Ownership Radar

AI ACCOUNTABILITYFIT

Purpose Give every AI output a single human owner, a clear backup, and a fast escalation route.
Intervention type Role-clarity mapping, psychological ownership
Lead Team or workflow manager
Time 15 min prep · 60 min group run · 10 min publish

Expected outcomes

  • User: Knows which outputs carry their name and what to do if they fail.

  • Team: One-page map with owner with an approver for each AI output.

  • Business: Error-to-fix time drops. Reduced risk.

  • Org: Audit trail shows a human fingerprint on every AI decision.

What to bring to the session
  • Output List Sheet (top 5–10 AI outputs with IDs)

  • Radar Board (Build in the steps)

  • Risk Tier Table (See below)

  • Escalation Trigger Cheat Sheet (See below)

Steps

1 | Setup

Step 1: Choose which AI outputs to map

  • Pick up to 10 key AI outputs.

  • Focus on ones that leave the team or affect decisions (e.g. “drafted reports”, “fraud alerts”, “flagged transactions”).

  • Write each one on its own sticky or on a board. Label them clearly.

Step 2: Prepare the Ownership Radar Sheet:

Draw a table with the below headers. Get familiar with what a filled out version looks like:

2 | Run the Radar Session

Step 3: Define Ownership

Read this out loud at the start of the session
“Owner means: the person who clicks Send, Accept, or Submit on this AI output. They are responsible for the final version. They don’t have to write it or approve it, but they own it.”

  • Add this to the board or screen.

  • Ask: “Is that clear to everyone?” If not, pause and clarify.

Step 4: Place Owners
For each AI output:

  • Ask the group: “Who hits Send?”

  • Write that name in the owner column.


Tip: If two people argue for the same output, split the task into two clear parts.
Each part gets its own Owner.

Step 5: Add a Backup

  • Ask: “If the Owner is away, who steps in?”

  • Write their name in the backup column.

  • Place it next to the owner name for that output.

  • Encourage rotation every quarter to avoid over-reliance.

Step 6: Mark the Approver (if needed)

Only add a Approver if one of these is true:

  • Legal says it’s needed

  • The risk is high (loss in money, safety, data)

  • The business important is high (impact, value)

  • The system requires it

Step 7: Draw the hand-off line

  • For each output, follow a simple logic showing:

    • Owner → (Approver, if any) → Archive or storage

Step 8: Set the Escalation Trigger

Some outputs are more important than others. Here you define which and outline where they do next.

  • Ask: “When should this go up the ladder?”

  • Pick one clear rule. Example:

    • “If confidence < 80%”

    • “If value > £50,000”

    • “If the task is overdue by 2 hours”

  • Write the rule under the output.

Tip: Not sure where to start? Use the Escalation Trigger Cheat Sheet below.

Step 9: Decide the Signature Cue

  • Ask: “Where will the Owner’s name appear?”
    Choose one:

    • In the product UI (“Verified by Jamie – 14:03”)

    • In the comments (“// approved-Lisa-0930”)

    • In the system log

    • In the google doc with a backup of the chat

    • In the project log of key AI outputs

Step 10: Capture and publish the map

Make sure every team member listed is aware of their role in the ownership of AI outputs.

Resources

Escalation Trigger Cheat Sheet

Important to consider: Always include a manual override rule for surprises.

Other methods within the fit block