A Meeting Prep Workflow (10 Minutes Before)

TL;DR: A meeting prep AI workflow means pulling the calendar invite, one relevant email or CRM note, and the other person's name into a single prompt before the call, then asking it four fixed questions: who's in the room, what did we already agree, what should this meeting produce, and what's the one thing you can't walk in without knowing. Ten minutes.

Why does a 10-minute prep routine beat winging it?

Most people don't skip meeting prep because they're lazy. They skip it because gathering the context (the last email thread, the deal notes, who's actually going to be in the room) takes longer than the meeting itself, so it gets triaged away in favor of whatever's on fire that morning. The cost shows up anyway, just later: the first five minutes spent re-deriving context out loud, or worse, contradicting something you agreed to last time because nobody reread the thread.

A meeting-prep AI workflow doesn't replace judgment. It compresses the reading you'd do anyway into a prompt that hands back the handful of facts that matter, so the reading happens before the call instead of during it. This post covers only that ten minutes: walking into the room prepared. Running the meeting itself is a different job with its own companion guide: the agenda, the facilitation, how you handle a pushback in real time.

What do you gather before you open the invite?

Before you write a single prompt, collect the raw material. Five things, roughly in the order they're fastest to find:

1. The calendar invite itself: title, attendee list, and anything attached to it. Half the useful signal in a meeting is already sitting in the invite and never gets read.
2. The last message on this topic: the most recent email, Slack thread, or CRM note that touches this account or project. You want the last state of the conversation, not the whole history.
3. Any notes from the previous meeting, if there was one: a summary, a transcript, or just your own scrawled bullet points from last time.
4. What the other side has said publicly recently: a product update, a hire, a funding round, anything that changed since you last talked. One search, not a research project.
5. Who's actually attending, by name and role, if the invite doesn't already spell it out. A meeting with a new VP in the room needs a different brief than the recurring sync with the same three people.

None of this requires a calendar integration. Prompt Architects doesn't reach into your calendar or inbox, and it doesn't need to. You're copying five things you already have access to into one place. That gathering step should take less time than the prep itself.

What should the AI actually answer before you join the call?

A meeting-prep prompt is only useful if it answers a fixed set of questions every time, not whatever the model feels like summarizing that day. Six pull most of the weight:

• Who is actually in the room, and what's their stake in this? Not just names: whether this person is the decision-maker, an approver, or someone who just needs to be kept informed.
• What did we already agree to, and what's still open? The single most common failure in an ongoing relationship is treating a resolved question as still live, or forgetting a resolved one was ever settled.
• What does this specific meeting need to produce? A decision, a number, a signature, or just alignment. Naming the output before you join changes how you spend the time once you're in it.
• What's changed since the last time we talked? A funding round, a reorg, a competitor's launch: anything that would make an assumption from last time wrong today.
• What's the one thing I can't afford to forget or get wrong? Every meeting worth prepping for has exactly one fact like this, and it's usually the thing that gets lost between the invite and the call.
• What's my ask, if I have one? Not every meeting needs one, but if you're walking in wanting something specific, naming it before you join sharpens everything you say once you're there.

Answer those six and the prep is done. Everything else (the agenda, how you open, how you handle a pushback) happens once the call starts, and belongs to a separate workflow rather than this one.

A 10-minute meeting-prep prompt you can reuse

Here's a template that turns the five things you gathered into a one-page brief, structured around the six questions above. Copy it, fill in the brackets, and paste in whatever you collected: the invite text, the last email, your own notes from last time.

You are prepping me for a meeting. Here is what I have:

MEETING: [paste the calendar invite title, time, and attendee list]
LAST CONVERSATION: [paste the most recent email, Slack message, or CRM note on this topic]
PREVIOUS MEETING NOTES: [paste notes or a transcript summary from last time, or write "first meeting"]
WHAT'S NEW: [paste anything public that changed recently: an announcement, a hire, a launch]
MY GOAL FOR THIS MEETING: [one sentence: what you want to leave with]

From only the material above, give me:
1. Who's in the room and what each person's stake likely is.
2. What we already agreed to, and what's still genuinely open.
3. What this meeting needs to produce, in one sentence.
4. Anything in the material above that contradicts something we previously agreed to.
5. The one question I should ask if I only get to ask one.
Do not invent facts that aren't in the material I gave you. If something isn't there, say so instead of guessing.

That last line matters more than it looks. A meeting-prep prompt without an explicit instruction not to invent details will happily hand back a confident, plausible-sounding brief with a hallucination baked into it, and a wrong assumption stated with confidence in a meeting is worse than no prep at all.

How much prep does each meeting type actually need?

The table above is a starting point, not a formula. Prep time should scale with two things: how much is riding on the meeting, and how unfamiliar the room is. A discovery call with someone who has never heard your pitch needs the full workflow above (our collection of prompts built specifically for discovery calls is worth pairing with it if that's your regular seat); a recurring standup with the same three people rarely does. The mistake isn't under-prepping the low-stakes meeting. It's running the same ten-minute routine on every meeting regardless of whether it's earning its keep, until prep itself becomes the thing people skip.

What should you have open when the call starts?

Three tabs, not eight. First, the brief itself: the output from the prompt above, not the raw material you fed it. Second, whatever you'll use to write new notes during the call, because the brief tells you what you already know, not what happens in the next thirty minutes. Third, the one source document you might actually need to point to mid-conversation (the contract, the previous proposal, the ticket number) rather than digging for it live while the other person waits.

Skip opening the original email thread or CRM record unless you specifically expect to need it. If the brief did its job, the material you extracted is already in front of you in a more useful shape than the source it came from, and reopening the source is usually a nervous habit rather than something the meeting actually requires.

Is this the same as asking AI to summarize the account?

Not quite, and the difference is discipline, not sophistication. Asking a model to summarize this account produces whatever it decides is relevant, which changes every time you ask and skips whatever wasn't in the one document you happened to paste that day. A meeting-prep workflow answers the same six fixed questions every time, built from a deliberately assembled set of sources rather than whatever was closest to hand. That consistency is the entire value: it isn't that the model reasons better with a template than without one, it's that you stop forgetting to ask the questions that actually matter.

The other real difference is what you leave out on purpose. A generic account summary tends to include everything the source material contains, in roughly the order it was written. The prep brief above deliberately drops most of that: the pleasantries, the scheduling back-and-forth, the parts of the thread that don't bear on this specific meeting. It keeps only what would change what you say in the room. That's a curation decision a template makes for you once, instead of one you make, or skip, under time pressure every single time.

When should you skip the routine entirely?

Skip it for meetings where nothing changed since last time and you already know that going in. The third stand-up this week with the same three people isn't a candidate for a five-source prep routine; it's a candidate for reading yesterday's message on your way to the call. Skip it, too, for a meeting you're running and already know cold. A brief built entirely from your own material tells you nothing you don't already know, and building it anyway is busywork wearing a productivity costume.

The harder case is a genuinely open-ended conversation, one where over-preparing makes you rigid instead of ready. If the point of the meeting is to find out something you don't yet know how to ask about, a tight brief built around fixed questions can narrow your listening instead of sharpening it. In that specific case, still gather the context, but skip the what's my ask step and let the conversation tell you what the ask should have been.

How do you stop rebuilding this prompt every week?

The template above works the first time you use it and gets annoying by the fifth time you retype the same five bracket labels. That's what a saved prompt library is actually for: not novelty prompts, but the boring recurring ones you'd otherwise rebuild from memory every single time. Save the template once, and every future meeting becomes filling in five blanks rather than writing five paragraphs of instructions from scratch. Our guide to building a personal prompt library covers the setup in more detail than fits here, if you haven't put one together yet.

The other piece worth automating is the material that doesn't change meeting to meeting: your own role, your product's name and what it actually does, the standard disclaimer you always need to include. That's what saved variables and a persistent context are for, reusable blocks the prompt pulls in automatically instead of facts you retype by hand on every single use. We've written separately about keeping context in every AI chat if that's the piece you're missing.

Put those two things together: a saved prompt template for the structure, saved variables for what doesn't change. Then the ten-minute routine actually holds up as a routine, instead of becoming one more thing you meant to keep doing.