Most people walk into meetings cold. They glance at the calendar invite, see a name they half-recognize, and spend the first five minutes piecing together context. AI-powered meeting prep eliminates that entirely. Instead of scrambling, you arrive knowing who you're talking to, what you've discussed before, and what matters right now.
This guide covers everything: how AI meeting prep works under the hood, what gets researched, the privacy model that keeps your data safe, and how to use it across every type of meeting you have.
What Is AI Meeting Prep?
AI meeting prep is the process of using a large language model to automatically research your upcoming meetings and generate a concise briefing before you join. If you're new to the concept, our introduction to AI meeting prep covers the basics.
Traditional meeting prep is manual. You open LinkedIn, search for someone's name, skim their profile, dig through your inbox for past threads, and maybe check Slack for recent messages. It works, but it takes 10-15 minutes per meeting. When you have six meetings a day, that's an hour and a half of research — or, more realistically, you skip the prep entirely.
AI meeting prep compresses that entire process into seconds. One hour before your meeting, the system pulls attendee information, scans your communication history, and synthesizes everything into a briefing card that appears alongside your full-screen reminder.
How It Works, Step by Step
The pipeline runs automatically in the background. Here's what happens, in order:
- Event detection. Your calendar is monitored for upcoming meetings. One hour before a meeting starts, the research pipeline kicks off.
- Attendee extraction. The system pulls every participant from the calendar invite — names, email addresses, and domains.
- Background research. For each attendee, the AI searches for professional background: current role, company, relevant history. Internal team members are flagged so the briefing focuses on what's useful.
- Communication history. Your email threads and Slack conversations with each attendee are scanned for recent context. This surfaces what you last discussed, any open threads, and relevant decisions.
- Document context. If you use Notion for meeting notes, relevant pages and databases are pulled in to provide additional context about projects and prior decisions.
- Synthesis. Everything feeds into the LLM, which generates a concise briefing: who each person is, what you've discussed recently, and suggested talking points.
- Delivery. The briefing appears on your reminder card. When the full-screen reminder fires, the prep is right there — no separate app, no extra clicks.
What Gets Researched
The quality of a briefing depends entirely on the inputs. AI meeting prep pulls from four primary sources:
Attendee backgrounds. For external attendees, the AI researches their professional profile — current title, company, and relevant career history. For internal team members, the system notes their role without the external research layer. Every attendee name links to their LinkedIn profile so you can dig deeper if needed.
Email history. Recent email threads between you and each attendee are scanned for context. The AI identifies the most relevant conversations — not just the most recent. If you exchanged pricing details three weeks ago and small talk yesterday, the briefing highlights the pricing thread.
Slack threads. Slack conversations add a layer of context that email often misses. Quick decisions, informal agreements, and project updates that live in channels and DMs get surfaced. See our deep dive on Slack context in meetings for how this works.
Notion documents. Meeting notes, project briefs, and shared documents from Notion provide institutional context. If your team tracks decisions or action items in Notion, those feed directly into the briefing. Our guide to Notion meeting context covers setup and best practices.
The BYOK Model: Bring Your Own API Key
One of the most important architectural decisions in AI meeting prep is who controls — and pays for — the LLM. Remind uses a BYOK (Bring Your Own Key) model, which means you connect your own API key from whichever provider you prefer.
Three providers are supported:
- Claude (Anthropic) — strong at nuanced synthesis and long-context analysis
- OpenAI (GPT-4 and later) — widely used, reliable for structured summaries
- Gemini (Google) — good integration with Google ecosystem context
The BYOK model has three advantages. First, cost transparency: you see exactly what each briefing costs on your own API dashboard — typically a few cents per meeting. Second, no vendor lock-in: if one provider raises prices or degrades quality, you switch keys in Settings and move on. Third, privacy: your meeting data flows directly from your machine to your chosen LLM provider. Remind never stores your API key on a server and never proxies your meeting content through its own infrastructure.
Privacy Considerations
Meeting data is sensitive. Attendee names, email threads, and internal Slack conversations are not the kind of information you want sitting on someone else's server.
With the BYOK model, the data flow is straightforward: your Mac sends the research context directly to your chosen LLM provider, which processes it and returns the briefing. Remind itself never sees the content of your meetings, emails, or messages. Your API key is stored locally in the macOS Keychain, encrypted at rest by the operating system.
Calendar and email access runs through Nylas, which acts as a secure relay for reading calendar events and email threads. Nylas is SOC 2 Type II certified and processes data in transit — it doesn't store email bodies or calendar details long-term.
The practical result: your meeting prep data touches exactly two services (your calendar/email provider and your LLM provider), both of which you've already chosen to trust.
Use Cases
AI meeting prep works differently depending on the meeting type. Here's how each plays out:
Sales calls. The briefing surfaces the prospect's role, company size, and any prior conversations. If you sent a proposal last week and they replied with questions, those questions appear in the briefing. You walk in ready to address objections instead of asking "where did we leave off?"
Investor meetings. Background research pulls the investor's portfolio, recent deals, and fund focus. Email history highlights prior term sheet discussions or intro threads. You spend less time on context-setting and more time on the pitch.
Client check-ins. These recurring meetings are where context decay hits hardest. What did you promise last month? What blockers did they mention? The briefing synthesizes your email and Slack history into a running summary, so every check-in builds on the last one instead of starting from zero.
Internal 1:1s. Even meetings with your own team benefit from prep. The briefing surfaces recent Slack threads, shared project updates, and any email follow-ups. Manager 1:1s in particular benefit — you arrive with specific topics rather than a vague "anything on your mind?"
No-attendee meetings. Blocked time for project work or solo focus sessions still get useful prep. When there are no attendees to research, the AI pivots to researching the meeting topic itself — pulling context about the project, recent decisions, and relevant background to help you hit the ground running.
How Remind Implements It
Remind is a native macOS menu bar app that ties calendar monitoring, AI research, and full-screen reminders into a single pipeline. Here's the specific implementation:
One hour before a meeting, the research pipeline triggers automatically. Attendee information is extracted from the calendar event. For each attendee, the app runs background research using your configured LLM, pulls relevant email threads via Nylas, and searches Slack for recent conversations. All of this feeds into a single LLM call that generates the briefing.
The briefing is cached locally and attached to the meeting. When the full-screen reminder fires — typically 2-3 minutes before the meeting — the briefing card is right there alongside the Join button. One glance and you know who you're talking to, what you've discussed, and what to bring up.
You can also trigger research manually. Click any meeting in the menu bar dropdown to generate or refresh a briefing on demand. If a meeting was added last-minute and the one-hour trigger already passed, manual prep has you covered. For a broader look at how meeting reminders work on macOS, see our complete guide to meeting reminders on Mac.
Manual Prep vs. Automated Prep
The difference isn't just speed — it's consistency. Here's how they compare:
| Manual Prep | AI-Powered Prep | |
|---|---|---|
| Time per meeting | 10-15 minutes | ~5 seconds |
| Consistency | Skipped when busy | Every meeting, automatically |
| Sources checked | 1-2 (usually LinkedIn + inbox) | 4+ (LinkedIn, email, Slack, Notion) |
| Historical context | Whatever you remember | Full thread history surfaced |
| Cost | Your time (~$50-100/hr equivalent) | ~$0.02-0.05 per briefing |
| Delivery | Separate tabs and apps | On the reminder card itself |
Manual prep is better than no prep. But the reality is that most people only manually prepare for high-stakes meetings — the investor pitch, the big client call. The everyday 1:1s, the recurring syncs, the ad-hoc catch-ups? Those get nothing. AI-powered prep closes that gap by making every meeting a prepared meeting — so you can stop showing up cold for good.
Getting Started
Setting up AI meeting prep in Remind takes about two minutes:
- Download Remind and connect your calendar. Remind supports Google and Microsoft calendars through Nylas — one OAuth flow and you're connected.
- Add your LLM API key. Open Settings, choose your provider (Claude, OpenAI, or Gemini), and paste your key. The key is stored in your macOS Keychain and never leaves your machine.
- Connect Slack (optional). One click to authorize your workspace. Slack context will be included in every briefing where you have relevant conversations with attendees.
- Wait for your next meeting. One hour before it starts, you'll see the research indicator in the menu bar. When the full-screen reminder fires, your briefing will be there.
That's it. No configuration files, no prompt engineering, no training period. The system works with whatever's on your calendar, using whatever LLM you prefer, pulling from whatever communication channels you've connected.
AI meeting prep is just one piece of a well-tuned Mac workflow. For more tools that complement your meeting game, see our list of the best Mac productivity apps for 2026.
The best meetings are the ones where everyone shows up prepared. AI meeting prep makes that the default instead of the exception.