Lesson 1 · Why pilots use checklists, and why your team should

Experienced people make mistakes most often on routine tasks, when they're rushed or interrupted. Checklists don't assume people are careless. They assume people are human. AI tools add a new risk: their output sounds confident whether it's right or wrong. A short checklist turns "I think it's fine" into "we checked."

Takeaway: a good checklist is short, used every time and owned by the team.

Lesson 2 · Before you prompt: data, purpose, permission

• Data: Am I about to paste customer, employee, financial or confidential information? Is this tool approved for that data?
• Purpose: What decision or document will this output feed into? The higher the stakes, the more verification it needs.
• Permission: Does our AI use policy allow this tool for this task? If there's no policy, ask before you start.

Lesson 3 · Briefing the AI: a clear prompt

Pilots brief the departure before takeoff. Brief the AI the same way:

1. Role: "You are a customer-service writer for a plumbing company."
2. Task: "Draft a reply to this complaint."
3. Context: facts, policies, tone and audience.
4. Constraints: length, things to avoid, "say 'I don't know' if unsure."
5. Format: bullets, an email, a table.

Lesson 4 · Trust, but verify: cross-checking output

On the flight deck, one pilot sets and the other verifies. For AI output:

• Check every number, name, date and quote against the source.
• Open every link and citation, because AI can invent them.
• Read it as the customer would: is the tone right? Would it embarrass us?
• Record who reviewed it before it goes out.

Lesson 5 · Crew roles and the "stop" call

Crew resource management means anyone can speak up, whatever their rank. Make it explicit: who uses the AI, who reviews the output, who approves it, and the rule that anyone can pause AI use if something looks wrong, with no penalty for doing so.