AIF1RSTAcademy
All missions
AI LeadersGovernanceSupport level: Guided

Your Team's AI Working Agreement

Give your team one page of shared rules for AI — what it may touch, what stays human, how outputs get reviewed — and pick one supervised pilot.

Campaign:
Build an AI-Ready Team
Estimated effort:
~180 minutes total

The problem

Your team already uses AI — unevenly, quietly, and without shared rules. Nobody knows what's allowed, what's checked, or what's risky.

Why it matters

Team adoption is a leadership act, not a tooling decision. The agreement is where human oversight becomes practice instead of a slogan.

Context

Adoption without agreement creates silent risk; prohibition without alternatives creates shadow use. A one-page working agreement beats both. You lead the drafting; the team makes it real.

Objective

Draft a one-page AI working agreement with your team and select one supervised pilot task to learn from together.

Pain point

Inconsistent, invisible AI use across the team: duplicated effort, unreviewed output, and unmanaged risk.

Opportunity

A team with explicit AI rules moves faster than one that improvises — and builds trust with leadership, clients, and each other.

Expected transformation

From improvised individual AI use to one shared, written agreement and one supervised team pilot with a named reviewer.

Prerequisites

None — this is a first mission.

Risks and guardrails

  • Writing rules alone that the team ignores.

    Guardrail: Checkpoint 2 requires at least one working session with the team — the agreement is co-drafted, not decreed.

  • Letting client or company data leak through personal AI accounts.

    Guardrail: The agreement's first section defines what data may never enter external tools.

  • Turning the agreement into a prohibition list that pushes use underground.

    Guardrail: Every 'don't' pairs with an allowed alternative — the agreement enables supervised use.

Checkpoints

  1. 1

    Map current use honestly

    Survey your team (anonymously if needed): what AI tools are already in use, for what, with what data. No blame — you need the real picture.

    ~45 min

  2. 2

    Co-draft the one-pager

    In a working session, draft: allowed data, review rules, tasks that stay human, approved tools, and who to ask. One page maximum.

    ~75 min

  3. 3

    Pick one supervised pilot

    Choose one team task for assisted work under the new rules: scope, reviewer, success measure, and a review date.

    ~60 min

Deliverables

  • The one-page AI working agreement — Allowed data, review rules, human-only tasks, approved tools, escalation contact — signed by the team.
  • Pilot definition — One task, one reviewer, one success measure, one review date.

Evidence

  • The agreement and pilot definition with roles instead of names; internal tool names may be generalized.

Evidence is private by default. You choose if it ever becomes anything else — and that choice requires your explicit consent.

Success criteria

  • The current-use map reflects input from the whole team.
  • The agreement fits one page and pairs every restriction with an alternative.
  • Human-only tasks are explicit.
  • The pilot has a named reviewer role and a review date.

Review

Self-review with rubric · Peer review

Self-review against the success criteria is available today. Reviewed verification beyond self-review opens as the Founding Community's review circles form.

Reflection

  • What did the honest usage map reveal that surprised you?
  • Which rule generated the most discussion, and what does that tell you?
  • What would make the team revisit and update this agreement?

Impact measurement

Baseline: no shared rules, unknown usage. Result: signed agreement + one running supervised pilot. Category: risk reduced / workflow improved (self-reported).

Capability rewards

  • Governance · 40 XP
  • Human-AI collaboration · 30 XP
  • Communication · 30 XP

What this unlocks

  • AvailableThe agreement format as a template for other teams you lead.
  • In preparationThe next mission in Build an AI-Ready Team: running the pilot and reviewing it together.

Privacy

Private by default. Survey answers stay anonymous; the evidence uses roles, never individual names.

Accessibility

The working session can be sync or async, in person or remote. All artifacts are plain text. Self-paced.

Need help? The help ladder

Help arrives in graduated steps so the work stays yours — open only as far as you need.

  1. 1Objective reminder

    Draft a one-page AI working agreement with your team and select one supervised pilot task to learn from together.

  2. 2Guiding question

    What did the honest usage map reveal that surprised you?

  3. 3Hint

    Survey your team (anonymously if needed): what AI tools are already in use, for what, with what data. No blame — you need the real picture.

  4. 4Partial example

    The one-page AI working agreement — Allowed data, review rules, human-only tasks, approved tools, escalation contact — signed by the team. · Pilot definition — One task, one reviewer, one success measure, one review date.

  5. 5Template

    The agreement format as a template for other teams you lead.

  6. 6Walkthrough

    Work checkpoint by checkpoint in order — each checkpoint description is a step of the walkthrough.

  7. 7AI coach

    Opens with the AIF1RST Guide and community review circles — not yet available.

  8. 8Mentor or staff

    Opens with the AIF1RST Guide and community review circles — not yet available.