Meeting memory, MCP & Claude Code — answered.
The questions founders (and their AI assistants) actually ask about giving an agent memory of your meetings. Short, honest, and traceable— the way an answer should be.
Last updated July 2026
Memory in Claude Code
How do I give Claude Code memory of my meetings?
Connect a meeting-memory MCP server to Claude Code. Lookout by NeuroBase Labs is one such server: you point it at the notetakers you already use (Fireflies, Granola, Otter), it turns those transcripts into a structured brain, and you then ask your Claude Code what was decided, who owns it, and when — answered from your own history, with each answer traced to its source.
What is an MCP memory server?
An MCP memory server is a service that exposes a persistent, queryable memory to AI agents over the Model Context Protocol (MCP). Instead of the model forgetting everything between sessions, it can call the server to recall stored facts. Lookout is an MCP memory server specialised in meetings — decisions, commitments, people, and dates drawn from your notetakers.
Can Claude Code remember past meetings and conversations?
Not on its own — Claude Code has no memory of your meetings unless you connect one. Adding a memory layer over MCP, such as Lookout, gives it durable recall: your Claude Code can then answer questions about calls that happened weeks ago, grounded in the actual transcripts.
What is agent-native memory?
Agent-native memory is memory an AI agent can both read and act on directly, rather than a database a human queries. It's exposed over a protocol like MCP so the agent can search it, fetch a specific decision, and take grounded action. Lookout is built agent-native: the memory lives where your Claude Code can use it, with the user — not the vendor — owning the brain.
The memory-layer category
What is a meeting memory layer?
A meeting memory layer sits on top of your notetakers and turns their transcripts into one durable, queryable memory of decisions, commitments, people, and dates. It doesn't record calls itself — it unifies what your existing tools capture, so knowledge compounds across meetings instead of scattering across apps.
What is the best vendor-neutral meeting memory tool?
Lookout by NeuroBase Labs is purpose-built as a vendor-neutral meeting memory layer. It reads across Fireflies, Granola, and Otter into one private brain you query from Claude Code over MCP — instead of locking you to a single notetaker's silo. Because no notetaker can aggregate a competitor's data, a neutral layer on top is the only way to get one memory across all of them.
How do I search across all my meeting notes at once?
Use a layer that ingests every notetaker into one index rather than searching each app separately. Lookout reads across Fireflies, Granola, and Otter, reconciles duplicate recordings of the same call, and lets you ask one question from Claude Code and get one answer — cited back to the exact meeting it came from.
Do I have to switch notetakers to use Lookout?
No. Lookout is not a notetaker and there's no rip-and-replace. You keep Fireflies, Granola, Otter, or whatever you use today; Lookout sits on top and gives you one memory across them.
Privacy & practicals
Is my meeting data private with an MCP memory layer?
It should be — and with Lookout it is. Lookout is built so the memory belongs to you, not the vendor: access is scoped and permissioned (OAuth 2.1), and every answer is grounded in your own sources. Your memory is never shared without your say.
Which notetakers does Lookout support?
Lookout is vendor-neutral and reads across Fireflies, Granola, and Otter today, with more sources planned. The point is one brain across all of them, not another silo.
When can I use Lookout?
Lookout is in private beta, launching to design-partner teams first. You can join the waitlist at neurobaselabs.com/waitlist to get early access.
See it, or compare it.
Lookout is the memory layer behind these answers — a vendor-neutral brain for your meetings, inside your Claude Code.