Work with me
Review and ways of working became the bottleneck.
I work with engineering teams adopting Claude Code, Cursor, or Codex.
Tell me about the teamI ran 15 parallel agents. Review took longer than the building, and one merge blocked the others.
I read my own AI coding sessions, deleted the setup, and rebuilt it from what the sessions showed. Coworkers adopted that setup.
I wrote that up in Claude Code worktrees.
This is what worked for me, and why I offer it to a team.
Audit
I instrument the team's AI coding tools with telemetry. People opt in, and the data is anonymised. I read everyone's sessions. You get shared team skills, a note for each engineer, and the two or three changes that will move the team most. About a month later I measure those same things again.
Workshop
We scope work for parallel agents. We settle naming, pseudocode, and data decisions before implementation, and we keep review from becoming the bottleneck.
Advisory
A regular call with an engineering lead who is rolling out AI coding.
Get in touch for scope and pricing.
You show me how the team uses the agents.
We do the one we picked on the call.
We look at what changed.
Engineering teams adopting Claude Code, Cursor, or Codex, and the lead rolling that out.
Claude Code, Cursor, and Codex.
Telemetry is opt-in and anonymised. I use it to see how the team works, then I hand the findings back. The sessions stay with the team.
I live in Sydney, Australia. The call, the audit, and advisory are remote. A workshop can be remote, and it can be in person when the trip is worth it.
Get in touch for scope and pricing.