About ai-learning-bits

ai-learning-bits is a professional blog about the tools that now sit between developers and their code: agentic coding assistants, the models behind them, and the architecture that makes any of it hold up in production.

It is deliberately opinionated and deliberately practical. No model-of-the-week hype, no leaderboards mistaken for truth. The goal is the working engineer's question: given this task, which tool, and how do I drive it without getting burned?

What's here

• Tool deep-dives — Claude Code, GitHub Copilot, Codex, and Gemini, each judged on what it's actually good at, not its demo.
• Architecture — how to build agents that survive real users: context, tools, memory, evals.
• Local models — running capable coders on your own hardware when the cloud isn't an option.
• The benchmark — a sortable, honestly-caveated comparison of frontier and open coding models.

Who writes it

Jakub Jirák — developer, plugin author, and a heavy daily user of every tool reviewed here. The takes come from shipping with them, not from press releases.

If a post can't change how you'd actually pick or drive a tool tomorrow, it shouldn't exist.