Brainstorm with AI (and Get Ideas That Aren't Everyone Else's)

Ideas for a team event that aren't everyone else's, then test your two favorites.

Level 1 · Chat Free · 30 minutes · Updated Oct 31, 2026

Ask AI for ten ideas and you’ll get ten ideas in five seconds. That’s the easy part. The hard part is that they tend to be the same ten ideas everyone else gets: reasonable, a bit generic, and a little too easy to accept. In the next 30 minutes you’ll run a brainstorm where AI widens your thinking instead of replacing it, and end with two ideas you’ve actually tested.

Remember the intern: smart, fast, first day on the job. Ask them for ideas and they’ll fill a whiteboard in a minute. But they’ve never been to your office party, don’t know your budget, and suggest what most people would suggest. Your job today is giving them real limits, asking for range, and keeping your own ideas in the mix.

What you’ll get

What you need

• Claude, any plan. Just a normal chat. Other tools work too; see the end.
• A real problem that needs ideas. We’ll use this one throughout: ideas for an annual team event for 40 people, half of them remote, with a budget of about $50 per person and half a day.
• A pen and paper, for Step 1.

The three rules

Part 1: One brainstorm, done properly (30 minutes)

Step 1. Write three ideas of your own first (3 minutes)

Before opening the AI, write down three ideas, however rough. For our event: “a cooking class over video”, “a quiz with mixed remote/in-office teams”, “volunteering half a day”. These are your anchor. Later you’ll check whether the AI added anything, or just repeated you in nicer words.

Step 2. Give it the real constraints (4 minutes)

Open a new chat in Claude and paste:

I need ideas for our annual team event. Don't give ideas yet. Just confirm you understand the constraints and ask me up to 3 questions.

Group: 40 people, about half remote in different time zones.
Budget: about $50 per person.
Time: half a day.
Goal: people from different teams actually talk to each other, not just watch something together.
What didn't work last year: a long video call with a speaker; people turned their cameras off.

The “what didn’t work” line is gold. It tells the intern what to avoid better than any description of what you want.

Step 3. Ask for range (5 minutes)

Now give me 20 ideas, grouped into 5 different categories (for example: competitive, creative, helping others, learning something, low-key social).
Include at least 3 ideas that seem a bit unusual or risky.
Don't include anything similar to these three ideas I already have: [paste your three ideas].
One line per idea.

Categories force variety. Excluding your own ideas forces the AI past the obvious.

Step 4. Score them against your criteria (5 minutes)

Put all 20 ideas, plus my 3, in a table scored 1-5 on: people from different teams talk to each other; works equally well remote and in person; fits $50 per person; easy to organize in 3 weeks.
Don't pick a winner.

Now you pick, not the AI. Choose your top two, and be honest if one of your own ideas still wins. That’s a fine result.

Step 5. Push your two favorites (7 minutes)

For these two ideas: [idea A] and [idea B]
1. What would it take to run it? List the steps and roughly what each costs.
2. What are the three most likely ways it goes wrong on the day?
3. Give me a cheaper, simpler version.
4. Mark any cost or fact you're not sure about as [check].

This turns “sounds fun” into something you can propose.

Step 6. Check the facts yourself (6 minutes)

AI is good at ideas and unreliable on specifics. Anything it gave you about prices, venues, suppliers, booking times or tools is a guess until you check. Open the real websites, or ask a colleague who has organized one before. Update the plan with the real numbers.

Part 2: Use it with your team

AI works well as one voice in a group brainstorm, not as the whole brainstorm.

1. Humans first. Give people 5 minutes to write their own ideas silently before anyone shares, and before anyone opens an AI tool.
2. Then the AI round. Paste everyone’s ideas into the chat and run Step 3: “20 ideas unlike these.”
3. Mix, then vote. Put all ideas on one board, unlabeled, and let the team vote with the criteria from Step 4.
4. Assign the checking. Whoever champions an idea checks its facts before the next meeting.

Why checking matters

With brainstorming, the risk isn’t one wrong fact. It’s ending up with the same ideas as everyone else, and not noticing.

• AI ideas help individuals but make results more alike. In an experiment published in Science Advances, people who got story ideas from an AI wrote stories rated more creative and better written, with the biggest gains for the least creative writers. But the AI-assisted stories were more similar to each other than stories written without AI. The authors describe it as individuals gaining while the group ends up with a narrower range of ideas (Doshi & Hauser, 2024). That study was about short stories, not work ideas, but the same pull toward the familiar is why this tutorial starts with your own ideas and explicitly asks for ideas unlike them.
• Specifics are guesses. Prices, venue names and booking times in AI ideas often sound precise. They are not checked facts until you check them (Step 6).

Checklist

Before you propose your ideas:

• I wrote my own ideas before seeing the AI’s
• I gave real constraints: budget, size, time, what didn’t work before
• I asked for different categories, not just more ideas
• I chose with my own criteria, and my own ideas were in the running
• My top ideas have a cost, a list of risks and a cheaper version
• Every price, venue or supplier detail has been checked against a real source

What you just learned

On the surface you came up with event ideas. Underneath, you practiced giving context: real limits, including what went wrong before, make ideas better. And you practiced checking in an unusual form: checking that the AI added range rather than just echoing you, and checking facts before an idea becomes a plan.

The same method works for: names for a product or project, topics for a newsletter, ways to cut a cost, ideas to improve a process, and questions to ask in an interview. Change the constraints in Step 2; keep “your ideas first” and “ask for range.”

Using a different tool?

The prompts work anywhere. Brainstorming needs nothing but a chat.

Features change often. This table was checked in October 2026.

Sources

• Generative AI enhances individual creativity but reduces the collective diversity of novel content (Doshi & Hauser, Science Advances, 2024)