Never walk into a meeting cold again

Walk in with context, and never manually file a follow-up again — Dori extracts and links it for you automatically.

For: consultants, founders, and business leaders running back-to-back client or team meetings

If you’re the one carrying every client relationship, every commitment, and every “what did we agree on” in your own head — this is for you.

Before a client call, you’re digging through old email threads and Slack DMs trying to remember what was agreed last time. After the call, notes live in a doc nobody — not even future you — will ever search again. Digging through old threads to reconstruct “what did we agree last time” can eat 10-15 minutes before a call has even started. A follow-up you promised gets buried under the next thing, and it slips.

How the meeting note builds itself

Connect a meeting source once, and every meeting from then on arrives with a summary, decisions, and action items pulled out automatically — already linked to the person and account it was with, filed under the right project without you doing anything, and searchable months later in plain language.

What’s actually driving this, and what it costs

Not everything here is an AI call. Matching attendees to people you already know is handled by matching logic against your vault, not an AI model — so it doesn’t cost anything per meeting and doesn’t depend on which provider you’ve connected. Filing a note under the right project works the same way.

The part that genuinely uses AI is turning the raw transcript into a summary and extracted decisions — that step runs on whichever AI provider you’ve connected, and it’s the one place a meeting actually costs something to process.

Questions you can ask afterward

This is what it’s actually for. Once even a handful of meetings have gone through Dori, you can ask things like:

• “What did we agree with Meridian last quarter?”
• “What’s still outstanding from my last call with Priya?”
• “Summarize every meeting I’ve had with Meridian this year.”
• “What did I promise to send after Tuesday’s call with Priya?”
• “Who else was on the call where we discussed pricing with Meridian?”

You ask these the same way you’d ask a colleague who happened to be in every meeting you’ve ever had — see Ask anything.

Quick setup (one time, about two minutes)

Your first meeting with Dori running

Say you’ve got a renewal call with Priya at Meridian.

Before the call. Open Priya’s page — it shows history, open threads, and follow-ups from last time, so you’re not reconstructing context from memory. Meridian’s linked account shows the commercial context alongside it: where the relationship stands, what’s been proposed. See relationships and Accounts.

During the call. Nothing to do. Your connected meeting source records as usual; Dori picks up the recording afterward.

After the call. A structured note appears with:

• A summary of what was discussed
• Decisions and action items pulled from the transcript
• Attendees matched to the people already in your vault — Priya is matched by name against people you already have, not guessed

That note is already linked to Priya, and to the Meridian account.

What this looks like after ten meetings, not one

This is where it compounds:

• Every meeting note is scoped to its project or account automatically — you never manually file one into the right place.
• Ask what you agreed with Priya six months ago and get the answer with the source note attached, instead of scrolling through old messages.
• Follow-ups from past meetings resurface as tasks instead of relying on you to remember them.

The mental load this removes isn’t the meeting itself — it’s everything you used to have to do afterward to make sure it wasn’t forgotten.

Go further

• Before a high-stakes meeting, get a second opinion on your talking points from Council.
• Want to draft the follow-up somewhere else — ChatGPT, Claude, wherever? Bundle the meeting into a Handoff instead of re-explaining the context by hand.
• Keep re-tagging these meetings the same way? That’s exactly the repeated correction Automations is built to catch and turn into a rule.