Peon post-mortem

document
POST-MORTEM #001 — the last one
service
peon.dev
severity
SEV-0 / permanent
status
RESOLVED — by shutting down
uptime
2021-07 → 2026-10

01 tl;dr

Peon was a great tool built on a great idea. It worked really well — before AI. Today it no longer pulls its weight, and I don't have the energy left to rebuild it for the AI era. So I'm shutting peon.dev down. For good.

02 what peon was

A virtual colleague for developers. You told it what to do. It did the task on a schedule, opened a merge request, and you reviewed it, merged it, and went back to the work that mattered.

It started out as PHPMate, grew into Peon, and was 100% open source all along.

03 timeline

  1. INFOPHPMate is born. Mission: give developers their time back.
  2. OKCarvago hands Rector over to the bot. ~30 min saved per developer, every day.
  3. OKRenamed to Peon. Recipes, custom tasks, open source.
  4. WARNAI coding assistants arrive. The bar for "automation" moves overnight.
  5. WARNAgents now open their own merge requests. Peon's unique edge gets thin.
  6. FATALMaintainer energy budget exhausted. Graceful shutdown started.

04 what went well

05 root cause

No single bug. The world just changed. Peon was built for a time when automation meant deterministic scripts on a schedule. Now an AI agent can read a codebase, reason about it, and fix things no script ever could. Next to that, Peon lacks the utility to be worth running.

Bringing it into the AI era would not be a patch. It would be a rewrite. With other, bigger projects to run, I can't give it the energy it would need. Better to end it honestly than to let it rot quietly.

contributing factors: AI ate the use case · finite energy · bigger projects

06 action items

07 thank you

To everyone who used Peon, filed an issue, sent feedback, merged one of its merge requests, or just believed in the idea — thank you.

It was a good run, and I'm proud of what it did. ♥

// if you want to get in touch, reach me at
j.mikes@me.com

— Jan Mikeš