Skip to content
PodSubs
ShowsPeopleGuestsJobsNewsEvents
More
BlogPodCred
Menu
ShowsPeopleGuestsJobsNewsEventsBlogPodCredNetworksSign in
Join nowSign in
  • Shows
  • People
  • Jobs
  • News
  • More
    • Guests
    • Events
    • Blog
    • PodCred
    • Networks
    Join nowSign in

Code & Curiosity by Maximilian Schwarzmüller

Code & Curiosity by Maximilian Schwarzmüller artwork

2025 - My Look Back

Dec 23, 202527 minAudio
  • Apple Podcasts

Another year's over. You know what that means! Time for a look back. MY look back in this case. And it's been quite a year. In good and bad ways.

Audio and video play straight from the show's own hosting. PodSubs stores and serves no media.

Transcript

From the show. Tap a time to play from there.

So another year's almost over and you know what that means, time for me to look back at it, and yeah, obviously that video will be about AI. Not just about AI, but a lot of AI because, you know, it's the thing that influences everything and has an impact on everything, And believe me, I'm not a fan of having, uh, AI everywhere. But it is what it is. But yeah, let's take a step back. You, you might remember this year was meant to be the year of

AI agents. It was the term you heard and read everywhere, especially at the beginning of the year. You still, of course, hear and read it everywhere, year started, everywhere on any- every website you could read, "2025 is the year of AI agents." And of course, an AI agent was basically anything that used AI for anything. You wrote a script that uses AI to summarize a PDF? Congratulations, you built an AI agent. Now, of course AI has made

significant advances to some degree when it comes to agentic workflows. Now, what is an agentic workflow? Again, depending on how you define it, it's anything that uses AI, but I would say it is any kind of workflow where AI makes decisions. And keep in mind, we're still talking about token generators, but of course those tokens can describe tools the AI wants to use. And one big category of agents that you definitely know is of course all these coding assistants.

I mean, they did not start this year, you know that as well as I do. But in this year, we saw lots of development on Cursor, we saw the release of Claude Code, and all these tools are of course about giving us developers assistants that help us write code, that can generate code based on our instructions or, if you're not that much into writing code, they can of course generate code based on pretty vague instructions and you can enter the

territory of live coding. And live coding definitely was another big theme and thing this year. The term was coined by Andrej Karpathy in 2025 and as he described it, the idea really was that you don't really read the code, that you go with the vibes, you just basically see if it works or it doesn't and you just tell the AI what to change or make it work and... Well, that is how you can build software.

Now, to be fair and very clear here, he made it very clear that he still also writes code himself. But of course, you could use live coding for pieces of software that are maybe just a one-time thing you need, maybe just a utility script you need, or maybe if you're exploring some idea, building a prototype, you could definitely try live coding for that. Now, I have tried my fair share of live coding and I will say, results have been rather mixed

and it's definitely not my thing. Wasn't my thing, still isn't. I will say though that of course it can be quite helpful to let AI build a feature where you maybe don't really know how to tackle it because you need to use some browser APIs, for example, which you don't know yet, help of the AI, which is not the same as just letting the AI build something and then use it and move on if it seems to work. That's a different thing. But yeah, live coding was a huge thing and still is a huge thing.

We have platforms like Lovable, v0, and many others which are, in the end, live coding platforms. Yeah. It's here, it will probably stay here. We saw lots of people experiment with it, we definitely have live coders out there that do live coding for a living, more or less I guess. But yeah, let me be very clear here. Live coding is not a good idea. Being an expert, knowing how to write code, and

using that expertise and that knowledge in combination with AI is so much better than live coding. That's... I can only recommend taking this path because your knowledge plus AI can get you really good results. Just relying on AI, probably not so much. Of course, coding assistants like Cursor or Claude Code are not the only agentic systems we saw develop this year. OpenAI, for example, also integrated an agent into ChatGPT that can do stuff

for you, that can click around on websites and do work for you. Now, maybe it's just my bubble, but it doesn't really seem like a feature that everybody's using all the time. But yeah, it's here, and combined with deep research agents, which I think we also had last year which definitely became more of a thing this year, you can offload some tasks, especially some research tasks, to AI. Though as everything and always, it's

typically not a super good idea to just take those AI results and use them as they are, but instead it's much better to use them as a starting point and build up on them with your own knowledge, maybe your own additional research, and then AI, in my experience, can save you some time in certain situations. And I definitely, for example, like using-... Gemini Deep Research for doing research for a video like this. I still bring in my own expertise. There are many points I

cover in this episode which were not brought up by Gemini, and also not by ChatGPT, but yeah, it's a nice additional help. From that angle, we could call this year the year of agents, but to be very honest, not in the sense all these newspapers meant at the beginning of the year. It doesn't really seem like all those companies out there are using agents for everything. Quite the opposite. There recently was an article about, uh, Copilot by Microsoft not

really delivering on its promise or not being as popular as Microsoft would hope. It really seems to be the case that companies want to use AI, they all feel like they have to, of course, and therefore, they are really searching for use cases for AI and it's hard to find them. Coding and, um, AI-assisted coding is definitely one of them, but many others might not be there yet and we don't have all these agents that do all the work humans used to

do. And by the way, when we talk about coding, I am well aware that in many companies out there, there are executives that basically tell their teams, "You either get double the work done more, or we'll have to let some people go use AI." And that, of course, is a horrible situation because whilst AI can be a very good assistant for coding, and I've said that more than once, I say it in almost a- all my videos, whilst it can be a good assistant,

it's not a good replacement. And yeah, it can make you more productive, but twice or three times the output? I'm not so sure about that. But yeah, it is what it is. That is something which will probably get better as companies realize that even when it comes to coding, AI is not the magic bullet, but just another helpful tool. Or, of course, all these AI models get much, much better, but until that point, it's a

really, really bad situations. But yeah, agents everywhere? I'm not so sure I don't see them. Now all that, of course, does not mean that AI is not getting better and that there is no chance of more agents doing useful stuff in the future. But when we look at this year, we definitely saw the release of many, many AI models. (laughs) It, it feels like a new model was released basically every week, right? Especially now in November, we got like Gemini

3, we got new models by OpenAI, by xAI. We get so many new models, and not just in November. It's really model inflation out there. If you're a web developer like I am, it feels like the JavaScript framework wars we had in 2019 around that time. Uh, but even worse because we didn't have as many frameworks as we have models now. Obviously, that will only continue, but the question is will those models get better and better? To some extent, to me it feels like they're not really

I know they are continuously getting better in all these benchmarks, but we've seen more than once this year new model is released, does very well in a benchmark, and then kind of falls flat in the popular opinion. Isn't really living up to its promise and people are not really, uh, happy about the model. We've seen that more than once. One prominent example would be LLaMA 4, which was released by Meta, uh, in, in the first half of the

year. LLaMA 4 did amazing in all these course, especially interesting because it's an But people didn't really like it. It really felt like it was fine-tuned for these benchmarks as it often feels these days with model releases and, well, I'm not sure if anybody's still using LLaMA 4. Uh, the result, of course (laughs) , of this, uh, failure, if you wanna call it like this, was that Mark Zuckerberg started hiring

all the talents in the AI space by paying astronomic amounts of money for the top talent and we now have all that talent war going on between OpenAI and Meta and Google and bunch of other companies. But yeah, regarding the raw performance of the we have yet to see if we'll see a steep increase in the next year, 2026, or if the curve kind of flattens out. Benchmarks alone certainly do not tell

the full story. Now, of course, there's a lot of money being thrown on the development of all these models and therefore, the year actually started with a shock when DeepSeek R1 was released because that was some Chinese underground model which nobody expected, which at the time seemed to be trained very cheaply and seemed to be able to compete with top-tier models back then. And of course now, at the end of the year, we can say that all these Chinese models, not just the DeepSeek models

related to video or image generation, play an important role in the AI landscape and China, despite all the problems they should in theory have with getting graphic cards, GPUs, plays a super important role and will be an important competitor in the year to come. And on the less, as mentioned, we'll have to see if AI models improve like this or maybe a bit flatter like this. Of course, large language models aren't everything,

text generation isn't everything. We saw pretty amazing improvements this year when it comes to image and video generation. I mean, if you compare the kind of videos you can View 3 by Google or of Sora 2 by OpenAI with what was possible a year ago, that's mind-blowing. It's so much better, so much, uh, better quality, higher character consistency, audio is better, and that's just the US models. If you take a look at some Chinese models

like the Kling models-Um, you can see some amazing development and results there too. And it's not just the case for video. Image generation was revolutionized first by OpenAI's GPT image model, but then that model now looks quite old compared to Google's NanoBanana, and now its NanoBanana Pro model, which excels at generating all kinds of images, including, uh, images that include a lot of text like this beautiful infographic

here, which was created with, uh, my site, buildmygraphic.com, which gives you a nice feature-rich editor and beautiful style templates to build graphics like this. So there is lots of development in the AI space when it comes to other modalities other than text, and of course, I'm also convinced that we'll see more useful tools being developed on top of AI models. Even if those AI models shouldn't become much smarter, we will definitely see better and better tools

models and therefore deliver more value. Just like Cursor is not just some AI model being used in its raw form, but instead an AI model that is heavily adjusted by engineering the context in a clever way, having good system prompts and all that stuff that leads to better results than if you were just to use the raw model through the OpenAI API or whichever provider you are using. So that was a

lot of talk about AI. I'm aware of that, as I mentioned. I'm sorry. But it is obviously an important part of we're living in if we like it or not. And believe me, I would be very happy if we were not living in a world that's full of AI slop and all that AI hype. But it is what it is. And of course, I'm fully aware that this also was and is another very difficult year for software developers, especially for web developers, especially also for junior

developers because we're coming still from that all-time high during the pandemic where all the companies couldn't get enough developers. Well, turns out they had too many. So you all probably know that chart about the software developer jobs. And, well, at least that seems to finally have reached the bottom, but now of course there is AI. And as I mentioned before, there are these discussions or the pressure inside of companies about

using AI instead of developers for certain tasks. And especially junior developers, therefore, are not in high demand because, hey, AI can do it, right? Wrong, totally wrong. And who's going to be the senior of tomorrow if there is no junior being hired today? But yeah, it is what it is, and therefore, unfortunately, this was another difficult year, and next year is probably going to be diff- difficult too, which is really bad. There is no other way of putting that. Now despite AI, thankfully,

there are also some other interesting topics to talk about when looking back at this year. For example, we, of course, saw new versions being released for Next.js, React, Angular, and all those other frameworks and library stats are important for us developers. It's also fair to say that it looks like React and TypeScript and Next.js won the AI race. I mean, take a look at the React downloads. AI clearly has a favorite.

And I've talked about this in other episodes before, this is a huge problem in my opinion because no matter if you like, um, the fact that we had times where a new framework seemed to come out every day or not, it's definitely not good to have one clear winning stack, which is React, TypeScript, Next.js, ChatGPT, and Tailwind. That's it. That is definitely not good, and I don't see that changing, unfortunately. And that is not to say that any of these technologies would be

They're not. They're great. Competition and innovation just also is good, and it looks like that's going to get harder with AI. But hey, okay, th- there was AI again. Let's, let's move away from AI. So yeah, we saw these new versions bringing us new features, but especially React and to some degree Next.js, but it was React, had some problems, some security problems. It was just a few weeks ago that a couple of vulnerabilities were reported related to React

and React server components. I did create a separate episode about that, so take a look at that if you wanna learn more. That, unfortunately, wasn't the only security issue related to web development we saw this year. Instead, we also had multiple NPM-related attacks, node modules-related attacks, where some packages were infiltrated with malicious code in a way that that code would spread automatically to other packages and

ultimately potentially reach your machine to then steal crypto or crypto credentials and stuff like that from your machine. And of course, that all is probably just the beginning. I mean, vulnerabilities and attacks have always been a thing, but especially with AI, I see a huge danger of that trend accelerating for two main reasons. For one, AI can, of course, help with developing

malicious code. It can help run such attacks. It can help with social engineering. I mean, you can create realistic voices days. So that will be a huge problem going think.But of course, the other problem potentially is that AI can also lead to gaping security holes in software. And just to be clear, the React vulnerability and the NPM attacks are very

likely not linked to AI in any way. I still see these problems coming up in the future for the two reasons I just described. And we saw with the React attacks and with the NPM attacks how vulnerable this web development, this front-end web development ecosystem really is. And especially in a world where more and more applications are powered by React, well, it's pretty clear that that is a quite interesting

target for an attack and for looking for vulnerabilities. And the same, of course, is true for NPM and Node modules because even we experienced developers often just install packages. And of course, we trust many of those established Why wouldn't we? So even we are very vulnerable to these attacks. Now take it a step further and think about people that don't know anything about coding too much about it and just live code their way through the

world. Yeah, these are quite interesting targets, I'd say. So yeah, that was a thing, that will be a thing, I think. Speaking of problems and attacks, not really an attack but still kind of worrying was that we had two big outages this fall, AWS and Cloudflare. AWS took significant chunks of the internet down this October for multiple hours, almost a day overall.

Cloudflare kind of did the same on a slightly smaller scale. Again, is AI a problem here? Were those engineers using AI for these configurations that blew, uh, things up? I don't know. Probably not, but we don't know. But it's kind of worrying how vulnerable the overall internet is and how dependent it is on a few big providers. And we knew that before. It's not new and it's not the first AWS outage.

But these two big incidents which happened only a few weeks apart, yeah, they really show us what the internet has become. It's been that place of decentralization and it kind of still is, but on the other hand, it kind of still isn't. If AWS, if one big region of AWS gets taken out, boom. Here go 30% of all websites as it seems. So yeah, that is

definitely also something we saw this year, that happened this year implications in one way or the other. And that is kind of, well, showing us what the internet has become, I guess. Speaking of what the internet has become, and here I have to go back to AI briefly. I apologize. Significant portions of all traffic on the internet and of all content on the internet are already generated by AI. So we got bots reading content generated by other bots generated by AI.

Great new world, I guess. So that, of course, is also something that accelerated this year. I mean, just two or three years ago, we had almost no AI-generated content on the internet because it sucked. But now it's, it's everywhere. And guess what? That will also only accelerate. I mean, if you're on the same bubble as I am on X, everybody's just building AI slop there. Great. That is, uh, a lovely new future.

And don't get me wrong, you can generate very useful stuff with AI. I love that I'm able to generate images or icons with AI. Of course, I am aware that we also have some problems here if we think about copyright and so on. But it is useful, I will admit that. And the same is true for some text, being able to summarize a PDF document. There are many use cases where AI is very useful. I wanna be very clear about that. But the AI slop and the internet becoming more of an AI

machine full of bots and AI-generated content, yeah, I've said it before, it's, it is what it is. I'm not super pumped about it though, to put it like this. Now, I could go on for much longer. I could talk about TanStack, which has a little wave of popularity, at least in my bubble now at the end of I could talk about the TypeScript port to Go instead of Rust, which was announced at the beginning of the year

soon going to be complete. And you can already try the Go version of TypeScript, it looks like. It's awesome. It's much faster. I could talk about the, the trend back to monoliths and VPS instead of microservices or managed services. And all these are just small trends, small building blocks, of course. In large enterprises, we're still working with microservices probably for good reasons. And as I mentioned before, React and Next is still king. But-These are all nice little things which

still make this overall web development world really nice to be in, to put it like this. I'm super excited about all these things. I'm excited about the fact that we got BetterAuth, an amazing authentication library which became super popular this year, which I'm using in all my projects, which makes authentication such a breeze. I enjoy working with Bun and Hono. I really love what you can do with PostgreSQL, that you can use extensions like pgvector to build a vector

database and build RAG applications, because again, there are useful things you can do with AI. I also very much like the fact that you can use Cursor and Cloud Kode or OpenCode or all these other AI coding assistants to be more productive and to outsource some tasks to AI. I'm absolutely loving the fact that we got a bunch of amazing new features built into browsers, like the popover API and many, many other cool APIs, and also CSS properties and values that you can

use today in all major browsers to make your life easier. The problem there just is that AI probably doesn't know about these features yet, and maybe never if it just is that machine that builds React and Next applications, and then learns based on what it built But we'll see how that turns out. I just wanna say, web development is, in theory, in great shape. We got all these amazing features that we can use in browsers today.

Many things are much easier to build than in the past. We got mature frameworks. We got innovation through frameworks like TanStack Start or the BetterAuth library. We got so many amazing stuff going on can be such a useful tool. I really, really hope that we'll be able to make good stuff out of all these building blocks, that we'll be able to build up on this great foundation in the future, and that we'll not get consumed by all the

AI slop and all the BS around AI, and all the problems that come from AI and companies trying to force AI on us in every aspect of our life. I really hope that executives will learn that AI is a great assistant, but not a great replacement for developers, and that we can leverage the potential of all these amazing things we got today, because I truly believe that this is the

best time ever to be a developer and a web developer. But it's also one of the most difficult times, unfortunately. So yeah, that was my look back. I might have missed some things you found important too. Please let me know. Please let me know. I try to be comprehensive, and this video is already super long. I will also create a separate video where I take a look at the next year, and I create a bunch of videos where I share opinions and thoughts about everything all the time. So yeah, if you like that, follow me and

have a great end of the year and a great start into the new year.

Open the transcript file

Hosts and Guests

Nobody is credited on this episode yet.

Credit the people on this episode

Be the first to rate this episode

More episodes

All episodes
  1. We will have to accept AI-generated content

    Oct 15, 202612 minAudio

    I think we'll have to get used to (and accept) AI generated content. Not to slop though.

    • Apple Podcasts

    Credit the people on this episode

  2. Is software dead?

    Oct 13, 202612 minAudio

    There's a lot of talk about software being dead. Unsurprisingly, it's complicated.

    • Apple Podcasts

    Credit the people on this episode

  3. GPT 6 Astra - Good Beyond 3D?

    Sep 9, 20268 minAudio

    GPT 6 Astra was released. It's a good model, I have to say. Especially for 3D work it seems to be a real game changer. But due to all the hype, you may be disappointed...

    • Apple Podcasts

    Credit the people on this episode

  4. You may ignore Omarchy but not its impact

    Aug 26, 20268 minAudio

    Omarchy got a lot of hype and visibility recently. But even if you don't care or don't use it, you shouldn't ignore it.

    • Apple Podcasts

    Credit the people on this episode

  5. SvelteKit 3, Solid 2 - Do they matter anymore?

    Aug 18, 202614 minAudio

    We're not going back to the JS framework wars. BUT new frameworks and framework versions are important. For a different reason now!

    • Apple Podcasts

    Credit the people on this episode