My Day at KotlinConf 2026 in Munich! - Vibe, Tech Interviews & More #anzeige

Hinzugefügt: 2026-06-04

[00:00:00]Does this whole trend of AI actually worry you more or hypes you more?

[00:00:05]>> Right now it's somehow complicated to to keep up overall more worried. it's um helping me more I would say because >> at least with the information that I have right now I think um Hey guys, welcome back to a new video.

[00:00:30]Today we are at Cotlin Conf here in Munich, my hometown. So I of course have to be there. We have some cool interviews with Jet Brains people with other speakers here on the conference with people who attend the conference just asking questions about everything that keep that is keeping us busy about agent decoding of course about everything that is changing in Cotlin Cotlin multiplatform compose multiplatform and maybe even something about Cotlin for back and there also some cool changes coming there. So I'm going to take you through my day here in a cool vlog.

[00:01:36]So we just came here from the keynote and that's in the end the first real talk that is that is opening the conference where all kinds of speakers are covering everything that is uh changing here. So just a summary of all the relevant changes and you can then decide which kinds of talks you want to dive into where these topics are just covered in in more detail. For example, Cotlin 2.4 has been covered here. We have some uh changes there like value classes with multiple parameters that seems very cool. Uh there are some updates to agentic development with coupe. So the cotlin agentic development framework for the back end side you can use that on either kour on spring boot on all these cotlin based backends or where you can at least work with cotlin.

[00:02:18]some cool updates to using Gemini inside of Android Studio and in general you really feel that the focus of the conference has shifted strongly towards agentic development AI it's in the end a topic that's that's keeping all of us busy just a little disclaimer from me before diving into this we had some really really nice and interesting interviews here with people who really contribute to this ecosystem to cotton libraries to cotton multiplatform as a technology to cotton as a language from jet brains people who shared some internal details that aren't so well known at this point and these interviews got a little bit long which is why I thought about how much of those I want to include in this vlog but I've decided because it would be a shame to not include that much to include most of them here in the vlog and rather include little timestamps here so in case you're not interested in a particular interview then feel free to skip. So now enjoy the vlog.

[00:03:12]>> What do you do as a carton developer? Oh god, that put me in a trap straight away. Uh, so recently in my new job, I've switched to become a React Native developer, but my job is to basically try and convince everyone at my company to switch over to Cotlin. Uh, which I will do. I guarantee I'll do it. I'll find a way. Um, but yeah, before that I was a a native Android developer for like 14 years or something stupid like that. So I love Cotlin. I miss Cotlin desperately, but um, yeah. So, >> and are you already into all this, you know, agentic coding workflows? Is this a thing for you? Um, yeah, it really is.

[00:03:44]Uh, so I am I love Claude, right? I sit in Claude all day to be honest. Like I haven't written properly, you know, like written typed out myself a line of code in probably about a month now, I would say.

[00:03:55]>> Um, until December or the beginning of this year, beginning of January. Until then, I was not using agent so much because u from my experiments, they were not so accurate. So I had to do a lot of cleanup after them. But since then I've been using both codeex juni from jet brains and uh cloud code the most popular options I would say and I find it that if I if I take it step by step and not giving it a too big of a prompt or too big of a ask then it works really well for me and going incrementally I can also review the code easier. That's the primary benefit because otherwise you end up with a 4,000 line pull request that maybe you don't take the time to review and then uh that can lead to problems. So I'm using that but I'm not a power user with three terminals opened at the same time yet.

[00:04:48]>> You can actually only learn the language features if you write the code yourself.

[00:04:51]>> Exactly. Yeah. Yeah. Yeah. Yeah.

[00:04:52]>> But if you're a junior developer right now, it's it's a very tough decision what to get into. get into the AI agents and have code that you don't understand >> or write code yourself which also feels wrong nowadays.

[00:05:01]>> And does it overall like this whole trend towards AI driven development hype you more or worry you more?

[00:05:06]>> Overall more worried?

[00:05:07]>> No, it's not scaring me. It's it's um hyping me more I would say because in the end if you do it right it can yeah increase your your speed of development.

[00:05:17]>> I think it hypes me more. In the beginning, I was a bit um not scared but sad that um I was thinking I won't enjoy coding as much because when I got into mobile development, I really enjoy like creating the UI, linking it to the to the back end, to the logic and everything doing that. So I was thinking, oh, so now AI is going to do that for me and I won't enjoy it anymore. But actually, I do find joy in building features, in seeing the result, even if I don't write the entire code myself. So, I'm excited about the future and I hope um I'm not hope I don't think it will reach the level that it can review itself and prompt itself to build an app from start to end without human intervention anytime soon. Maybe that's the point that will get me scared. But for now, I'm I'm excited. Um but I'm trying to trying to be an optimist about it. Like for the longest time, I held out and I was like, "Oh, I'm not using AI. I'm not using it." And then I was like, "Okay, use chat GPT's codeex." And I was like, "Oh, this is pretty cool."

[00:06:16]And then like Android Studios inbuilt agentic tool as well and now obviously claude code.

[00:06:21]>> If there was a button you could press and AI would be gone forever would you press it? Uh no.

[00:06:26]>> Okay. So you you say AI is overall a net positive for society in general.

[00:06:32]>> I think so. Yes. At least with the information that I have right now. I think um it brings a lot of benefits with of course of course the costs just like social media. People are talking about a lot about social media and how um damaging is for some people and if you use it wrongly or if you like get addicted to it. But with the right amount of usage uh and a bit of discipline, AI is going to uh help us more than it would like make our lives worse. Of course, this is also depends on how these AI companies evolve over time and how they kind of share the power. But um so far I'm excited and I wouldn't uh destroy it.

[00:07:16]>> Oh wow. AI is gone forever and what we go back to the preAI days basically.

[00:07:22]Yes. Yes. Yes. Just because then it's just just for selfish reasons, right?

[00:07:28]Like the power and the the power in the knowledge set that I had before AI was so um selective, right? Android devs, we used to have jobs out the wazoo. And now like with AI it's like okay, yeah, you do AI, okay, great. You do Android, fine. No job for you. Whatever. So yeah, just for selfish purposes.

[00:07:43]>> No, I don't think so.

[00:07:44]>> I don't think so. Cuz you overall enjoy the trench more, the technical progression.

[00:07:49]>> Yeah, it's even if it's challenging, I think it's enjoyable. Yeah.

[00:08:08]>> So, who am I with here?

[00:08:10]>> I'm Simon. and I'm a developer advocate at Jet Brains and I've been working with Colin for 11 years uh mostly on the server side uh with Ktor >> and yeah I'm Sebastian Doo I'm working at Broadcom as a spring former core committer I recently joined the spring AI team and I have been working on introducing the cotlin support in spring for 13 years now >> cool and now in my audience we have a lot of native endo developers a lot of kmp developers some of them are into backend development but not everyone if there is now someone who wants to get into back and with a language they already know which framework would each of you pitch them and how >> so I think kater is a really great framework for building services if you're already familiar with kmp android because uh kmp android specifically uh have been using colum for a long time already and they use corins a lot uh and I think this is a benefit to them because skater is built on top of corin so it allows them to use their existing knowledge about concurren currency in Corines to build efficient services in Kin.

[00:09:13]>> Yeah. And Spring is basically the the leader in the GVM space. We have both Java and Cotlin developers mostly Java developers but we have a very strong Cotlin community with big companies famous company I can't name today but who are working with with Cotlin and it's maybe more for yeah bigger project more involved etc. But we have also some mobile developers that are using it because just there is some spring knowledge and basically yeah you can apply the spring knowledge that you have already in Java also to cotlin and we try to find the good trade-off between idiomatic spring idiomatic cutling and yeah find find a good trade-off there.

[00:09:50]>> Okay Sebastian regarding cotlin backend development what do you think Kato gets right and spring boot doesn't? Yeah, good good question. I I try to not look too closely to cater because I try to yeah get my own inspiration about stuff.

[00:10:05]Of course, we discuss we are we have a very close collaboration between Jet Brains and Spring. I I think the the pure idiomatic DSLish uh configuration etc. It's it could be appealing especially for small to medium project. on our side we are more on a trade-off with some DSLs but also some yeah spring boot regular configuration model which is a bit more a bit more implicit I would say and and so it's really a matter of taste so but I think the configuration DSL are interesting we are providing some but not for everything >> you probably guess my question what do you think springwood gets right but cater doesn't >> um what I think spring does really well uh is that I think it has somewhat an opinionated uh approach. Uh so this leaves I think less uh open questions for the user. Uh and Gator has a much more unopinionated approach. Uh but that also means that typically you need to form an opinion um about some things to build Gator services. So I think that is something that uh Spring does very well uh and they have integrations with almost anything you can imagine in the GVM ecosystem. Uh so that is definitely also a big benefit. Uh in spring >> I as someone who have like internally deployed backends as both cater and spring boot so I've tried them both. Um my experience with spring boot using cotton for it is that if you want to really use cotton for it in a cotton ediomatic way that you have to accept trade-offs. Um, if you want to use Corins all over the place, you can, but you're limited to a few little APIs that you have to use in order to have the full full advantages, have reactive database drivers and these kinds of things. And the conclusion that I drew for for us at PL coding is often times it's just easier sticking to what has already worked for Springwood for decades. Um, sticking to the blocking APIs, but using cotlin because cotlin is just nicer to use overall. Do you agree with that or what is your take as someone who really contributes to this ecosystem? Yeah, so that's a very good question and we get a lot of similar question from the community. Uh, currently we have two different worlds.

[00:12:14]We have the blocking world with GPA, GDBC etc. Spring MVC and we have this reactive world where basically we we expose our reactive stack based on reactor with cool routines flow etc and both have different use case. What we are beginning to work today is on beginning to unify a bit that those stack by basically leveraging virtual threats which are becoming basically the golden default on on on server side but they still like very powerful API like cut inflow and so we try to basically uh in the upcoming months build modernize I would say our blocking stack with Spring MBC to basically um double down on virtual threats provide a way to expose that with cotlin flow or with reactor flux or with regular blocking but modern API that are based on virtual threads and then you we just have a single uh uh yeah stack to maintain and then you choose your API so we are really trying to provide the the best of the two worlds without having to choose a definitive stack and uh and it's perfectly fine to not use cool routines everywhere but if you need cool routines for streaming etc you should be able to combine virtual threads and coin flow and cool routines in a clever way. So we we try to we are going to try to work that on the upcoming months to simplify a bit the choice and remove this reactive versus blocking stack uh thing.

[00:13:44]>> My experience with KTOR and to be fair I used it the last time one one and a half years ago. So maybe something has changed but my main issue I had with it were version conflicts. Um if you have to work with like coin kor cotton all of them share the same cotton version but I then sometimes I had to update a specific library but that broke another one and I like it was very difficult to find a version combination that worked for everything that that was my experience and that for for quite some things you you had to build functionality from scratch because it doesn't have that rich ecosystem like Bring the rain.

[00:14:44]>> So I'm here with >> I'm Nick. I'm um leading the product management for Jet Brains Air here in Jet Brains.

[00:14:50]>> I air was actually a new product to me.

[00:14:52]Um I haven't seen it before and as someone who doesn't know it like my intuitive impression was it's kind of a a jet brains based cursor. Is this true or do you want to correct this in some way?

[00:15:02]>> Well, it depends on what you how do you define the jet brains cursor? So we when we started to work on uh JetBrint Air uh we thought that there's a needs coming from the developers to have agents being like a first class citizen in the space yet we're aiming to work with professional developers. So they want to be in context of their project. They want to have um a transparency of like what agent is doing with within the project and have the ability to actually precisely manage the context they're giving back.

[00:15:36]>> And is this like a complete substitute for an existing IDE or does it extend that?

[00:15:41]>> I would say it I would say it extends that because you cannot just say like okay now I will fully switch to agenda coding right it never happens in reality especially in brownfield projects. So most likely what I observe right now and what I observe within jet brains is that people are still working within uh idea but then for specific tasks uh they just delegate that using air within the same space within the same project they might work all together on the same feature.

[00:16:10]So start with the agent do this boiler plates uh and then just opening ID and fix small things or just adjust them u so it actually on the bar of the quality they need. Um so I would say it's not a substitution. Uh we'll see how it goes in the future. Uh for now definitely this is like a a partner in crime for the IDE so you perform the job but it's a bit more than just like a small chat interface within. So if you are working with agents on a daily basis then uh you will definitely try air and I would say you might like it a lot compared to what you currently have.

[00:16:48]>> For someone who's maybe not fully into these agentic workflows in how far does air offer functionality that we don't get when working with an agent like Gemini agent directly in an IDE like Android Studio >> I mean air is not trying to rebuild the whole um agent harness. So what we do actually is um we provide you um rich interface and environment where you can work with agents you currently work with. Um what changes is the way how we help you to perform this task. So just remember we're trying to build uh products which make developers more productive and make it more enjoyable.

[00:17:23]So uh when you work with agents uh especially in terminal at some point you might bump into some frustrations with the terminal UX and stuff like that.

[00:17:33]with air. For example, I will give you an example of a workflow. If you want to perform the task within the framework of plan, implement, review, then air uh makes it very native for both the developer and the agent. Uh for example, you start planning with Gemini and then uh you come up with a plan, you implement that let's say with the cloud agent, it's all connected within the user interface. So it it feels like you're going step by step with that and it's all configurable. So you don't need to sweat around like can I do that or can I not do that. It's it's just a way we try to build the proper habits and like a proper framework so you are not overly sweating around how should I do that properly. We just guide you through the process and we make it very native.

[00:18:19]When the agent performs something you have the full review you can get into the code. You can get uh and navigate like what kind of implications that might have. Then you may ask the agent to review and it has like the native comment interface. So agent practically gives the comments back to the same change and then just revert it to the session of the agent implementing. So it feels like a natural way of doing things instead of trying to build all this tooling for yourself mixing with multiple agents or within the uh the terminal you currently are.

[00:18:51]>> Thank you so much for your time.

[00:18:52]>> Thanks a lot for having me. Yeah.

[00:18:54]Cheers. Nice to meet you.

[00:19:03]Uh, I watch his videos. They are a good teacher and good. They really appreciate what what he created for the community and if you have the chance to take his his courses, there are a ton of valuable content in what he create. Thank you.

[00:19:22]>> Yeah, like the compos master class.

[00:19:24]>> I think it was the compos. It was part of the master class. Yeah.

[00:19:27]>> Yeah.

[00:19:28]>> How did you like it?

[00:19:29]>> Uh it was it was good.

[00:19:41]>> So I'm here with >> uh Emil Fl. I'm a product manager for Cotlin multiplatform. Uh and together with Anton I work on Cotlin multiplatform at Jet Brains. Hi folks.

[00:19:49]Uh I'm Anton Mave working together with me on Cotlin multiplatform as a project lead. What are the main areas of Kartland multiplatform that you want to focus on most in the next time? So be it tooling, build system, CMP, etc. Oh, that's actually great. That's actually great question. Like we spent uh last couple of months uh like coming up with some ideas of what we should focus on and one of announcement what we made uh on the keynote is cotin tool chain and cotin tool chain is basically uh there are two kind of uh groups of problems that people are experiencing with technology. Uh first like when you starting with Kimpy like when you start uh like where you install uh Android Studio install Xcode you need to learn about swift you need to learn about something cocoa pot a lot of to to learn and you need to start a lot of tool and we want to focus on the like great on boarding experience great setup from the start and this is what cotling tool chain is about. So this is the things that we kind of export with amper as a our tool for on boarding for starting new green field projects. So we take one of the direction what we focusing will be focusing is basically great experience for newcomers for new projects and this is cotental chain is about and second uh is uh existing project and I will give you a chance to cover that.

[00:21:20]>> Sure. So if we look at where a lot of our users are secretly cotlin multiplatform is all the parts that cotlin android are are not so we have to figure out the parts that make it into swift make it into the iOS ecosystem maybe make it into the web ecosystem and that's really where we spend a lot of our time like finding finding those other parts um and a lot of our existing users are using the combination of Android and iOS uh and iOS development is something where we have continually been improving a lot we uh since last year have the cult multi platform plugin which allows you to do swift development in Intelligj. Some people might get flashbacks to app code. Uh but yeah, it's basically making that iOS development experience really nice.

[00:22:00]Thinking of the debugger, compilation speeds, that's really where um existing projects especially they become bigger, those things become more and more important and I feel like that's aside from the newcomers, that's where the existing projects and maybe what we call brownfield, so existing projects adopting cotlin multiplatform really benefit from a lot. Um so yeah I think I think that's where a lot of our investment is and also where we have a lot of really smart people uh solving things like thing like swift export or SPM import is is is all in that category of helping um those kind of projects >> regarding this whole ecosystem around cotlin k& are there any plans that you can share that haven't been shared much yet in the public what is going to come in the next 6 to 12 months >> actually one piece of ecosystem is uh the kimpy libraries and libraries In general uh we have shared that the number of libraries is growing. It's like it's not twice the growth uh in a year but still that's like almost twice as many libraries today. But there are still multiple use cases coming when uh some KP libraries are missing. And this is uh like our uh our plan is uh to see how we can uh improve the situation with libraries to fill the gaps uh to uh work with the library developers. Um if there's some for instance some library which missing some support on some uh library ecosystem this is uh uh we would very much like to see how we can uh help it uh to move forward the ecosystem to have all the key uh use cases covered with kimp libraries.

[00:23:43]That's one of a direction which uh yeah we haven't announced but this is what we are working uh on actively.

[00:23:51]we'd like to add some.

[00:23:53]>> So, um, one area that is on complexity for many users, but especially when you come from Android and you have to publish your iOS app and you have to sign your app for the first time, those kind of tasks are hard and uh I think we could do a better job with removing those kind of steps and those kind of frictions for for our users. So um I see it's something where I see a lot of uh potential for us to make our ecosystem like basically end to end a nice experience. You start with a cotline tool chain and you might end with a golden tool chain as well but uh no promises just knowing that there's pains there and we can do a lot better.

[00:24:31]>> Yeah. Do you regarding pains is there like the the biggest pain for a solved problem from the previous two or three years? So the biggest the biggest unexpected challenge that happened to you to something that you've now they solved >> actually yeah the challenge is on boarding uh getting people started and this again to cotton tool chain is our big uh big stakehold we have already proven that and like I personally very much involved in this tool chain project and emper design and I do believe it's the big uh big milestone for us.

[00:25:11]This is alpha already and we will uh move it forward with things like hey basically you start with the milson you start with common line you install cotlin you create your project you install couple of skills and then you go to the app store I would say something I take for granted because I actually only started a year ago um but the fact that you can confidently write swift code in intelligj for your project it's pretty pretty insane so the fact that we have this uh well-rounded solution for a lot of the multiplatform needs for for iOS development and also for for JavaScript and TypeScript we already pretty good because we have other tooling in that environment. Um but debuggers in their run configurations I I only touch Xcode when I have to publish and sign that's kind of the place that is left um and I think uh not to bash on Xcode or anything in that regard but it is another tool another environment you need to get familiar with and um I think we like to avoid that. So we like to keep people in one environment and um if that is Xcode, you know, use the Cotlin tool chain from the CLI and work from there like that's really be able to let engineers just spend time on building their product instead of the tooling and learning the things around it. Um so yeah, I think a year ago this was a question of how this would even work and now these kind of things exist which is pretty crazy like you can just command click and you go to the right functions.

[00:26:32]It's uh that's pretty cool. Actually I recalled another another thing that's uh kind of uh very pressure accumulated and we weren't sure if we can solve it how much it uh time will take but they actually the team who was involved with they were did brilliant job is uh swift package manager support. Yeah. Yeah.

[00:26:50]Like like because cocoa pots is already obsolate and the more and more libraries have been published with uh package Smith packages and we were like we were hesitant and we were sure but folks did great job. So folks could us who made it happen.

[00:27:08]>> Awesome. Thank you so much for your time.

[00:27:10]>> Yeah.

[00:27:11]>> Thank you.

[00:27:12]>> All right that was it. long day is over, but it's always a great experience being here on Cofflin Cough because it really shows you that there are real people behind all those numbers that I just see in my YouTube dashboard. But, uh, understanding that there is really a real person behind every single number of all those hundreds of thousands of views, that that's crazy getting all that positive feedback. Thank you to that if you've approached me here. Thank you if you leave positive comments. U, thank you for Jet Brains for the invitation. And I'm of course curious what kinds of technologies that have been shown here that I've uh had interviews about you are the most interested about and actually also hearing on my channel here in the next time. Thanks for watching. See you back in the next video. Bye-bye.