// engineering notes on AI systems & automation

Building systems
that run in loops.

I'm Petr. I build AI agents, automation pipelines and the infrastructure that keeps them running while I sleep. This is the engineering log: what I built, what broke, and what I'd do differently.

loop_007 · [agents] · · 11 min read

Replacing n8n with Claude Code and a CLAUDE.md file

Our long-form content pipeline kept dying on API timeouts. So I threw out the workflow builder and gave an agent persistent memory instead. Here's the architecture, the failure modes, and the numbers after 30 days.

read →
loop_006 One prompt, 14 landing pages: local content that isn't spam
[automation]
loop_005 GitHub Actions as a poor man's agent orchestrator
[infra]
loop_004 The 90% automation target: what actually resisted
[meta]
loop_003 Branded PDF reports from raw analytics, no humans involved
[agents]
loop_002 Lead scoring with a registry API and one cron job
[automation]
→ all loops

SUBSCRIBE_

One email a month. What I built, what broke, what I learned. No roundups, no reposts.