How to Prep for Any Meeting in 15 Minutes With AI (The Exact Prompts I Use)

Why meeting prep keeps falling apart

Let’s name the real problem. It isn’t that you don’t know how to prepare. It’s that there’s no time to. Microsoft’s research found the average knowledge worker is interrupted roughly every two minutes during core hours, around 275 times a day, and that emails, chats, and meetings already swallow about 60% of the working day. [1]

So prep gets deferred until the five minutes before the call, when you skim the last email in the thread and hope for the best. We’ve all done it. You’re not disorganised. You’re outnumbered by your own inbox: the average worker now receives around 117 emails a day. [1]

AI fixes this not by being clever, but by being fast and structured. The whole point of the routine below is that it’s the same four steps every time, so you stop reinventing prep and just run the playbook. Fifteen minutes, start to finish.

One note before we start: never paste confidential client data or anything sensitive into a public chatbot. Use a paid or enterprise tool with privacy controls for anything that isn’t already public, and keep the genuinely sensitive details out entirely.

Step 1: Build the one-page brief (3 minutes)

Walking in cold is the enemy. The first move is to turn whatever you have (an email thread, an agenda, a few notes) into a tight brief you can read in thirty seconds.

Paste the raw material into your AI tool and use this:

Why this works: it forces clarity on the one thing that matters most, the objective. Half of all meetings start without anyone being able to state why they’re happening. This prompt makes you answer that before you walk in.

What you can do right now: grab the next meeting on your calendar, paste the invite and any related email into a chatbot, and run this. You’ll have your brief before your coffee’s cool.

Step 2: Generate the questions that matter (3 minutes)

Good meetings are driven by good questions, not by whoever talks most. AI is genuinely useful here because it doesn’t share your blind spots.

That last instruction is the one that earns its keep. The “question nobody will raise” is often the awkward one about budget, timeline, or who actually owns the decision. Having it ready, phrased calmly, is the difference between a polite meeting and a useful one.

Step 3: Pressure-test your own position (4 minutes)

If you’re going into a meeting with a recommendation, a pitch, or a point of view, this step is where AI shines as a sparring partner.

The phrase “don’t be polite about it” matters. Left to its defaults, AI tends to flatter you and agree. You have to give it explicit permission to push back, or you’ll get mush. (This is the same reason giving AI permission to say “I don’t know” reduces made-up answers, a habit we cover in our work on keeping AI honest.)

You’ll walk in having already heard the hardest version of every objection. Nothing in the room will surprise you. That composure is what people read as “well prepared,” and you got there in four minutes.

Step 4: Pre-write the follow-up (3 minutes)

Here’s a trick most people miss: write the follow-up email before the meeting, not after. It sounds backwards. It’s the opposite.

Two things happen. First, drafting the follow-up forces you to picture what a successful meeting actually produces: decisions, owners, dates. That clarity changes how you steer the conversation. Second, after the meeting you just fill in the blanks and hit send in ninety seconds, while everyone else lets their action items evaporate.

If your meetings tend to spawn a dozen scattered tasks, pairing this with an AI notetaker closes the loop nicely: the notetaker captures what was said, your pre-written template captures what to do about it.

Which tool to use, and a 2-minute setup tip

You can run this routine in ChatGPT, Claude, Gemini, or Copilot. They’re all more than capable for prep work. The bigger upgrade isn’t which tool, it’s setting it up so you don’t paste your context every single time.

If you use Claude, create a Project for “Meeting Prep” and drop in a short note about your role, your company, and how you like briefs written. Every chat inside that Project then starts with that context baked in. Our guide to Claude Projects walks through this. ChatGPT’s custom instructions and saved memory do the same job.

This is where the routine goes from “neat trick” to “default habit.” When the setup cost drops to nearly zero, you actually do it before every meeting instead of only the scary ones.

The mistakes that waste the time you just saved

A few honest warnings, because this routine can backfire if you treat the output as gospel.

Don’t read the brief aloud like a script. The AI brief is your warm-up, not your lines. People can tell when you’re reading at them. Use it to get oriented, then talk like a human.

Don’t trust facts it volunteers. If the AI adds a statistic or a “fact” you didn’t give it, assume it might be wrong until you check. For prep, lean on it to structure your information, not to supply new claims.

Don’t over-prep small meetings. A 15-minute check-in doesn’t need a stress-tested position. Match the effort to the stakes. The full routine is for the meetings that actually matter: the pitch, the budget conversation, the difficult stakeholder.

Start with your next high-stakes meeting. Run the four prompts once. If you walk out thinking “that was smoother than usual,” you’ve found your new default. The hour you used to lose to prep was never the problem. The lack of a repeatable system was.