Public profile
Adrian
A growing catalog of my orchestration failure modes. Tests taken, verdicts, trends. Updated when episodes ship.
The newsletter argues that learning orchestration in the agentic era requires publishing the numbers. This is my page. Every orchestration test the newsletter ships, I take first. My verdicts are public and permanent. The profile does not "complete" — it accumulates.
Orchestration tests — verdicts
The closet-hours test (v1) — ep 001
Tests not yet shipped
Each new episode that adds a test appends a row here. My verdict is taken and posted before the episode goes public.
The distribution — ep 004
The replacement test — ep 011
The presence test — ep 012
The delegation test — ep 015
The ledger test — ep 020
Failure modes currently catalogued
- Over-building — evidenced in episode 001. Hiding from the small human to build things that are not due. Magnitude: unsettled, still counting tokens.
The list grows. It does not shrink. That is the point.
How to read this page
Each row in Orchestration tests — verdicts is a specific failure mode I have been tested on, with my result preserved for public accountability. Verdicts do not age off — they accumulate into trends. When I retake a test, both the old and new verdicts are kept.
The Adrian Indicator at the top shows current-state metrics: sleep, tokens, open sessions. These update when the data updates (phase 1: manually; phase 2: hourly via a local daemon).
The failure modes block below the test table summarizes the pattern that emerges across tests — the kinds of orchestration mistakes the data reveals I am making.
There is no orchestration score. There is no readiness percentage. There is no gamification. The profile exists to be looked at honestly and updated honestly. That is the entire design.
Subscribe
Biweekly. Free. Paid tier arriving in a few months when it has been earned.
email only — no tracking pixels, no referral nags, no popups ever