The words behind an AI employee.

Short definitions for the way we talk about the work, the loop, and the controls that keep it safe.

AI employee
A model plus the operating system it needs to hold a job, run by a human operator to an outcome. Not a chatbot and not a tool you babysit, a role that delivers work.
Blueprint
The homepage tool that researches your company and returns up to three grounded AI-employee proposals, each one only shown if it clears a strict review.
Guardrails
The rulebook, sign-off gate, kill switch, spend cap, sandbox, and traces together, the safety system that lets an operator run on real work.
Horizon move
A next-stage capability most companies plan a year or two out, stood up now because the signals say it is time.
Kill switch
The control that stops an operator instantly. Operators fail closed, never open.
Operator
The forward-deployed human who runs the loop and owns the result. Every AI employee has one on top.
Rulebook
The lines an operator never crosses, set before it runs.
Runbook
The written steps for the work an operator does over and over, so the same job is done the same way every time.
Sense, act, check, learn
The four steps of the loop. Sense what is happening, act on it, check with a person where judgment is needed, and learn so the next run is sharper.
Sign-off gate
The point in the loop where a human approves anything that carries risk before it goes out.
Spend cap
A hard limit on what an operator can spend, so it can never run up a bill.
The loop
The single signature device behind every operator: sense the trigger, act on the work, check with a human on anything risky, and learn from what worked.