The journey
One entry per turning point. Three phases:
| Phase | Span | What it was |
|---|---|---|
| work-ai | 2025-05 → 2025-12 | Pairing with an AI on SRE work. Cursor, then a personal Pro licence. |
| enablement | 2025-12 → 2026-03 | Claude Code becomes the daily partner. Skills, docs and a video to bring the team along. Evenings go public on GitHub and agent-zone.ai. |
| personal-lab | 2026-03 → | The organisation licenses Anthropic; my own subscription turns toward my own curiosities. The lab begins. |
The third phase is where the projects live.
- Cursor at workwork-ai
First AI pairing partner, on real SRE work. Within weeks I was paying for my own Pro licence.
- Claude Codeenablement
Introduced to Claude Code. Within two weeks I had downgraded Cursor and bought the top Claude tier. Never looked back.
- Teaching the teamenablement
Skills, documents, one-on-ones, group sessions, and an anime explainer on YouTube to lower the entry bar.
- First public reposenablement
Evenings and weekends go public on GitHub. A local-model benchmark, Temporal patterns in Go, and a knowledge site for agents.
- The lab beginspersonal-lab
The organisation licenses Anthropic. My own subscription turns toward my own curiosities. This is where the story really starts.
- Losing $HOMEpersonal-lab
A local process renamed a parent of my home directory mid-session. Everything not in git or a backup was gone. The lab grew up that week.
- The lab grows uppersonal-lab
v5 lands, the fleet gets a worker library and a weekly fire drill, LiLM's factory is built in a week, and this site gets a hostile review before it goes live.