Skip to content

Resource

The one question to ask management about the AI agent.

What is it allowed to do, and how do you know? These questions help you prepare a board discussion about agent access, approvals and launch conditions. Ask for one concrete workflow and the records behind each answer, then use the briefing builder to write up the risks and decisions.

Briefing guide

Questions, records and launch conditions.

  • Agent permissions and controls
  • Board questions
  • Actions the agent can take
  • Supporting records to request from management
  • Launch options
  • Editable risk matrix, owner register, and local JSON/SVG/print exports

Discussion prompts

One-page board briefing

Three prompts for the conversation with management about permissions, approvals and launch conditions.

The board question

What can the agent read, write, delete, send, execute, approve, retrieve, or remember, and what artifacts support management's answer?

Launch options

Proceed, pilot-only, or pause pending controls. Tie the answer to authority, approvals, logging, rollback, and customer impact.

Records to request

Show tool inventory, data classes, approval gates, incident rollback path, and the most recent abuse-case test results.

01

Make them show you a workflow

Ask management to walk one customer request from sign-in to the data returned and any action taken. Note where access or approval is checked and which component records the result. If the answer is a policy statement rather than a record, ask again.

02

Turn the discussion into a report

The AI Risk Briefing Builder lets you describe the system, rate each risk with the reasoning shown, name owners, and record the decisions leadership has to make. Preview the matrix and decision register, then export JSON, a presentation SVG or a printed report. A redaction switch strips all entered text and dates from the exports if you need it. The ratings are judgments on an ordinal scale; the tool doesn't measure probability or predict financial loss.

Next step

Turn the discussion into a briefing.

Build an AI risk briefing