# AI delegation charter template - TheAICommand artefact

Name: AI delegation charter template
Date: 2026-08-15
Source: https://theaicommand.com/library/artefacts/how-to-write-an-ai-delegation-charter
Article: https://theaicommand.com/leadership/your-teams-ai-delegation-charter
Licence: Free to use and adapt internally. Attribution appreciated. No warranty. Educational purposes only.
Disclaimer: General information and education only. Not legal, compliance, financial, or professional advice.

---

### Section 1: the charter, one row per recurring task

| Task | Current owner | AI's role | What stays human | Review trigger |
|---|---|---|---|---|
| | | | | |
| | | | | |
| | | | | |

**Task.** The specific recurring piece of work, named the way the person who does it would name it. "Draft the weekly client update", not "reporting".

**Current owner.** The person who owns it today. A role, not "the team". Naming an owner is what stops a task falling into an accountability gap the moment AI touches it.

**AI's role.** One of four plain settings:

- *none*
- *draft*: AI produces a first pass, a person finishes it and owns it
- *assist*: AI helps inside the task while the person drives
- *do-and-check*: AI completes it inside a set boundary and a named person checks it before it counts

**What stays human.** The part of this task a person keeps no matter the setting: the judgement, the client relationship, the final sign-off. Writing it down makes the boundary a choice rather than a thing that quietly erodes.

**Review trigger.** When this line gets looked at again: a date, plus the events that force an earlier look, a new tool, a new rule, or a near-miss.

A charter is not a decision-rights map. A decision-rights map answers who decides, placing AI on a ladder of inform, recommend or act. A charter answers which tasks the team hands to AI and which stay human. A task can sit on the charter as draft-and-check while any decision buried inside it, whether to send or whether to escalate, stays governed by the map.

### Section 2: the session plan

1. Book a 45-minute team session within a fortnight of the honest AI-and-jobs conversation, while the trust is fresh. Bring no charter of your own.
2. Ask everyone to bring their own recurring tasks in advance, each with a rough sense of how often it runs, so the room starts with real work rather than abstractions.
3. Pool every task on one surface, then split the obvious first-draft candidates, explanations, summaries and first passes, from the judgement-heavy work where a person's read is the value.
4. Draft the charter, then argue with it line by line as a team. The argument is where the boundaries get found.
5. For every task, agree the AI role, the part that stays human, and a review trigger. Where the team disagrees, default to the more human setting and note the reason.
6. Red-team the draft for over-delegation and deskilling before you adopt it, and fix what it flags.
7. Put the charter where the team can see it, name an owner for keeping it current, and set the first review date, quarterly at the latest.

### Section 3: what never goes on the charter as AI alone

Three categories stay human on purpose, and the charter should name them as deliberately as it names what it hands over.

- **Anything where a wrong result could reach a customer, harm a person, or corrupt the record before a human sees it.** These never sit at do-and-check. They are draft or assist at most, with a person between the output and the world.
- **The judgement, the relationships and the accountability.** AI can change what work gets done, but it cannot own who is affected or answer for the result.
- **The tasks that are how someone builds a skill.** Hand every first draft to a model and the junior who learned the craft by writing first drafts never learns it. Protecting those reps is a deliberate charter decision, and the review trigger is where you catch a team that has quietly automated away its own apprenticeship.

### Section 4: the charter-is-ready checklist

Adopt the charter only when every line below is true.

- Every row names a specific task, a current owner, an AI role, what stays human, and a review trigger.
- The team wrote it together, and no line was set by the leader alone.
- Nothing is set to do-and-check where a wrong result could reach a customer or harm a person before a human checks it.
- The tasks that build a junior person's skill are kept human on purpose, and the charter says so.
- No decision is disguised as a task. Decisions live in the decision-rights map, not here.
- Each line has a review trigger, and the whole charter has a quarterly review date and a named owner.
- It is visible to the team, not filed in a drawer, because a charter no one can see is not a shared agreement.