You got the tap on the shoulder. Maybe from your manager, maybe from an executive, maybe you volunteered because you're the person on the team who's been playing with ChatGPT. Either way, you're now the unofficial AI person for your department. You don't have a budget. You don't have a team. You don't have a plan. You have curiosity and a vague mandate to "help the team explore AI." Here's your playbook.
Week 1: Map the Pain
Forget AI for now. Spend your first week understanding what your team actually does and where it hurts.
Talk to five colleagues individually. Not in a meeting. One-on-one, informal. Ask three questions:
- "What's the most repetitive part of your week?"
- "Where do you spend time on tasks that feel like they should be faster?"
- "What information do you wish you had quicker access to?"
Write down their answers. Don't suggest AI as a solution yet. Just listen.
You're building a pain map, and the pain map becomes your use-case list. The best AI applications aren't the ones that sound impressive. They're the ones that solve problems people already complain about.
Week 2: Pick One Task
From your pain map, pick one task that meets all four criteria:
Repetitive. It happens at least weekly, ideally daily.
Low stakes. If AI gets it wrong, nobody gets hurt. No compliance risk, no financial exposure.
Data available. The inputs exist in a form AI can process (text, spreadsheets, documents).
Quick to test. You can try it in under an hour with existing tools.
Don't overthink this. The goal isn't to find the perfect use case. It's to find a quick win that builds credibility.
Examples that work well:
- Drafting routine emails or status updates
- Summarising meeting notes or long documents
- Classifying or sorting incoming requests
- Creating first drafts of reports or presentations
- Extracting specific information from large documents
Week 3: Build Your First Demo
Use an accessible AI tool. ChatGPT, Claude, or whatever your organisation has approved. Take a real task from your pain map and do it with AI.
Document everything:
- What the task was
- How long it takes manually
- How you used AI (include the actual prompts)
- How long it took with AI
- How good the output was (be honest about limitations)
This documentation becomes your pitch. Not a slide deck. A real example with real numbers from real work.
Your first demo should make someone say "wait, that took you how long?" Not "wow, that's impressive technology." Practical beats impressive every time.
Tim Hatherley-Greene
Chief Operating Officer
Week 4: Show One Person
Not a presentation. A conversation. Find the colleague from your pain interviews who seemed most curious (not most technical, most curious) and show them what you did.
Sit next to them. Walk through the task. Let them try it with their own work. Be honest about what worked and what didn't.
If they're interested, help them try it with one of their own tasks. If they're sceptical, that's fine. Thank them for their time and find someone else.
You're looking for early adopters, not unanimous buy-in. In a team of 20, you need 3-4 people actively trying AI to create momentum. The rest will follow when they see results.
Weeks 5-8: Build Evidence
Now you have a small group experimenting. Your job shifts from demonstrating to documenting.
Track specific outcomes:
- Time saved per task (be precise: "45 minutes per week on status report drafting")
- Quality improvements (if applicable)
- Tasks that didn't work well with AI (equally important to document)
Build a simple case study. One page. The task, the approach, the result, the limitations. This is your internal marketing material. When someone asks "what's AI doing for your team?" you hand them this.
Keep experimenting. Try new use cases from your pain map. Not everything will work. The ones that fail teach you about the boundaries. Document those too.
Weeks 8-12: Expand Carefully
You've got results. A few colleagues are using AI regularly. You have evidence. Now expand, but carefully.
Run a lunch-and-learn. Not a training session. A "show and tell" where your early adopters demonstrate what they've done. Peer demonstration is more convincing than anything you can present.
Create a shared resource. A simple document or Slack channel where your group shares prompts, tips, and lessons. Make it low-effort and practical. Not a knowledge base. A shared notebook.
Connect with other champions. If other departments have AI enthusiasts, find them. Share approaches. Learn from their experiments. A cross-department network of champions creates organisation-wide momentum.
What to Avoid
Don't overpromise. AI is useful for specific tasks. It's not magic. Setting realistic expectations builds trust. Overpromising destroys it.
Don't go rogue. Work within your organisation's policies. If there's no AI policy yet, flag it. "We're experimenting with AI tools for internal tasks - is there guidance on approved tools and data handling?" Being proactive about governance builds credibility.
Don't force it. Some colleagues won't be interested. That's fine. Adoption spreads through pull, not push. Focus your energy on the willing, and the results will gradually pull in the cautious.
Don't make it about you. Your goal is to make your team more effective, not to become the AI person. The best outcome is when your colleagues are using AI independently, without needing you. That means you've succeeded.
Don't ignore the feelings. Some colleagues will feel threatened, anxious, or dismissive. These responses are normal. Listen. Acknowledge. Show, don't argue. "I get that. When I first tried it, I felt the same way. Want me to show you what it actually does?"
The 90-Day Goal
By week 12, aim for:
- 3-5 colleagues using AI for at least one regular task
- 2-3 documented use cases with measurable outcomes
- A simple governance conversation started with your manager or IT team
- A connection with at least one champion in another department
That's enough. Not transformative. Not impressive on a slide deck. But real, measurable, and sustainable. The organisations that adopt AI well will do it through hundreds of these small, practical wins rather than a single dramatic programme.
Being an AI champion isn't about being an AI expert. It's about being curious enough to experiment, practical enough to find real applications, and generous enough to help your colleagues see the value. You don't need a budget. You need a pain map and a willingness to show people what's possible.