Skip to content
Petar Gusić
Field notes2026

Writing

Notes on delivery systems, org design, and practical AI adoption in growing engineering teams — from the operator seat, not the analyst seat.

Now publishing

Series 2October 2026

What happened when product managers drove the agent

A four-week pilot where product managers wrote their own A/B test code with an AI agent, and a developer reviewed every change. Six posts, with the numbers — published as they land.

47experiments filed
38dev-approved
15 minmedian review
4weeks
Read the series →
Latest post02 / 06
An AI coding agent waited 30 minutes for a developer, then answered itself

Same ambiguity, same guardrails, same model — the only variable was whether a human replied. The second time nobody did, and the agent downgraded its own stop and committed. The fix took five days and touched no application code: no answer is never permission.

Read on LinkedIn ↗
Next · The amber gate we removed, and the tripwire nobody called

Complete

Series 1March 2026

What 47 people taught us about AI coding agents

A five-week AI pilot with 47 volunteers across ten departments at a European scale-up. Six findings, with the numbers — including the ones that made me change my mind.

47volunteers
10departments
5weeks
150survey responses
Read the series →
Inside06 findings
  1. The most consistent gains weren’t in engineering
  2. Engineering trust plateaued at 3.59 out of 5 — and that’s the right answer
  3. Three evaluations in three years, and the tooling answer kept reversing
  4. Some engineers saved 10–12 hours a week. Others got almost nothing
  5. The ceiling wasn’t model quality. It was usage limits
  6. The #1 complaint tripled over five weeks — the best news in the data