What is a remit?
A remit is the written scope of a Digital FTE, and it is the first document a manager writes when hiring one. It answers three questions in three lists: what the role owns, meaning the function, the properties, the tools, and the outputs it is accountable for; what it escalates, meaning the categories that reach a person regardless of confidence; and what it never does, meaning the actions that stay human at every level of the autonomy ladder.
The word is borrowed from ordinary management, where a remit is the area of responsibility given to a person or a committee. It fits an AI employee for the same reason: a worker without a written scope cannot be judged, corrected, or trusted, because there is nothing to hold it to.
How does a remit work on ftes.ai?
On ftes.ai the remit is written in the first-run wizard, which pre-fills everything the platform already knows about the organization and asks only for what is new to the role, and it lives afterwards in the Setup tab beside the autonomy ladder, the budget, and the connections. The Digital FTE reads its remit before every proposal; a proposal outside it does not happen, and a case the remit does not cover is escalated rather than guessed at.
The remit is also where the always-escalate list is kept for a role: for a Customer Support FTE, refunds, account access, complaints, and distress; for a Recruiter FTE, anything about a specific candidate's outcome. That list is fixed and does not shrink as other classes of action earn autonomy.
Why does the remit matter more than the model?
The remit matters more than the model because most of what looks like an AI quality problem is a scope problem. A Digital FTE that drafts replies in the wrong register, chases the wrong accounts, or proposes matches above a threshold the finance lead would never approve is usually working exactly as written; the writing was loose. Tightening the remit fixes the behaviour for every future proposal at once.
It also matters for governance. The remit is what an auditor, a security reviewer, or a new manager reads to understand what the role was allowed to do, and the audit trail is read against it. A role with a written remit can be shown to have stayed inside it; one without cannot be shown anything.
How is a remit refined over time?
A remit is refined from send-back reasons. Every time a manager sends a proposal back, they give a reason, and at the weekly review in the first month those reasons are read grouped by category. A reason that repeats is a rule the remit is missing: a tone the brand never uses, a customer segment that should be protected, a threshold that was set too high.
The rule is added to the remit, the next proposals inherit it, and the send-back rate for that category falls. This is the loop that turns a rough first remit into a precise one within a few weeks, and it is why the first month's approval time is worth spending closely.
What is a remit often confused with?
A remit is often confused with a prompt. A prompt is an instruction for one piece of work; a remit is the standing scope for all of a role's work, and it is enforced by the approval gate rather than hoped for by the model. A remit is also confused with targets: targets are the outcomes the role is accountable for, held in the Targets tab with evidence, while the remit is the boundary within which it pursues them.
It is sometimes confused with the autonomy ladder too. The ladder decides which classes of action inside the remit need a person's approval each time; the remit decides what is inside at all. A class of action cannot be promoted on the ladder if the remit does not include it.
- Prompt: one instruction for one task
- Remit: the standing scope for every task the role does
- Targets: the outcomes it is held to, with evidence
- Autonomy ladder: which in-remit actions still need approval
Where do you see the remit in the workspace?
The remit is in the Setup tab of each Digital FTE's workspace, in its three lists, with the always-escalate list marked as fixed. Changes to it are logged in the audit trail with who made them and when, so the version a proposal was made under can always be found. The remit template pieces in the ftes.ai library give a starting draft for each role.
Related terms: always-escalate list, never-automated list, autonomy ladder, targets, send-back.