Lookout
Your meetings, remembered — inside your Claude Code. Lookout unifies the notetakers you already use into one private memory you question over MCP, every answer traced to its source. Launching to design-partner teams first.
Works with Fireflies, Granola, and Otter— and the calendar you already keep.
Also works anywhere MCP does — Claude Desktop, Cursor, Windsurf, and any MCP-capable client.
Connected over OAuth, isolated per user, recalled on your own tokens.
From scattered notes to one queryable brain.
Connect your notetakers
Fireflies, Granola, Otter — bring the tools you already use. Lookout reads across all of them. No rip-and-replace, no new habit to learn.
We thread the memory
Every transcript becomes structured memory — decisions, commitments, people, dates — woven into one private knowledge graph that belongs to you.
Recall in Claude Code
Ask your Claude Code what was decided and when. Lookout answers over MCP — and anywhere else MCP is supported — with every claim traced back to the meeting it came from.
Two steps. Real commands. No dashboard.
Private beta. This is exactly what the connect flow does at launch — add the server, approve once in the browser, and your memory is a tool you can call.
01Add Lookout to Claude Code
Run this in your terminal. A browser opens once to authorize — that's the whole install.
claude mcp add lookout --transport http --url https://lookout.neurobaselabs.com/mcp
One-time browser consent (OAuth 2.1 + PKCE). Lookout gets a short-lived, scoped token — no API key ever touches your disk. Revoke in one click, anytime.
02Ask your memory
Your memory is now a tool. Ask Claude Code a question — it calls Lookout and answers from what your meetings actually said.
> What did we decide about the Northwind pricing tier?
-> lookout.recall("Northwind pricing tier — final decision")Recall runs on your own Claude tokens. We never see the conversation — every answer comes back traced to its source meeting, speaker, and date.
Ask a real question. Get the moment it was decided.
Left, you ask inside Claude Code. Right, Lookout answers from your own memory — with who said it, which meeting, and when.
> Remind me — what did we actually land on for Northwind's pricing tier, and who pushed for it? -> lookout.recall("Northwind pricing tier — final decision") matched 1 decision · 1 commitment · 1 source
Northwind goes on the Growth tier at $1,400/mo, billed annually — not the custom Enterprise quote. The team agreed the annual commit was worth trading ~15% off list to close before quarter-end.
Priya Nair (Head of Sales) proposed it; you approved on the call.
Priya to send the annual Growth-tier order form to Northwind by Fri, Jun 20 — marked done Jun 20.
Illustrative example. In beta, Lookout reads only the meetings you connect — nothing here is real customer data.
Built so you'd trust it with the room.
Your tokens, not ours
Recall runs inside your own Claude Code, on your own tokens. We serve the memory; we never see the conversation. The only spend on our side is ingesting your meetings — never your questions.
Isolated by design
One browser consent (OAuth 2.1 + PKCE), a short-lived scoped token, no key on disk. Every row is walled by per-user Row-Level Security at the database — so one account reaching another's memory isn't restricted, it's structurally impossible.
Read and write
Most memory tools only let you read. Lookout lets you edit — correct a decision, add a note, fix an attribution — straight from Claude Code, with a confirmation on every write. This one lets you keep it true.
Question your meetings like a colleague who never forgets.
What did we commit to in last week's review?
Answered with the exact source moments — never invented.
Every decision we've made about pricing, with sources.
Answered with the exact source moments — never invented.
Who owns the launch, and when did they agree to it?
Answered with the exact source moments — never invented.
Summarize everything said about the Acme account this month.
Answered with the exact source moments — never invented.
Your notes shouldn't live in someone else's walled garden.
They should live in your brain.
Questions founders ask.
Lookout by NeuroBase Labs is a vendor-neutral AI memory layer for founders. It unifies your meeting notes from Fireflies, Granola, and Otter into one private brain you recall and act on inside your Claude Code over MCP, so decisions, commitments, and context never get lost.
Lookout runs as an MCP server you connect to your Claude Code. Once connected, you can ask your Claude what was decided, who owns it, and when — and it answers from your own meeting history, with every answer traced back to its source moment.
Lookout is vendor-neutral: it reads across Fireflies, Granola, and Otter (more sources planned) so you get one memory across all of them instead of three separate silos. You keep the tools you already use.
No. Lookout sits on top of the notetakers you already use — there is no rip-and-replace. You keep your tools and gain one brain across them.
Yes. Your memory is yours — stored privately and never shared without your say. Lookout is built so the memory layer belongs to you, not the vendor, and every answer is grounded in your own sources.
Lookout is not another notetaker — it is the memory layer on top of the ones you have. Notetakers each remember their own calls in their own silo; Lookout threads all of them into one queryable brain you use from Claude Code.
Lookout is in private beta, opening first to design-partner teams. Request early access to get in — the connect command and MCP endpoint are already final, so nothing you set up now will change when we open more broadly.
Early-access pricing is $19/month, locked in for founding users — the rate you join at is the rate you keep, and it's free for design-partner teams during the beta. Recall runs on your own Claude tokens, so we never meter your questions; our billing only covers ingesting your meetings.
Get in before launch.
Private beta — free for design-partner teams. Tell us which notetakers you use and we'll bring you in as we open seats.