devblog · written while building

On programming, AI, automation and what surprises me.

Programmer. Architect. I have been making software since the tail end of the demoscene, and using AI in real work since 2024. I am building Refio, a local AI agent for IntelliJ, because I want to understand in practice how it all works. This is where I write down what I learn.

Deterministic where you can, model where you must

Archive 8 posts
23
Jul 2026

Why I still score my own benchmark, even though two LLM judges do it for me

Last time I wrote about my benchmark (the code that kills your browser), I skipped the boring part. Sixteen models, six attempts each, one task – that is 96 HTML files somebo…

AI · Programming
10 min
05
Jul 2026

Let Claude Code write, and Codex check it

After Anthropic’s recent moves – higher prices, hard-trimmed limits – a question I keep hearing more and more comes back: “Codex or Claude Code?”. And…

AI · Programming
10 min
21
Jun 2026

One small trick and you stop losing context in local LLM models

Recently I was messing around with the Refio plugin and testing it on local LLM models. In theory everything worked: the model was supposed to get the whole project context –…

AI
6 min
04
Jun 2026

Code that kills the browser – observations from a small LLM benchmark

There’s a moment when the model returns code, the file opens, and for two seconds you think it works. Sometimes it actually does. Other times the browser tab goes blank, or y…

AI
6 min
10
May 2026

Quantization for the impatient – why int4 doesn’t speed up image generation on GPU

Long time no post… on image generation. Recently I sat down with an old codebase and put together a quantization pipeline for diffusion models out of curiosity – SDXL, …

AI
9 min
04
May 2026

Every LLM writes the same Todo App. And I mean that literally

I’m building a benchmark of local models. Simple task: build a Todo App in a single HTML file – add, delete, mark as done, filters, localStorage. Trivial. Each model ge…

AI
6 min
02
Apr 2026

How Refio manages conversation context, or how to fit a big project into limited AI memory

How Refio manages conversation context, or how to fit a big project into limited AI memory AI models can do a lot these days. They can analyze code, write documentation, hunt for b…

Uncategorized
10 min
28
Dec 2025

Refio – a local AI assistant for IntelliJ and 6 RAG techniques worth knowing

Genuinely practical RAG? It’s not just about building a vector index and hoping the LLM will find everything on its own. When you’re working on a project, you’re …

Uncategorized
7 min