AI WORLD HQ
If you manage people, you have probably already been asked some version of "should we be using AI for this?" — and you may not have had a confident answer. The good news is that leading an AI transition is mostly a people job, not a tech job. The five steps below give you a structure you can start this week without buying new software.
Before you start
You'll need four things, and you almost certainly have them already:
- A regular team meeting (in person or on video)
- A shared document tool your team already uses — Google Docs, Notion, or Microsoft SharePoint all work
- Access to a text-based AI tool for testing. The free tiers of ChatGPT, Claude, or Gemini are fine for the steps below; no paid plan is required
- Roughly 90 minutes total spread across one week, not all at once
No new apps to install. The goal of these steps is to set up habits, not to deploy software.
Step 1 — Run a 15-minute "show and tell" meeting
Block 15 minutes at the top of your next regular team meeting for a quick AI "show and tell". In advance, ask one team member to share one prompt (the instruction you type into an AI) that saved them time this week. The rest of the team listens, asks one question, and that's it. No slides, no demos of new tools.
💬 Example message to send the team the day before: "For our next meeting, please bring one AI prompt that saved you time this week. Just share what you typed and what you got back. 60 seconds max, no prep needed."
You'll know it worked when at least one person shares a real example and the rest of the room has a follow-up question. That single exchange is the start of a learning culture.
Step 2 — Map your team's repetitive tasks
Before suggesting any tool, list the five most repetitive tasks your team does each week. A "repetitive task" is anything you'd describe the same way three weeks running — meeting notes, status updates, weekly reports, draft replies to common questions. Open a shared doc and create a simple table with three columns: Task | Who does it | Time spent per week. Fill it in together in a 20-minute team session.
💬 Example row to seed the table:
| Task | Who does it | Time per week |
|---|---|---|
| Summarize meeting notes | Everyone, in turn | 3 hours |
| Draft weekly client update | Report writer | 2 hours |
You'll know it worked when your team can point to at least three tasks that each take more than one hour per week. Those are your candidates for the next step.
Step 3 — Build a shared prompt library
Take the top task from Step 2 and turn it into a reusable prompt template. Save it in a shared folder your team already uses — Google Drive, Notion, SharePoint, or even a pinned channel in Slack all work. Pick one task, such as summarizing meeting notes, and write a short prompt that anyone can copy, paste, and tweak. Include placeholders so people know what to swap in.
💬 Example prompt template for meeting notes (copy and adapt): "Summarize the meeting notes below into 5 bullets. Highlight decisions made and any open questions. Keep it under 150 words. [paste meeting notes here]"
You'll know it worked when two or more team members use the same template in the same week and get useful results without extra help. That is the moment a personal habit becomes a team habit.
Step 4 — Write a one-page AI policy
Write a short, plain-English document — one page maximum — that sets the ground rules for AI use. Post it where your team already looks for guidance: the team wiki, your intranet home page, or a pinned message in your team chat. The policy should cover three things only: (1) what data is okay to paste into an AI — a sensible default is "never paste anything you wouldn't email externally" — (2) the human-review rule, and (3) who to ask when stuck. Anything longer and people won't read it.
💬 Example policy line to include verbatim: "Every piece of AI-generated content must be reviewed, edited, and approved by a human before it goes to a client, a senior manager, or any external use."
You'll know it worked when a new team member can answer "What's our AI policy?" by pointing to one specific place. That's the test that the policy is real, not just buried in a shared drive.
Step 5 — Schedule a monthly AI check-in
Put a recurring 30-minute meeting on the calendar — once a month, same week each month — and title it clearly (for example, "Team AI check-in"). In each meeting, ask three questions: What's working? What's not? One thing we should stop doing? The point is not to chase new tools; it's to notice patterns — which prompts are saving time, which ones created extra work, and what the team may be starting to trust too much.
💬 Example agenda block to paste into the calendar invite: "30 mins — AI check-in. 1) One prompt that helped this month (5 min). 2) One thing that wasted time or went wrong (10 min). 3) Decide: keep, change, or drop our current prompt library? (15 min)."
You'll know it worked when team members start bringing small fixes and tweaks to the library without being asked. That's when the process runs itself.
Common mistakes
Mandating one tool for everyone. Different team members think and write differently. Fix: let people choose their own AI tool, as long as the output meets the policy in Step 4.
Skipping the policy because it feels too formal. Without written rules, people will paste sensitive data into public tools without thinking. Fix: keep the policy to one page — three rules is enough to start.
Trying to automate ten tasks at once. The team gets overwhelmed and nothing gets properly tested. Fix: pick one task from Step 2 and run it through the prompt library for two weeks before adding anything new.
Treating AI output as final. AI can confidently make up facts — this is called a hallucination (when an AI invents details that sound right but aren't). Fix: the human-review rule in Step 4 is non-negotiable, even for "obvious" outputs.
Only celebrating wins. If you only hear about successes, you'll never learn where the tool is failing your team. Fix: make Step 5 a safe space to share what didn't work, not just what did.
Wrap-up
Leading your team through an AI transition is mostly about habits, not software. The five steps above — a 15-minute share meeting, a workflow map, a shared prompt library, a one-page policy, and a monthly check-in — give you a structure you can begin tomorrow without buying anything new. Choose one step and do it before next Friday. That single action will tell you more about what your team actually needs than any software comparison ever will.