Have your AI prep you for your next meeting
Walk in knowing where you left off. Your AI reads the invite, pulls your recent email and Teams with each attendee, and hands you a catch-up brief — and when it's a first meeting with someone new, it says so, so you're not scrambling for history that doesn't exist.
Who it’s for
Anyone with back-to-back meetings who has ever opened a call and thought “wait, where did we leave this?” — founders, account managers, consultants, anyone juggling threads across email and Teams.
The pain
The context you need is scattered — an email thread here, a Teams chat there, a calendar note somewhere. Reconstructing it in the two minutes before a call is stressful and easy to get wrong.
The recipe
In your AI (Claude, ChatGPT, or Cursor) with LMCP connected, paste:
“Before my next meeting, catch me up on the attendees — pull my recent email and Teams with each of them and flag anything still open. If I have no history with someone, just tell me it’s a first meeting.”
What the agent does
- Reads the invite — the meeting, time and attendees —
list_calendar_events - Pulls recent context for each attendee —
search_emails+teams_read_chat_messages(andsearch_contactsto identify them) - Branches on what it finds (see below) and hands you a short brief
Two kinds of prep — the agent picks the right one
1. You have history → a catch-up brief
When there’s prior email or Teams with the attendee, the agent summarizes where you left off: the last exchange, decisions made, and anything still open — so you continue instead of repeating.
2. First meeting / external contact → “this is a fresh start”
When the person isn’t in your contacts and there’s no email or Teams history — a new external contact — the agent doesn’t hand you an empty result and let you assume you forgot something. It tells you plainly: this is a first meeting, no prior context, and pivots to what’s actually useful — who they are (from the invite, their domain/company) and what to walk in with. So you prep to introduce, not to follow up.
What you see
- A short brief per attendee: caught-up context, or “first meeting — here’s who they are.”
- Open items flagged, so nothing that needs a decision slips.
Why it works so well with LMCP
- Both email and Teams, locally: the context lives across Mail and the Teams desktop cache — LMCP reads both on-device, no Graph API or tokens.
- Knows the difference: internal colleagues (rich history) vs. external first-contacts (none) get different, correct prep — no misleading empty results.
- On your Mac: your inbox and chats never leave the machine.
Variations
- “…also check Slack and WhatsApp with them” → add
slack_search_messages+whatsapp_search_messages. - “…draft talking points from the open items” → the agent turns the flagged items into an agenda.
- “…prep me for everything on my calendar tomorrow” → a brief per meeting, first thing in the morning.