How to use AI for meeting prep: agendas, research, and follow-ups

Most people walk into meetings underprepared. Preparing properly takes 20-30 minutes, and when you have six meetings a day, that math doesn’t work. So you skim the calendar invite, open your laptop, and wing it.

AI changes that equation. With the right prompts, you can prepare for a meeting in under five minutes. Not superficial preparation, either: understanding who’s in the room, what was discussed last time, what you need to accomplish, and what you’ll say when the conversation goes sideways.

Researching attendees before a meeting

When you’re meeting someone for the first time (a potential client, a new vendor, a colleague from another department), knowing something about them changes how the conversation goes. You ask better questions. You spot common ground faster. You avoid awkward moments.

Before an external meeting:

“I have a meeting tomorrow with Sarah Chen, VP of Marketing at Meridian Health Systems. We’re discussing a potential consulting engagement around their digital marketing strategy. Based on what you know about typical VP of Marketing priorities at mid-size healthcare companies, help me prepare: (1) What challenges is she probably facing? (2) What questions should I ask to understand their situation? (3) What topics should I avoid or be careful about in healthcare marketing?”

For a meeting where you have LinkedIn information or a bio available, paste it in:

“Here’s the LinkedIn summary for the person I’m meeting with tomorrow. [paste summary] Based on their background, what are 3 good conversation starters that show I’ve done my homework? Also, are there any obvious connections between their experience and my work in [your field]?”

For internal meetings with people you don’t know well:

“I’m joining a cross-functional project team meeting for the first time. The team includes people from Engineering, Product, and Customer Success. I’m representing the Data team. What questions should I ask in the first meeting to understand the project without sounding like I haven’t read the brief? I want to seem prepared but also genuinely learn about what’s already been decided.”

A privacy note worth saying out loud: pasting someone’s LinkedIn summary, a job-application bio, or any third-party personal information into a public AI tool is the kind of thing some employers’ AI policies forbid. Check your org’s policy before you do this in a work context. Pasting a publicly-visible bio into a conversational tool isn’t doing anything dramatic, but the policy exists, and being on the right side of it is the freebie.

Drafting agendas that keep meetings on track

Meetings without agendas wander. Everyone knows this, and almost nobody writes agendas because it feels like busywork. AI makes it take about two minutes.

For a recurring team meeting:

“I run a weekly 30-minute team standup for a marketing team of 6 people. This week’s topics should include: Q2 campaign planning status, the website redesign timeline (it’s behind schedule), and a new request from Sales for case study content. Draft an agenda with time allocations. The campaign planning needs the most time. Include a 3-minute slot at the end for anything else.”

For a one-on-one with your manager:

“I have a 1:1 with my manager on Thursday. Things I want to cover: I’d like feedback on the presentation I gave last week, I want to discuss moving up the timeline on Project Atlas, and I need to flag that I’m feeling stretched too thin across three projects. Draft a brief agenda I can share in advance. Keep it professional but not overly formal.”

For a client meeting where you need to deliver a progress update:

“Draft an agenda for a 45-minute client status meeting. The project is a website redesign, we’re in week 4 of 8. Topics: show the homepage design (approved last week, now built out), discuss the product page layout options (I’ll present 2 directions), and align on the content migration plan (they need to provide copy by April 10). I also need to tactfully bring up that they haven’t given feedback on the sitemap I sent 10 days ago.”

The value of sending agendas in advance goes beyond organization. It signals respect for other people’s time, sets expectations, and gives quieter team members a chance to prepare their thoughts. AI removes the friction that stops most people from doing it.

Summarizing previous meeting notes

You had a great meeting two weeks ago. Decisions were made. Action items were assigned. Now you’re about to meet with the same group and you can barely remember what was discussed. Sound familiar?

If you have notes (even messy ones), paste them in:

“Here are my rough notes from last Tuesday’s project meeting. They’re disorganized because I was typing fast. Clean these up into: (1) a brief summary of what was discussed (3-4 sentences), (2) decisions that were made, (3) action items with who’s responsible, and (4) open questions that still need answers. Here are the notes: [paste notes]”

If you recorded the meeting and have a transcript (Zoom and Google Meet now have native transcription, and standalone tools like Otter, Fireflies, and Notion AI work too):

“Here’s the transcript from our 45-minute product planning meeting. Summarize it into: key decisions, action items with owners, topics where we didn’t reach agreement, and anything that needs follow-up before next meeting. Keep the summary under 300 words.”

For preparing to reference a previous meeting in the current one:

“I’m about to go into a meeting with the same client I met two weeks ago. Here are my notes from that meeting. [paste notes] Give me a quick refresher: what did we agree to, what was each person supposed to do before this meeting, and what questions were left open? Also flag anything that seems like it might have fallen through the cracks.”

This last prompt is particularly useful because it helps you spot things that were promised but probably weren’t done. Walking into a meeting knowing that the client was supposed to send you their content by last Friday (and didn’t) puts you in a much better position than realizing it mid-conversation.

Generating talking points

Sometimes you need to present an idea, make a case for something, or lead a discussion, and you know what you think but haven’t organized it yet. AI is good at turning scattered thoughts into structured talking points.

For presenting a recommendation to leadership:

“I want to propose that our team switches from Jira to Linear for project management. My reasons: it’s faster, the UX is better for our developers, and we’d save about $400/month. The likely objection is migration effort and the learning curve. Give me 5-6 talking points I can use in a 10-minute presentation to my director. Focus on business impact, not personal preference.”

For leading a discussion on a sensitive topic:

“I need to lead a team discussion about why we missed our Q1 targets. Revenue was 15% below goal. The main reasons were: we lost two big deals in March, our sales cycle got longer, and we had some staffing gaps. I want to be honest about what happened without making it feel like a blame session. Give me talking points that acknowledge the miss, identify root causes, and pivot toward what we’re changing for Q2.”

For contributing to a meeting where you’re not the lead:

“I’m attending a budget review meeting tomorrow. My department is requesting a 12% increase to hire two more engineers. I need to justify this to the CFO in about 3 minutes. Our current team of 4 engineers is handling 30% more support tickets than last year, and response times have doubled. Give me concise talking points that connect headcount to business outcomes, not just workload.”

Writing follow-up emails and action items

The meeting ended. Everyone agreed on next steps. By tomorrow, half the room will have a different recollection of what was decided. A follow-up email fixes this, and AI writes them in about 90 seconds.

Immediately after a meeting, while it’s fresh:

“I just finished a client meeting. Write a follow-up email summarizing what we discussed and listing action items. Here’s what happened: We agreed to go with Design Option B for the homepage. The client will send final copy for the About page by Friday. I’ll have the product page mockups ready for review by next Wednesday. We’re meeting again in two weeks. Next meeting we’ll review the product pages and discuss the checkout flow. Keep it concise and professional.”

For after a job interview or networking meeting:

“I just had a 30-minute informational interview with a product manager at Stripe. We talked about their approach to platform partnerships and she mentioned they’re expanding their developer relations team. She recommended I connect with her colleague Marcus on LinkedIn. Write a thank-you email I can send this afternoon. Reference something specific from our conversation so it doesn’t feel generic. Keep it brief.”

For after a meeting where disagreements need to be documented carefully:

“Write a follow-up email for a meeting where we didn’t reach full agreement. The marketing team wants to launch the campaign on May 1. The engineering team says the landing page won’t be ready until May 8. We compromised: marketing will start with email-only on May 1 and add the landing page when it’s ready. The engineering lead, Tom, still has concerns about rushing. Summarize this diplomatically, making sure both perspectives are acknowledged and the compromise is clearly stated.”

The follow-up email is where most meeting value gets lost. People leave the room aligned and slowly diverge because nobody wrote down what was agreed. Spending two minutes with AI after every important meeting solves this problem. For the writing-craft side of follow-ups specifically, see writing and communication.

A scope note about this article: it covers before and after the meeting, not during. AI in real time inside a meeting is a different problem with different tools (live transcription, live translation, real-time prompt-helpers), and the choice is more environment-dependent (do you have a meeting bot? are your participants OK with that?). The before/after pattern works in any context.

What to do with the eight minutes you save

The math is simple: about five minutes before each meeting (review the agenda, summarize last time’s notes, write down what you want out of this one) plus three minutes after (clean up the transcript, send the follow-up email). Eight minutes that prevent the slow drift between “we agreed on X” and “wait, did we?” by the next meeting.

The genuine insight, if you take only one prompt from this whole article, is the missed-commitment scan. Asking AI to flag what was promised but probably didn’t happen, before you walk into the next meeting with the same group, is the single most useful prompt for anyone who runs ongoing client or project work. The agenda and the attendee research help. The follow-up email helps. But meetings that go sideways usually go sideways because something fell through the cracks two weeks ago, and finding that thing before it’s a surprise is the whole game.

If you’re new to AI in general, organization and productivity covers the broader daily-rhythm angle, and tips for better results covers prompt-craft specifically.