About
About me
I'm Mick Brzeziński — fifteen years of shipping software, the last few focused on LLMs and agents.
My take: the model was never the hard part. The real work is the engineering around it — the harness, the evals, the guardrails that make it hold up on real data.
By day: production agents and pipelines. After hours: my daughter's record label, an 1892 carol rebuilt for my wife, a phone that edits our home videos overnight.
Now
What I'm working on
-
Benchmarking four agent harnesses on one spec — same model, measured to the token.
-
Model evals that run on a Claude subscription instead of a metered API.
-
New releases for my daughter's record label.
Projects
Things I've built
My two-year-old's record label
~50 disco polo compilations, an animated series and an Android app — generated from one text file, at about $0.10 a song.
The World Cup predictor
An agent that called both World Cup openers to the exact score — judgment split from formatting, EV optimised, a human on the trigger.
Lead Scout
A $0 lead-research agent: scouts ~100 companies a week from public sources, scores fit, drafts the opener — and never sends without me.
The 2,520-call eval rig
A model-eval harness that runs on my Claude subscription instead of a metered API — golden set, deterministic grader, cost per call to the token.
Writing
Latest from the blog
-
Your AI built you a startup. It also built you a time bomb.
Vibe coding gets you to launch in an afternoon. The 30% it skips — security, access control, the parts that don't show in a demo — is exactly what breaks around Day 60. The data, the documented incidents, and a 5-minute gut check for any AI-built app.
-
I replaced the server with a prompt: a self-running AI news digest for $0
A developer-grade daily AI digest — crawl ~28 sources, let Claude edit, publish a static site — with no server to operate. The cron daemon is a scheduled routine, the database is a git branch, the host is a free CDN, and the editorial runs on a subscription rail at $0 metered API. The full teardown.
-
Your codebase could document itself — here's who's building that, and how
A wave of tools promises docs that keep themselves current — the repo as its own wiki. Here's the map: the main players (DeepWiki, Google Code Wiki, Driver, DeepDocs, Swimm, Mintlify) and the four distinct approaches they take, what each trades off, and the questions to ask before you trust any of them. No ranking — the space moves too fast for that.
Contact
Say hi
Want to talk agents, projects, or work? Drop me a note — I'll write back.