The vocabulary we use when scoping and operating an AI coworker.
The written record of every decision a workflow contains, what each one requires, and who or what is allowed to make it.
The three settings a decision can be assigned: runs independently, drafts for approval, or remains human.
Adapting AI to the operation a company already runs, rather than redesigning the operation around the software.
Everything a production AI workflow needs after deployment: hosting, monitoring, permissions, exceptions, cost, reporting, and improvement.
Who leads the work: fully agentic, human-in-the-loop, or agent-in-the-loop — the machine, the machine with human approval, or the human with agents extending them.
The optimization target: the level of machine responsibility that produces the best operating outcome, not the highest automation percentage.