S2E19
· 39:15
I think it's the first time I'm plugging in my microphone to this Mac.
Mikkel Malmberg:So, like, in forever?
Peter Suhm:Why would it work? Why would it work?
Mikkel Malmberg:Oh, no.
Peter Suhm:It's a new Mac.
Mikkel Malmberg:It's a new oh, it's because it's a new Mac. But apparently, you bought a slower connection one. Wait. Oh, it's good again. I think oh, no.
Peter Suhm:Let's try the five gs. Is this better?
Mikkel Malmberg:Yes, I think so. Looks like it.
Peter Suhm:Let's try this.
Mikkel Malmberg:Let's try this. Let's do it.
Peter Suhm:Back on the Riverside. Dude, my number one takeaway from last time we recorded Yes. Is that I want an in person podcast with someone. Oh, more yeah. Fun than like recording a Zoom meeting basically.
Mikkel Malmberg:Agreed. Agreed. Even though, like, we still have some things together. The the cater founder mode. Yeah.
Mikkel Malmberg:Yeah. So how have you how have you been? It was really nice seeing you in person.
Peter Suhm:Yeah. It was. It's been hectic. Mean, I guess I can talk about the talk that I did the day that we recorded last time.
Mikkel Malmberg:Well, yeah, that's The very reason that you went all the way north.
Peter Suhm:Yeah. To the North Wall. Yeah. I can't remember how much I talked about it in the podcast, but I had this I like I prepared like I created my presentation as a layer of a lab. Because that way I could do more like interactive demos of some of the AI stuff that I was showing.
Peter Suhm:Like, this is how we can use structured output in an LLM. And then like, had a text input, you could type something and then it returned some JSON on the slide. Right? Which was a kind of cool experience. And when I got to my demo of embeddings and vector databases, I could like search live on stage and on the slide, I would see the results of the search based on the embeddings in my postgres database.
Peter Suhm:And and then I could also show the the code snippets next to it. And it it was really, really nice experience as a presenter to not have to switch anything and just you just in your slides. You could do anything you need to do is sort of pre prepared on each slide compared to my last talk where I had to do like fifteen minutes of live coding in Cursor with AI models that may or may not work. Yeah. This was a really nice experience, something I'm definitely gonna do again.
Peter Suhm:People asked if I would turn it into a open source package. I think you already have one for the exact same thing. Mine was just Laravel. I don't think I wanna do that because I don't really care, to be honest. And I don't have that much time, sadly.
Mikkel Malmberg:No, no. That's the thing that you have to remember that like open sourcing shit is gonna it comes with responsibility.
Peter Suhm:Yeah. Either no one cares and then it's just a waste of time or they care and then it's also a big waste of time.
Mikkel Malmberg:Yes. That's exactly. That's the calculation. Yeah. Maybe we should say that if I if I seem a little bit distracted it's because there's a puppy there.
Peter Suhm:It's your ADHD is going crazy.
Mikkel Malmberg:It's because I have a puppy present.
Peter Suhm:The ADHD is present with us in the room.
Mikkel Malmberg:Yes.
Peter Suhm:In a furry.
Mikkel Malmberg:He has just ADHD has just found a licorice.
Peter Suhm:Oh, delicious.
Mikkel Malmberg:So at least we get some cleaning up. So this is Vera. She's a new newest addition to the family as if we needed any more living beings in this house. But, yeah, she's really cute. So that's why I'm looking her out a bit.
Mikkel Malmberg:This is the first time she's in my office, and it's hectic.
Peter Suhm:Yeah. Nice, man.
Mikkel Malmberg:But, at least she's cute. Thank you. I don't know. That's awesome. I've never had a dog before.
Mikkel Malmberg:I never had a dog growing up, neither did my wife. So we're just figuring out what
Peter Suhm:is You think it's a dog? Yeah. As best as
Mikkel Malmberg:you can
Peter Suhm:tell when you got it.
Mikkel Malmberg:But it's really it's really nice. It's really nice so far. So, yeah, we'll see. Apparently, they become teenagers at one point. Hopefully, not at the same time as my human kids become teenagers.
Mikkel Malmberg:But who was he? Probably before. Yeah. So how did it go with the talk? Was nice.
Peter Suhm:It went well. It's like a tough crowd, you know, how people are in your sort of part of Denmark. They just kind of stare at you, and then they don't say anything or show any emotion or clap or anything. Like you don't really thankfully we had like a couple of like sort of core Laravel community people there for from like our Laravel Denmark org so I had like my crew that was like clapping and asking questions and stuff.
Mikkel Malmberg:You're hype man? But
Peter Suhm:like the people the local people were just like staring at me. Yeah. And yeah, there was no question. Like normally I get like ten or fifteen minutes of questions and there was not a question. So then I just try to ask them questions and I was like, has anyone like tried any of these things or experimented with it and like no reaction.
Peter Suhm:Is anyone breathing? Afterwards like people were coming over and they're like, yeah, I did this crazy thing with embeddings and we had this issue and then we solved chunking in this way and I was like, why the hell did you not raise your hand so everyone could hear it because that's exactly like what would have been interesting.
Mikkel Malmberg:Joke But it's
Peter Suhm:just the culture thing.
Mikkel Malmberg:Yeah, the jokey trope is that like if person from Northern Jordan says, like, it wasn't bad, that's the highest amount of praise that you can ever
Peter Suhm:Yeah. That's insane. Yeah. That was a little bit thing, like, every time I so I try to, like, ask people afterwards, like, like, what do think of the presentation like I'm trying to get some feedback for this thing and everyone I talked to was like yeah I talked to my friend and he and he liked it. What?
Peter Suhm:So weird. Yeah. I never got to the friend,
Mikkel Malmberg:I think. No. No. No. No.
Mikkel Malmberg:It's just like
Peter Suhm:But then the office was crazy. Like, you could just tell the real estate is more affordable in the Oh yeah. Yeah. So it's like a massive office for 20 I think it was 44 people working there and like this giant office with like right on the waterfront with like all this stuff and gym and indoor trees and they had like a person making tapas for people at the meetup and they had wine, a wine fridge and I don't know, this is kind of crazy for a meetup. It's really fun.
Peter Suhm:I had
Mikkel Malmberg:What was the company?
Peter Suhm:Home Runner. Runner? They're like logistics ecommerce thing.
Mikkel Malmberg:Right.
Peter Suhm:It used to be called Cool Runner. Now it's called Home Runner.
Mikkel Malmberg:But it's B 2 B.
Peter Suhm:Does it It's yeah, it's B 2 B. Yeah. That's how I not planned on drinking, but they sort of dragged me.
Mikkel Malmberg:That's how every good story starts. Yeah.
Peter Suhm:I found like a really late night pizza place.
Mikkel Malmberg:Nice. Yeah. We have a few of those. We get like giant pizza slices.
Peter Suhm:And then the next day, I sort of like took the whole day sort of off except like the most important like email stuff and originally I had planned to take the train there but then the train was delayed and I was like I'm not I was already so upset with the trains that I was just like screw this and I just drove instead for the four hour hours that it takes to get there. So then the next day I had sort of the whole day free and I had a car and I was in a part of Denmark that I rarely go to so I did a huge like detour around all of Jotlin basically.
Mikkel Malmberg:Oh, nice.
Peter Suhm:I went to the beach which is like Oh, wow.
Mikkel Malmberg:Which one? Which side?
Peter Suhm:Oh, the west side.
Mikkel Malmberg:West side. Yes.
Peter Suhm:Yeah. That's the wild side. I went to the Mountain Of The Heaven, which used to we used to think was the tallest mountain in Denmark. And by mountain, I mean, it's a 160 something meters tall.
Mikkel Malmberg:Yeah. It's like a hill. In every other country in the world, it's a it's a big hill.
Peter Suhm:It's yeah. It's a small hill. But yeah.
Mikkel Malmberg:It's like we could only the only the Dutch we could impress with our mountains, quote unquote.
Peter Suhm:Yeah. That was kind of wild, just kind of fun day, like driving around. I had a really good time, actually, like I listened to some audiobooks, and I had a lot of sort of thinking to do both around OGKid and some of my stuff at Tailwinds. It was really nice to just have a a day with like no I had no time to like that I had to be back home. Yeah.
Peter Suhm:Because my train ticket was supposed to get me home at like in late in the evening. So I had sort of all day to just bum around. And so it was and I kept changing locations, which is helps me a lot when I'm trying to think through something something. Yeah. So I really enjoyed the time there.
Peter Suhm:And I had some really good sort of insights about what I want to do at Tailwind, some new ideas for for OGKid and yeah, so overall was a really nice trip even though it's only two days. Like I planned it in a way that I would get like the maximum amount of time away for the one night.
Mikkel Malmberg:Yes.
Peter Suhm:So I left at six in the and I was back at like seven in the evening something like that the day.
Mikkel Malmberg:That's, that efficiency.
Peter Suhm:Yeah. But then, my wife had to go to London, for a couple days. And then when she came back, just as she came back, this our smallest child got really sick and was sick for like two days. And then she got sick. And so, like, I came back with like this energy of like, wanted to be really productive.
Peter Suhm:And then I just had more than a week of just like the minimum amount of productivity. Like I barely got the most important things done and it was just so been so frustrating. So like just like this week I've had the same to do item for OG kid every day of the week that I just moved to the next day every day and I don't do it ever. And it kind of kind of made me realize that like like I keep saying OG kid is a side project but I keep doing dumb things where I treat it as like my job.
Mikkel Malmberg:Right. I'm
Peter Suhm:getting stuff for OGKid to the same to do list that I add like Tailwind stuff. Right. Whereas like the Tailwind job is actually my job and it's important and the OGKid is stuff is more for fun and it just shouldn't even be on the same list because it's just frustrating to Yeah. To not finish what's on that list because then I feel like I'm doing I'm not doing my job. Yeah.
Peter Suhm:And I kinda have been doing my job. I just haven't been doing my side project job, you know?
Mikkel Malmberg:Yeah. Yeah. It you should do like me and just have no lists. That's better.
Peter Suhm:Yeah. No. I definitely need a list for tailwind stuff. Just No tasks,
Mikkel Malmberg:I'm no wrangling. Just free reeling it.
Peter Suhm:Maybe if you're just burning crypto money with no direct purpose other than just
Mikkel Malmberg:Yeah. That wasn't fulfilling either. So I'm still searching for the for the real enlightenment. Yeah. It's it's true that I wrote a a library like it before.
Mikkel Malmberg:It's called Maximum Over Business, and it was way back in the day where React was still, like, create class and, like, class based things and wasn't so functional components and everything. And I actually wrote a library for it called called Dinkus. That's, like, supposed to be for, you know, the Logitech pointer things to like have have to support those. What's up with the door?
Peter Suhm:Give it a bone.
Mikkel Malmberg:Yeah. Maybe maybe a microphone. What do you wanna say? Yeah. So it's called Maximum O Business.
Mikkel Malmberg:And it was for the same reason that I wanted to show, like, web things. And having to go back and forth between Keynote and Zafari is just like, ugh. Yeah. Much more fun to have it integrated. But then And it's also hard
Peter Suhm:to make slides that look great
Mikkel Malmberg:Yeah.
Peter Suhm:Compared to just, like, making them with CSS and
Mikkel Malmberg:Oh, and now we can just, ask Claude to make it nice. Improve my slides. And but then, of course, the problems show up when whenever your own HDMI cable dongle shit doesn't work and they have to convert it to PowerPoint and it's just like, yeah, that's not gonna happen.
Peter Suhm:Yeah. Yeah. No. I deployed a version online and I had it on my computer locally and I felt like that was enough. For a meetup, for a conference, would have like a dumb copy as well.
Mikkel Malmberg:Yeah. Yeah. But for That makes sense.
Peter Suhm:For a meetup, it was fine. You know, worst case, I would just do like a manual demo thing at the meetup.
Mikkel Malmberg:Yeah. And in that case anyways. It wasn't bad. No. Yeah.
Mikkel Malmberg:So that's mostly it. I I mean, that's always how it goes. Right? Exactly at the moment where you think, like, I'm gonna log in and get shit done, and life shows up. And you get a dog and suddenly your podcast has a lot more squeaky noises.
Peter Suhm:Yeah. It's kinda nice ambient dog sound.
Mikkel Malmberg:Ambient dog sounds. Yeah. So anything else that's been on your mind?
Peter Suhm:Yeah. I mean, I've been OGKid has been on my mind. I had
Mikkel Malmberg:Aaron for his time. Now.
Peter Suhm:Which was a really cool
Mikkel Malmberg:Some prominent ones. Yeah.
Peter Suhm:I had so I had Aaron sign up and he even mentioned it very briefly on his podcast, which was, really fun. Yes. And his cohost Ian Lanceman is actually also a customer. He just hasn't used it yet. He just bought it because he's a nice guy I
Mikkel Malmberg:think. We like the idea. Called him the godfather of OG kid.
Peter Suhm:Right. That would be nice if you turn into that. So that I've been thinking about sponsoring their podcast, because I feel like there's a it's a really good sort of overlap with the people I'm trying to reach. Yeah, and when Aaron signed up, I was like, this is the best time I ever get sponsor their podcast, because now they can actually talk about them using it. So I pulled the trigger and dropped the the $1,000 on a podcast sponsorship.
Mikkel Malmberg:Nice.
Peter Suhm:First time which is less money that I've made from OG kit. So it still felt good. And it felt
Mikkel Malmberg:like I'm an investment.
Peter Suhm:It's definitely an investment. And I think like, it's not, I don't expect to like make the thousand dollar back like that month. But I do like, I feel like people will people won't necessarily be able to remember that it's OG kit, but just because Aaron can say that he uses it. I think people will be able to remember next time they have the problem that Aaron used something to solve this problem and they'll be able to figure out what that was. Yeah.
Peter Suhm:So that's why I feel like even if they don't like remember that it's called OG kid or what the URL is or whatever they they they'll be able to remember that they heard Aaron talk about solving the problem with some tool, and then hopefully be able to find it. So that was sort of and then, you know, it's it's a fun side project. And it's cool to be able to buy a podcast sponsorship for some of the money you made so far.
Mikkel Malmberg:So
Peter Suhm:it's just, yeah, it's cool. And then for tailwind, I've just been like, really busy with the partnership stuff that I'm doing there. And I'm sort of pivoting the whole model a little bit right now. Just because we have a lot of partners now. It's been pretty successful.
Peter Suhm:We hit a pretty big milestone recently, which was a million Canadian dollars in annual recurring revenue for the partnerships, which has mostly been me cold outreaching to people and be like, hey, do you want to sponsor us and people being like, I don't know. And then me following up and so actually doing sales like the good old way is how most of that money has actually come in. So it's pretty big milestone, pretty, pretty crazy to see it kind of working. But the problem is that we're sort of running out of, like, the the value of having your logo on our website goes down the more logos we add. So I'm trying to sort of split the program into two because there's also a lot of companies gold
Mikkel Malmberg:sponsors and the silver sponsors.
Peter Suhm:No, no, actually, not even sponsors at all. Because there's some people that care about the marketing aspects. But there's also a lot of companies that care about having access to our team. So I'm trying to pivot into more of a sort of enterprise programme where, where you don't really get any promotion, you're just sort of buying access to the tailwind team. And so we can like expedite things and give feedback and, you know, give feedback on a product or help like figure out as a technical issue or some strategy related thing And trying to see if that could be successful as well because it is something we have a little bit of demand for and that we for that thing, we don't really run out of like logos logo real estate.
Peter Suhm:We can have quite a lot of customers that have access to the team because in reality they don't use it as much because it's more of a like insurance and then they need it for like a project once in a while. So that's like kind of the next big thing. That was one of the things I figured out as I was driving through Jotland that there's like a good solution to that is like we actually should just have like a different product where people that don't care about having their logo on a website shouldn't even have their logo on a website because we should just save that for people that actually care about it. Like we've had really big companies like FANG level companies reach out and ask if we could like add classes for different things to tailwind, you know. And it's like, yeah, we should have a relationship with you.
Peter Suhm:If you care this much about tailwind, there should be a way for you to give us money on a regular basis and your team to talk to our team, you know. So that's basically what I'm trying to figure out.
Mikkel Malmberg:Give me one second, Peter. I think I'll just have to find some water for her or something. So short Cool. Alright. Let's try to see if we can make it through your last OTKIT
Peter Suhm:point. Yeah. For OTKIT, like like I said I had that thing on my to do list for like one and a half weeks now where I kept moving into the next day. I did actually work on it, but basically I wanted to launch a new I have a new pricing sort of figured out for OGKit. And in order to launch that, I need to build like a deeper polar integration because right now it's like very simple.
Peter Suhm:Like I did the simplest possible thing right now. Basically, the registration page is like behind a secret URL, which is kinda dumb. And then when you when you pay in Polar, I just redirect you to that secret URL so you can create your account, but it's not connected. Like, the app doesn't know about your Polar subscription. But the new pricing is gonna be more tied to you what subscription you're on, like how many pages you can use OGKit on basically.
Peter Suhm:And in order to do that I need to like sort of re or build the actual like billing logic and subscription logic on my end. Right? And the the thing that I realized is when when something is a side project and I only have like an hour here and there like in the evenings and stuff where I work on it and I also use AI agents and when they do a lot of work for me and I end up with a PR that's like 15 files changed and it's a really large div. And I only look at it like, you know, once every other day, like it it stresses the hell out of me actually. And it's just too much to wrap my head around.
Peter Suhm:Like, I would feel different. It was something I worked on for eight hours a day, even if the agent did a lot of the work just because I'm like in it. But when I like check-in after two days, okay. Okay. Let's see if we can like finish this PR.
Peter Suhm:It's like then like the context switching is just too much basically and it is like too much to load into my context. Yeah. So I realized that that that's just not not gonna work for me and it's not gonna be successful. So so basically what I realized this week is like for for OGK, like the only viable way for me if I'm gonna like use agents to write most of the code is to really like chip away at things in a very small way. Yes.
Peter Suhm:And like something that Derek from Savical has always been really good at is like whenever he's building something, he always like ships the whole feature before he actually ships the feature. But it's always there like behind feature flags and he prepares the database and he like the the first thing he'll do is like you know the migration for everything and just get everything ready in the back end and he's really really good at that like sort of little controlled pull requests like that it's even like a
Mikkel Malmberg:it's a good approach even without coding agents. Right.
Peter Suhm:But it becomes critical with coding agents because they're extremely good at just like quickly creating like a thousand lines of code, you know, that you just don't have time to like sort of internalize. So like, I really realized that it's I just I need to sort of know in advance what I want. It's fine to like do a big PR just to test something out. But I'm not going to ship that.
Mikkel Malmberg:I know
Peter Suhm:if I if I have an expectation that I'm going to merge that PR, it's going to sort of just hang over my head, which is going to be annoying. So I really need to like, I really need to sort of work more in a way where, okay, I know I need a subscriptions table for this. Let's just make that. Let's just ship it even if it doesn't do anything. Okay.
Peter Suhm:I need to know I I know that I need to listen for the Polar webhooks. What if we just, like, built the stuff that listens for the webhook, fires events, maybe we lock them to the log just so we have, that logic in the app already without doing anything with it. And then, okay, once I have some more time, I'll build the logic that actually listens for the webhook based events and actually apply that, you know, the relevant thing action to the subscription. And it just like, it's a more calm way to work for me just to like get things out there and just always Adam is good at that too. Like basically he he always tries to like let's just see if we can ship something like even if it's not if it's behind a feature flag or even if it's like just a really really tiny thing like let's always try to be in a shippable state.
Peter Suhm:Yeah. And he'll like famously if you see a PR he'll just like randomly merge it you know just because he doesn't like a PR just being there for a long time. He's like,
Mikkel Malmberg:that No, shouldn't no. Can, like, can so easily become like just a giant waste of time because nobody wants to touch it when it becomes too large. It's like big too big to fail.
Peter Suhm:Yeah. And it just like why add that stress to yourself when it's just you and the agent, but the agent is very good at creating a lot of code that then it just when it's in the PR like that, it just feels like it's a growing pile of code that I need to now own that I didn't even create and that's annoying. Yeah.
Mikkel Malmberg:Yeah. Yeah.
Peter Suhm:So I've changed
Mikkel Malmberg:A couple of things that I've because I've noticed the the same thing. So one thing is that like the the Giga PR you have, the branch, you could just throw it away because coding it again is not gonna take you Yeah. Long. So it's like just take take the things that you like and make the AI make a list, describe what what the good parts are and then use that to rebuild it. Yeah.
Mikkel Malmberg:Should almost cheap now.
Peter Suhm:Code twice with AI. Right? At least twice.
Mikkel Malmberg:Almost. Like, at least it's not a problem to write it again. Like Right. The time there's no wasted time in writing it again almost. It's like half an hour and then that's it.
Mikkel Malmberg:And now you're so much smarter smarter the second time around. And other than that, I've also found it to be very helpful to do the planning upfront. So, like, deliberately hold writing code. Like, tell it, don't write anything yet. We need to plan this first.
Mikkel Malmberg:And some of them have, like, plan modes, which help, but others don't and you tell Building them
Peter Suhm:the full version just quick and dirty and try it is like a good way
Mikkel Malmberg:to
Peter Suhm:as to if you think of it as planning rather than like the actual thing, it is actually like helpful.
Mikkel Malmberg:And it's because like it used to be how I figure out things was to just like start and then
Peter Suhm:Yeah.
Mikkel Malmberg:Oh, there's a problem here. Needs think about that. Then two steps forward, one step back almost, and then eventually you get to a point where it's good. Where now it's just like you have to prime the cannon in the right direction before you fire, because as soon as you fire it can go really far. And if it's in the wrong direction, it goes just in the wrong direction very fast.
Mikkel Malmberg:Yeah. But the thing is you can
Peter Suhm:just fire again.
Mikkel Malmberg:Yes. Yes. That's that's right. And it's so cheap now. I've actually just the other day found a project called beads, pearls like you put on a necklace, beads, beads, beads, is a terminal git storage based issue tracker that you're supposed to use with the agents.
Mikkel Malmberg:And for most of this stuff, like the trackers and the specs and the all the bullshit, it's just like I've never found it useful almost. Like I don't even have commands anymore. It's just like raw dogging prompts every every time. But this one actually seems like it's a better approach than having all the markdown files which I've been using, like the task lists and descriptions and everything. Because that's the way to like any medium to large task, you almost always have a way to reset the context, which means, like, start a new thread almost, like or exactly.
Mikkel Malmberg:And so, like, keeping the important parts between threads is almost has always been, like, just saving a markdown file for me. And the beat seems like it's like that, but automatic and then not directly in markdown, but in, like, a ish format with, a database ish structure. So, yeah, still still investigating that, seeing whether it's actually good, but it looks like it is actually helpful.
Peter Suhm:I feel like
Mikkel Malmberg:But, otherwise, the Markdown files is, like, good.
Peter Suhm:Art Are is really good at, like, figuring out your code base Yes. By just grabbing. Grab is, like, really had a comeback.
Mikkel Malmberg:Yeah. There's one called AST Grab as well, which the agents like to use as well.
Peter Suhm:Okay.
Mikkel Malmberg:And then, yeah, just regular rip grip or what it's called. Yeah. They're very fast to get up to speed and just smarter every day. It's crazy.
Peter Suhm:Yeah. But yeah.
Mikkel Malmberg:Actually, it yeah. I showed you a project last last week no. The other week when we met. There was like it's it's still stealth mode. It's a side another side project.
Mikkel Malmberg:So a dog and a new side project.
Peter Suhm:But it's cool. It's cool.
Mikkel Malmberg:Yeah. Yeah. Yeah. But it's I actually started that effort five years ago. I just looked up the first commit.
Mikkel Malmberg:Wow. Just so quickly. I I spent a few days and realized so quickly this, like, the scope of this is too big. Like, I will never get anywhere near done. But then now, like, it seems within reach.
Mikkel Malmberg:Like, suddenly the whole universe has shifted. And just think about it. It's like February. I was complete I was completely like, my workflow day to day was completely different in, like, six, eight months ago to now, which is so crazy. And at hole punch, the company I work at now, I had to do some, like, some just grunt work of, like, updating CI scripts and, like, just adding a linter that I also built and stuff like that.
Mikkel Malmberg:Which like it was it's so helpful to have the the coding agents for that as well. You could just tell it like, make a list of all the repos, then make a list of all the PRs, open a branch in each one and do this exact these exact steps, then open a TMux window in each project, and I can go through, like, all of the shit that you had to do by head. It's just like yeah. It's so crazy.
Peter Suhm:I just had I just asked Opus three five to do, is that what is it? Four five? Four five?
Mikkel Malmberg:Yeah. That's the new one.
Peter Suhm:Yeah. The new one anyways.
Mikkel Malmberg:Yeah. I just asked
Peter Suhm:it to, like, do a Black Friday sale for OTGit. It's 30% off.
Mikkel Malmberg:Oh, nice.
Peter Suhm:And it just like did the banner and like updated the links because that was just yeah. I didn't really have to do anything except like give feedback on the text it wrote in the banner.
Mikkel Malmberg:It's so crazy. It is completely nuts. And it's so weird to me that there are still people who are like holding back on this. I really like it's almost like, I don't know, deliberately wanting to stick to bulls and horses instead of buying a tractor. Like, why
Peter Suhm:It's also wild to me that people look at it and think that we won't need developers in the future. That's that's really not my takeaway.
Mikkel Malmberg:Also that. Like I came up with a
Peter Suhm:With great power comes great responsibilities, you know.
Mikkel Malmberg:Yeah. Like stating that normal people will be writing apps as soon as like no. Normal people are saying stuff like, my printer doesn't work. It only works when Lars uses it. And then Yeah.
Mikkel Malmberg:Two weeks ago, I needed to print in color and it didn't work because it was like the Saturday. And that's how normal people think about computer problems. And yeah. So there's a there's there's a bunch there's a long way to go before they'll be writing apps. So yeah.
Mikkel Malmberg:It's exactly. I read a nice blog post about it. It was a newsletter maybe that Internet friend Justin Searles posted, which were like the argument was sort of in the ways that that, like, good software development practices are as important as ever. Like even more so, because, like, testing, that's the way to make the agent, like, really fly. To have it, like, give it a feedback loop that it can run itself.
Mikkel Malmberg:And, like, that we just spoke about PR routines and, like, just having, like, a nice decent way to to structure your work in a way that doesn't mess everything up. Just like of course, that's still important even though it's a computer writing all the code. And even more so, because now you don't know the code. It's like, okay. So where do I even start when I need to learn this?
Mikkel Malmberg:And of and in many cases, the solution is more agents. So I've been sometimes when the PR becomes large and I'd like don't even if I do like a spike in trying something out, I don't but I don't even care about the code right now. I just want to see like what does it look like when it works. So I just like prompt prompt prompt prompt until it works. And then of course, there's a whole, like, pile of code that I've never looked at even because I just wanted to see how it works how it looked like when it worked.
Mikkel Malmberg:And so where do you start PI ing PR reviewing that, like, from the top is, like, sorted by the alphabet. There's not there's not gonna be the best way. So I'll just pipe the whole diff to NH it and then tell it to structure it in a way that like from the outside in, in a way that makes sense and is easier to to review and it just does so. It's like you just have to like every step of the development process you have to rethink in a way to think like how how can I do this smarter now? And likely, can.
Mikkel Malmberg:You just have to, like, think about it from the right angle, I guess. Yeah. It's just crazy. So yeah. That's the that's the news from here, I guess.
Peter Suhm:Yeah, I mean, for just to sort of wrap up OG kid, I guess, like, I I it's like my the favorite side project I've ever had in my life. Like, it's like, I really love it. Because it's like it's a it's a kind of hard problem, but it's also very simple. Like, it's a simple outcome and it's like it's a simple product because there's one API endpoint that everyone uses. Yeah.
Mikkel Malmberg:And Because it's GraphQL. Right? That's that's what you mean.
Peter Suhm:Exactly. Oh, I haven't heard that name in a long time. But it's it's like so the concept is like it's it's hard to explain but once you see it it's like pretty easy to understand I find and yeah it's like it is a fun problem to solve and it's also fun that people are actually willing to buy it. Have 28 customers now which is is cool. That's nice.
Peter Suhm:Yeah. So it feels it started to feel like a real thing. It just needs more work once I have some time.
Mikkel Malmberg:Yes. And,
Peter Suhm:yeah, maybe that's my last or maybe a life of date is, I don't think I mentioned this that last time, but we're going to Spain for a month. So in about two weeks, a bit more than two weeks, we're going to Spain for a month, we're going to spend, Christmas and New Year's there. And the first week I'm gonna take the week off and the next two weeks I think probably everyone at Tailwind has that time off but I still do support but it's you know limited compared to normal work when it's over the holidays. So maybe I'll have a bit more time to play around Oh. Side project.
Mikkel Malmberg:Don't Maybe jinx it,
Peter Suhm:I'll just
Mikkel Malmberg:Get a dog.
Peter Suhm:Out with my kids on the beach. Alright. Yeah. Or maybe a mix.
Mikkel Malmberg:Is that not what it's called in Spanish?
Peter Suhm:A what? A dog? A dog? I think it's called That's a rock. No.
Peter Suhm:Perro. Perro. And a hot dog is Perito Caliente.
Mikkel Malmberg:Oh, nice.
Peter Suhm:A little warm dog.
Mikkel Malmberg:A little warm dog. That's what I have in my lap right now.
Peter Suhm:A little warm dog?
Mikkel Malmberg:A little warm dog.
Peter Suhm:Name in Spain is Petrico. Little Petrico.
Mikkel Malmberg:What am I thinking about with the rock? Oh, it's Pierre, the French name Pierre, also means rock. And Steen in Danish also means rock and is the name. And of course, the rock in The States. So works everywhere.
Mikkel Malmberg:Yeah. Yeah. I think that's
Peter Suhm:Perito Caliendo on the beach in two weeks.
Mikkel Malmberg:That's really nice. That's really nice. La Playa.
Peter Suhm:Exactly. As I always say.
Mikkel Malmberg:And so does that mean podcast break?
Peter Suhm:I think so.
Mikkel Malmberg:Yes. So we'll see each other in the
Peter Suhm:I'll send a postcard. The new year almost. Yeah. We're coming back in the January.
Mikkel Malmberg:I wonder what development looks like at that point.
Peter Suhm:Yeah. Yeah.
Mikkel Malmberg:Could be something completely different again.
Peter Suhm:Maybe I don't need to come back.
Mikkel Malmberg:That's right. That's right.
Peter Suhm:That'll be kinda nice.
Mikkel Malmberg:Took a month off and then never came back.
Peter Suhm:Yeah, like going to Spain sort of out of season for a month, four people rent a car for a month, like an apartment. It's like not very expensive.
Mikkel Malmberg:No, I don't imagine.
Peter Suhm:Yeah. It's like less than people definitely less than people spend on a week in the summer.
Mikkel Malmberg:It's like, it's more expensive in like the energy department with two small That's the expensive part. It's not the money. Money doesn't But mean
Peter Suhm:I know that kind of sucks to be in Denmark as well when it's dark at three in the afternoon and it's cold.
Mikkel Malmberg:Yes. It is really, the darkness season mentally and environmentally. Yeah. Yeah. So, I'll report back.
Mikkel Malmberg:That's nice, Peter. Well, have a nice trip, and we'll see each other on the other side of the whole season.
Peter Suhm:Yeah. And I was thinking next time you go to Copenhagen, you should schedule it in a way that you could jump off the train here for
Mikkel Malmberg:an hour
Peter Suhm:to an in person podcast I'm literally That's at the train station, you right.
Mikkel Malmberg:I will. So we'll both enjoy your slow Internet.
Peter Suhm:Yeah. But we don't need Internet.
Mikkel Malmberg:I forgot. I forgot how real life works. Yeah. Well, we should, Peter, and let's find the time for that. And, otherwise, we'll see everyone the other side.
Mikkel Malmberg:Bye.
Listen to Onto Something using one of many popular podcasting apps or directories.