All episodes

What a week in AI? Well, okay, I guess every week is "packed with announcements".
But with Google IO, tons of projects and the Claude 4 model releases, there WAS a lot going on. Time to unpack!

There's a lot of hype around LLMs & generative AI - especially when it comes to "AI Agents".
I don't have a super agent (or an army of agents) that does work for me.
But I got a couple of use-cases that really help me get more done!

LLMs often default to a single tech stack (like React, Tailwind, ShadCN). This tendency could reduce competition and innovation in the long run, potentially leading to a development monoculture.

I've slept for too long on open LLMs like Gemma, Llama, Qwen, DeepSeek etc.
They're AMAZING!
Using them is simple (with LM Studio or Ollama) and they're not far behind the "big models" you can access via ChatGPT etc. At least in my experience.
I'm using them for all kinds of tasks...

React Compiler RC (Release Candidate) was released!
Almost one year after its initial release (back then in experimental status), it's now almost ready for production use.
And it can save you LOTS OF time of writing useMemo(), useCallback() and memo()!

Google released a couple of AI Agent related products & services!
The A2A protocol is maybe the most interesting initiative. Also because it kind of clashes with MCP. Or does it?

Microsoft & VS Code are striking back against Cursor & Windsurf. And not just by making GitHub Copilot better...

It's essentially ONE key feature I really don't want to miss anymore.
Well, maybe 1.5 features...

MCP has been a hot buzzword. But what exactly is that model context protocol thing about? And who do LLMs actually use tools?

Obviously, parts of the web dev community are upset: Why didn't they choose Rust?
Well, turns out they had good reasons!