like they're OK with you spending some time with engineering teams, they're OK with you taking complete

engineering projects. And then it's a good divide between some developer advocacy part, which involves like going out there explaining how the product works, like making sure that developers know how to use their product, but then going back working with the engineering team, working with the product team, spending months and just like building an SDK or whatever is important for your tools. So it's, it's a mix of it depends.

And like it, it's a mix of like what your company can offer you and what the product can actually benefit from your experience as well. That's. When you're creating projects or when you're creating software,

what about it gives you joy or gives you energy? It is the output for sure. Like this was one of the reasons why I could never connect with front end. OK, so I have always been a back end developer. I have had my time, but like I, I of course build whenever any project needs AUI. We are at a point where we cannot shy away from building AUI. We have all the tools right there and it's it's not rocket science. Like we literally have templates

out there. But I've never really been able to connect with front end because I feel there is still more steps to success when you're building a beautiful website or a product, product front face. Whereas with back end, it's AP is when we're talking about AP is back end databases. The result, the output, the, the success is so fast. Like you hit a query, you know if it's the right or wrong you, you hit, you hit a command, you know if you're doing the right thing or not.

You run a benchmark. The benchmarks will tell you if you're going in the right direction or not. So for me, it is fastest output that first connects me to building more, which is also something that my partner keeps telling me that I need to work on. Like you're not always going to get that quick dopamine hit. Sometimes you're going to have to push, push and then you're

going to get it later. And I think that's why I've been in the developer advocacy role, because for me here, it's I build a lot of projects, so I get more of this dopamine hit. Dopamine hit, Yeah. Yeah. And I think it makes sense for me. Like, I don't know why it is, but when I would solve a problem that in my opinion, I thought was complex or it was something

that I had never done before. Yet even though it wouldn't immediately have business value, even the just the way I wrote it or the logic that I put into place, it would be something that I was happy about. That would be something I would be proud of. And I feel like with a lot of tooling nowadays that that goal post of like output, it's moving because I'm not necessarily writing that creative part is moving away more and more to tooling to make me more

productive. But then this kind of enjoyment, it changes for me. I need to achieve more to feel kind of that same amount of joy. Yeah, or you just miss out on the joys that you could have had to like the non assisted way. Yeah, we are. Yeah. That's, I think this is also something we spoke about, right. Like previously when we were chatting about the joy of doing things has changed and like what we get joy from has now changed.

I don't really think it's good or bad. It's just that the goal posters change now and I feel IDs like AI DS are still not there yet. If you have to support really complex solutions like C++ or if you're working with hardware or Rust for example, it's still is not going to give you the most

elegant solution. It might point you to some solutions, but it's really, it really performs well when you're using the most highly used languages like JavaScript or PHP or majorly JavaScript. Yeah, I think most of the success that I've seen with people building things has been with JavaScript. So if you still want to get that joy from writing code, I think maybe more languages like Rust or core languages like CC Plus Plus might still give you that

joy I feel. Yeah, yeah, I would have to move to those languages. Like I feel like if I can draw an analogy between like creating software and like being a painter, I would. I think as a painter I would enjoy painting and like the different type of brush strokes. Like that's what I would get my enjoyment out of. And now if I have an AI enabled painter, they can just click and drag and like put colors or they can completely make a new

painting in the form of minutes. Like that's what it feels like now. And then the output is like a beautiful painting or not. And it's still very subjective. Can I look at it differently because with this analogy, I could never paint anything good and now AI helps me paint good things. Like even if it's a one click thing, I do get to say that I build that like I I prompted that and that gives like. So it again depends from where

you're coming from. Like what for artists, definitely it is going to feel like it's. It's taking over a lot of their creative overhead. But for example, like my sister-in-law, she's an artist. And the last time we met her, which was six months ago, she absolutely hated Chat GPTI. She absolutely hated the fact that, like they, they're coming up with posters which like businesses are using. And no, like people won't, don't want to like, rather give her a contract to make that poster.

Yeah. And we spoke to her last week and she's thinking of starting her own business now. And she was, she was making business cards. And then she sent me and my husband some suggestions of like what she's doing. And we were shocked that she made it with Chad. And you're like, hey, she's like, no, you're like, yeah, it's really. I was busy. I needed some quick ideas to get out. And this was really nice.

I was surprised. And now I can, I trade over that like, of course I put that last stroke from my side and I'll make sure that it has my touch. But yeah, it's, it really caught my vibe. And I'm like, I think this is what this is a good way to think about using AI assisted tools to

to, to use them when you're when you have to get things done and then go back to things doing them manually when you really want to get joy out of them maybe. Yeah, I think so too. Like it is this shift then more towards OK, what are you trying to achieve and then the way to do it that can be accelerated by tooling by whether it's Shachi, BT or if I'm a software engineer, but generating code.

Basically in the end, we are going to focus more on on outcomes and hopefully in a good way because we're doing a lot of stuff now. I see start-ups kind of spring up like mushrooms and they're all trying to find the gold with regards to AI and what they can do. That is, I think, I don't know if I call it a hype cycle, but it's definitely a trend that I've seen. Again, there are winners, there are losers, some in like in any

kind of cycle, right? And I'm actually seeing a huge growth and like a a pump in security products and tools that just make sure like whatever, whatever you're like white coating or building with AI, it's secure. And yeah, they know nobody from like nobody's complaining. Like we want everyone to build more things. We want more authority. We want more businesses. Like we, we don't really want more monopolies, right? Like, so just giving the the power to people to build is, I

think a net good thing. Of course there will be people who will always find bad ways to do something. And I don't think this is new. Like this has been with like what every technology that has we've built with like, yeah, it's the way we look at like, do you we do we want to go ahead with it and do a net good for the community or do we want to look at like, oh, just the bad things? Yeah. It's funny that you say that because the, the possibilities that we have now, like it, it

changes behaviour. And with that behaviour change, generating code, multiple lines of codes, like it goes beyond the grasp of kind of what a human can input, unless that becomes your day job. But still the security aspect of it and that vibe or that trend that you've seen, I think it makes sense because people are trying to maintain and control or at least accommodate for this change in behaviour.

Yeah. This is something that we actually are seeing a lot of treble like the company that I work at, we, we work with a lot of APIs and we, our major goal is to make sure like people are building secure APIs because that's how we are communicating with most of our micro services and different logics. And well, the amount of APIs that people are writing has increased immensely.

But with that, the amount of insecure APIs that people are building, the amount of secrets that are just going out, like, that's crazy. Like, and there's one weird start that I had actually presented last month that I'm going to present this month as well in Berlin is when we actually looked at some numbers, like based on languages and like what, what frameworks or what, what developers are building more secure, writing more secure code.

Unfortunately, it was JavaScript that came out to be really, really insecure or just writing bad code. And I wonder that's why is that? Like, is it because of the ample amount of bad code that's just available to train on or or just because it's in it's language for everyone to pick up?

For me, I think it's the, it's the latter. Like I, I had a person on, he was from the Rust community and he said it's interesting. Rust is a language. It has this kind of big hurdle to get up and running to feel productive. It's very different from something that doesn't have many language specific things like Go for example, or JavaScript gives you this freedom and you can do many things in different ways. Like it has an up and down side.

But he said if I'm doing a job interview and someone has never done Rust before and I can see they actually wrote high quality Rust code, that makes them a good engineer by virtue of them being able to do it in Rust. For him, it already UPS the bar with what he sees in quality in a person, and he doesn't have that same feeling with the rest of JavaScript. Yeah. Well, I do write a lot of JavaScript code and I think it's it is the language of web. I will definitely say that for sure.

But I think overall there is a lot to catch up when it comes to building secure applications in in JavaScript for sure. I think again, for me, I feel the reason why I do not agree that trust is has this history of like building safe applications or like there's actually again an interesting talk that I was attending from a senior developer at Tesla and his whole job was one of his jobs.

Actually, I don't want to quote him on something wrong, but was to figure out how many security concerns or just malware is available in Rust crates. And he mentioned that sometimes even the official crates have a lot of like buggy like security concerns and this will just go past right Like nobody is going to check for them. And it's just, I think it's because Rust is a niche. That's why it has this personality of like being the smarter language.

I don't know. I personally feel like, yeah, it is great. There are use cases, but to each of it's own, like what I wouldn't like if I want to build a web app and I wanted like to do the like, I wouldn't build it in trust, I would build it in JavaScript. So yeah, depends on what use case you're going for while you're building something, yeah. I feel like Rust is, and I, I talked to him about it as well in that episode specifically.

I feel like it's a very new language compared to the languages that we have out there. And any language has to have years under their belt before they become mainstream. Because there's a lot of languages that I can't name them that I don't know about that are probably new. But the ones that are established, I feel like Rust is kind of one of the later ones that came out that is now established in the community. I actually don't know how old Rust is.

I know, I know. Go is 25 years old, maybe around 25. 25. Yeah, yeah, yeah. I think it was 2000s when like Rob Bike and the whole engineering team from Google built it. I don't know, I might be wrong, but I'm pretty sure it was 2000s, yeah. Gotcha. Yeah, I thought really it was like from 20 tens, but I'll, I'll look. It up? I don't know, could be. Yeah.

But yeah, when I'm thinking about, OK, how, how does a language get adoption and how does it get created, especially now with Productivity Tools, I feel like Rust might be one of the last languages to be like a higher level language specifically for humans. If we're going to get a new language, it might be something specifically for LLMS. I do, yeah. This is something that I agree. I also feel like it is always like, for example, I've been tinkering a lot. I'm not good at Python.

Like I didn't really spend a lot of time tinkering with Python or

like, but I've been now trying to understand the whole machine learning ecosystem through Rust. So if I'm building any, any CLI based tool or like an AI tool, I prefer, I actually prefer building it with Rust. And the there are enough tools to, to, to have the same result as Python in Rust now. And that's just because how well, how quickly you can pour things in Python, and I mean in Rust, sorry.

And yeah, I do agree. Like I'm I've been really enjoying using Rust for machine learning specifically. It's been a while since I've picked up a new language. Like I I haven't tried to pick up a new language with the tooling that is out there nowadays. I've always been fan of Go. I like TypeScript as well, but like Go is like my bread and butter.

That's it. I think it would be fun though, with the tooling that is now out there as kind of an accelerator to a learning process to pick up a new language and kind of feel

out the intricacies. Yeah, I how I picked up Rust and Go was again, like Rust was one of the major languages that my previous company used because the database was built in Rust, like it was first written in Go and then it was just built in Rust. But it's always fun to like have the same benchmarks in both the languages And that gives me like, that's that again, is something that gives me like a lot of joy and like it's fun to see and it's sort of like, it

hurts your ego if like your, your language isn't really doing well and comes to benchmarks. And Rust and Go is always something that go hand in hand, like during the compile time, Rust will definitely take more time, but when you have an executable, it's super fast. Like it's, it has literally no like start up time. And that's something that you'll you'll feel. Whereas with Go, compile times

are more faster. But then there are times when you will feel that before, like you'll feel the breaks before, like your application starting in. Yeah. So just running the same, the same tools with two different languages and checking benchmarks and like how much time both of them take to run, I think that really clicks. At least that clicks for me, yeah. Do you remember any interesting insights from those benchmark comparisons?

Yeah, it was Primogen actually, who did a whole stream about, I'm not sure what he was. He was running on two different systems like I think they that it was, I think it was ACLI tool itself. Like 1 was built with Go and one was built with thrust and Go actually performed well on that one. And the reason that he he figured out was because of like Go is language for CLI. Like I personally feel like the fastest way to build a quick CLI tool is just pick up Go. Do you think?

I mean, I, I've built a lot of CLI things in Co and I've seen companies also do that. Yeah, it's. Pure joy, right. Like, and I think the tooling in Go is really good for CL is I wouldn't again, like try to build a web app with core, like try that. But for CL is it's it's pure joy. Yeah. Actually, I don't, I don't really remember what it was, but the conclusion was that Go performed well compared to Rust. And I was like, OK, yeah, yeah, I'm on the right direction. Like, yeah.

Yeah, I like that. I mean, I've done a lot with Go, but I I want to switch more so to the content side because I, I am a lot on LinkedIn, not so much on Twitter.

I know you're more on Twitter and people have started complaining specifically on LinkedIn that there's a lot of AI generated content, AI generated comments. What is your thoughts on kind of AI generated content that nowadays? What have you seen? There are a few super generic things with the AI generated content that are like red flag, like unwielding in the age of and all of that. And that's just unfortunate.

Like if I was reading the same language maybe like 10 years ago in a novel, I would have loved that. I would have, I would have felt like I am connected to that thing. But now because it's so overused, yes, it does definitely put me off a content piece. Like I will not read A blog if it starts within the age of something something unless it is written by someone like who is well known.

Like for example, if like like Paul Graham writes like PG writes an article and it starts with like in the age of and I'm like, OK, let me give it a chance. Like like I know this person has a history of writing good things. I'll give it a chance. But if it's someone that I don't know, or even if someone like I, if yeah, it's, it's a serious red flag, at least for me. Yeah, I think it's hilarious how people really enjoyed it. And then content creation got easier, so then people started

doing that. Like writing is the simplest thing because also a model loves token output. It will just, it would just keep going. Do you want me to add on top of this? Do you want me to give more options? And if you just, if the aim is to write the longest thing ever, we are now really good at that. And then people try and do that. And then that's what you see. And then people have this kind of, yeah, they don't like it from a reading perspective.

We don't really write from a reading perspective anymore. We write to be writers. And everyone can be a writer nowadays. I don't like, I feel like we still like to be engaged in something and if it's really well generated by AI, where some like, I don't know if it is generated really well, but I cannot catch it, I will read through it and like I will, I will enjoy it unless like someone puts it in my brain that, Oh, you know what, like this was generated by AI. And then I was like, is my

system OK? Like, how did I enjoy this? Like, and then I'll try to assess like what was exactly that, which is again fine. My end goal is to enjoy a piece of article. Like I am OK, like reading AI generated like, because we read like things that chat GPD shares as well. But that's like something that we need, We need that piece of information which we don't have. And that's when we're reading it.

But if like if it, if it's concepts that are already out there, then I would probably go back for a 5 year old video or a five year old blog. And like I said, because I know that has directly come from someone's pure understanding and they're going to write it as raw as possible. And they're not going to put a layer of like polishing with it with AI because I don't need that. I want to know your thoughts. I want to read you as a reader.

And I'm not going to judge you if you have a spelling mistake unless like it's your interview application. I'm an interviewer, but I'm going to judge you if you're going to give me the same in the age of blah, blah, blah, blah. And like the same steps again and again.

Of course. I'm so yeah, to the point being like, I definitely keep an eye out for the kind of, I actually read very less now, which is unfortunate because I am, I'm, I want to make sure that I'm investing my time in the right thing. And that's where X like, for example, like it's so impulsive. And I don't think people even have the time to pass their tweets through AI before like, like you're just like writing posts when you're in like I most of my tweets are impulsive tweets.

And that's, that's what they call shit posting, I guess. And that doesn't really happen with LinkedIn. I doubt anyone like like, yeah, I don't really like tokens. No one's going to like write a two letter thing. It's always a long essay on LinkedIn, so it's again a lot of energy involved as well, I feel. It's interesting that indeed, I think people are looking for places to be authentic and to

also find authentic people. And if you have kind of a shorter context, I mean just I don't know how how Twitter is nowadays, but before I think a few years back you could only do X amount of characters. I think that has now increased, but if you could only do 50 characters or only do 100 characters, then it would be very hard to. Why would you do a prompt to get those 50 or 100 characters?

Then you just write kind of what's on top of your mind because it's only this small anyway and LinkedIn is not like. That no, even on X like there are these like bots that people now like make which can give AI generated answers. Oh yeah. And I have this block policy. Like I will specifically take time out to block you just because you gave me an AI generated response.

Like if you would have just said anything else like you would, even if you've said something like critical about what I wrote, I would have actually engaged and I would have like shared my thoughts on that because yeah, yeah, I'm there to like spend time as well. But if you're going to give me an AI generated response just for like getting more views or something, I am blocking you and I'm first going to call you out saying like hey, I am blocking you for creating AAI generated

response and then. Yeah. Do you use any AI in the content that you write? Yes, of course. Like when you're writing things, for example, like release notes or when you're writing things that will be read not just for learning purposes, but for compliance purposes or things like where your spellings

actually matter. And like this is something that will be passed from one documentation, again, is a very good area where you can use AI because that's not somewhere where I'm looking for emotion or to connect with someone. I'm, I need to know how things are done. And that's where AI does help to have a faster flow or just connecting the dots. And I'm OK with that. But yeah, again, I feel if you've been a writer and you've written like written your real thoughts, you will still be a

good writer. AI is not going to make you a bad writer. But if you're starting to write with AI, then I'm a little bit worried. Like then I would, I wouldn't. Writing is something I don't think you should do with AI specifically. Like make it write your code, that's fine. Like make it write your CSS code like that's the best thing.

But write your raw thoughts. I feel like your thoughts should be your own and that's where you should definitely shouldn't use AI. Yeah, I feel like this is this is the same with regards to, OK, what is your goal in the end? If you have to write content to have a landing page for a product that you're selling, I would say sure, use use AI. But if this is your thoughts and you're sharing your learnings or like your writing requires critical thinking, that is part

of who you are. And if you let AI kind of go HAM, is it then really your thoughts? Is it then really you being authentic or is it something that is just being regurgitated and that is kind of copy paste able everywhere else? You know what, like I have never really understood the meaning of the word like, you know, like in SEO marketing, like there's this word known as copy.

Like can you write me a copy? And well, it is really important for like SEO like generation and all that, but I've never understood that. Like what do you mean by I need to give you a copy for this? Like for me, because I've come from like pure like just writing background or lighting something when it was stuck for me and now it's unstuck and I want to get it out of my system for myself and maybe for somebody else as well. It's just a flow of things from

how like what is the problem? Like how am I solving it? And like, why did I even do this? And like what did I, what different ways I tried to do that? And I've connected to similar writing as well. I've never connected to a copy like but yeah, it depends. Like it's again what you're trying to achieve but but I personally have found found it very difficult to connect to a copy.

Yeah, I've never thought of the word copy, that it could also mean like it's a copy of what someone else already wrote, basically that it's a placeholder for something. Yeah, yeah, yeah, yeah. But I wonder if those two are related. I mean, I, I like what AI is bringing us. I also see that there's some

downtime or, or downsides for sure. Like the human to human interactions are now sometimes replaced by human to AI interactions. And that makes me a bit more fearful. We already kind of in this tech field have this stigma from an outside perspective that is just people typing and typing away. And now with with an AI buddy, it's like typing, typing away

and you have an AI buddy. Yeah, I feel like empathy is going to be a bigger topic going forward, that people still need to relate to each other, understand each other, have the patience to listen to each other. And AI might be taking away from that. Definitely. And that's where like a like for developer at Wacosphere, there's the whole domain of developer

relations. Empathy is a very big, very big part of our role to not assume that somebody knows something about your domain or technology or tool and to reach the developer where they are as long as like it makes sense for the business as well. Like, of course, you won't go completely to completely from A-Z, but to always have empathy about not assuming what they might know or what they might not know. And this is something that should be reflected again in like your documentation or like

anything. You have to start from a subjective 0 and then go to like 1000 when you're trying to like explain how your tool works. And I still like that. Like I still like when documentation has like these markers and pointers, like click that arrow, click that button and like, because even though like I'm a tech person, when I'm building IKEA, I want to just see images like I want to see. And it's so well, like IKEA documentation is so good.

Like they will match the same thoughts of like those screws even on their small images. And I use that like imagine if like the the the people that I get like no, you need to be able to like just figure that out and like why? Why don't you know that Seriously doing? Trial and error like figure it. Out and then we would have broken furniture. So I think we we, we are very fortunate that we are in this industry where we can break things and look at it scientifically, mathematically

and technically. But just assuming that everyone should be able to do that is really not the way to, like, go forward. I feel if we still want to be in a mixed community of different people. Yeah. Or else everyone's going to just like, put us out and be like, you just sit with each other. Don't talk. To us, yeah. In the end, like we're building for humans, we're building for people, and we're doing that in so many industries.

So the patience and empathy to understand each other, like finding each other as humans, that's going to be the, I think the skill that's going to be most valuable going forward.

Absolutely. And I think I do feel like if you're talking about purely like business perspective as well, there's going to be an I do feel like I see a niche coming up of human interactions now where no matter what, we're no matter what AI is helping us do. Like I see like these AI cooking robots, which is great for a family that does not have time or like they both of them are busy, they have kids and like, they still need food And like, they don't want to have like a

restaurant food or like crappy meals. They want to have good homemade food. And a robot could be a really interesting tool there.

But we when we're talking about like a special moment, like it's your, your kids first birthday or it's your, it's your 10th anniversary or something. And you want to go out of your way. What would you want? Like would you want a meal prepared by a rhubarb? Or would you want to go to that Noma chef and like that one restaurant that you've been like dying to go and like have that meal from that person's like me specifically from that person.

And I think that's like physical experiences are going to get more and more expensive, I feel. And I'm scared will access to them also get reduced for overall population? And will it be like more access to people who just have more money? Maybe, I don't know. Yeah, I really don't know. I, I just hope it doesn't go

away. Like the thing, the cooking example, I enjoy cooking, but then I also have moments indeed where there's not enough time and I order or, and the joy of creating something, especially if you have a partner at home or if you have kids like it, it should still be there. So I hope that never goes away, that you always have the option,

because I like having options. Yeah, and I do sometimes feel that way though, like if this meal would have been cooked, would I like just ship that thing more like? And I try to tap myself out like there's only so much I can do. Yes, there are different people, like who everyone has their priorities. And if a building is their your only priority where you just like wake up and you build and like that's it. Great.

Like, I think they are the kind of people due to which we have, like, EVs and like, there are some people who are making it their life's mission to just build things. And I know there's this kind of sacrifices. They're doing it for it. But is it the right thing for me? Like, it's something that I need to choose. Decide if yes and then go, go for it. But if not, then I shouldn't have to, like, beat myself up to it. Like, no, like, let me just eat that food.

Like, let me just take that take away because I should be building right now and not cooking a nice meal for myself. Yeah. Yeah, I get that. I've really enjoyed this conversation, Platin. This was a lot of fun. Same. Yeah. It's really good to, like, just tap out of the typical daily conversations about like, oh, what did you make? What did you build? What did you generate? But more about like, how is this affecting us as, like, individuals and developers?

Like, we take so much pride in calling ourselves, like, builders and developers, but I hope we're building ourselves to be good people as well along with this. Yeah. Yeah, I mean, if you see yourself as something that is also software and evolves, then, I mean, reflecting and thinking about things like these and having conversations similarly are necessary, I think. So, yeah. Thank you so much for coming on and sharing. And we'll we'll round off here.

If you're still here, leave us a like. Let us know in the comments section what you thought of this episode and we'll see you on the next one.