The professional ecosystem for Agent Operators.
The agents arrived. The people using them well are still mostly alone. We're building the connective tissue around them: Works on My Machine, the newsletter; Artificial Ruby, a monthly meetup in NYC; and engagements that put the practice inside real teams.
The work
You recognize this. Maybe you live it.
Someone has a Claude Code ritual that turns forty minutes into eight. Someone else has a custom GPT only they understand. Someone in ops keeps a spreadsheet that talks to three APIs and a prayer candle. Each one works. None of it transfers.
the Claude Code ritual
Works perfectly for one engineer. Nobody else can reproduce it.
the custom GPT
Useful, undocumented, only known to whoever built it.
the spreadsheet monster
Talks to three APIs and a prayer candle.
the research-synthesis flow
A PM has it; nobody else understands what's in the prompt.
the prototype pipeline
A designer's screenshots-and-vibes routine.
the support macro chain
Saves four hours a week. Or it did, until last Tuesday.
This is the work of an emerging profession.
The role
If this is your work, this is the language for it.
You design, run, evaluate, and improve the workflows that use AI agents, even if nobody you work with calls it that yet.
You sit between human intent and machine execution: mapping work, shaping context, connecting tools and data, supervising agents, verifying outputs, writing things down, and keeping workflows current as models change.
We call the role Agent Operator. The full definition, the nine-skill stack, and the maturity model live on its own page.
Engage