Granola raises $43M Series B to build AI-powered team intelligence layer on top of meetings
May 14, 2025 with Chris Pedregal
Key Points
- Granola raises $43M Series B to expand from individual meeting notes into a shared team intelligence layer that lets employees query across all colleagues' transcripts and reasoning over meeting context.
- The startup uses large context windows with Claude 3.5 Sonnet in reasoning mode to answer comparative and analytical queries where traditional retrieval-augmented generation breaks down.
- Granola plans to move beyond meetings into autonomous agents handling email, Slack, and Docs workflows, positioning transcription as a wedge into a broader contextual AI platform for knowledge work.
Summary
Granola, the London-based AI meeting intelligence startup, has raised a $43M Series B with participation from Lightseed and Nat Friedman, among others. The company launched less than a year ago with 18 people and is now transcribing millions of minutes of meetings per day.
From note-taker to team intelligence layer
Granola started as a Mac app that lets users take freeform notes during meetings while the product listens in the background and rewrites those notes into something polished when the meeting ends. Simple, minimal, deliberately Apple Notes-like. That product spread entirely through word of mouth — no built-in growth loops, no multiplayer features until now.
The Series B is funding the next move: shared intelligence across teams. Users can now query across their entire team's meeting history, not just their own. The pitch is that granola becomes the context layer a team actually uses — ask it what's on fire, how you performed in a sales call compared to teammates, or what a specific person said three weeks ago.
RAG works for lookup but breaks on analysis questions. For queries like "how could I improve my one-on-ones" or "how do my answers compare to my team's," Granola stuffs large context windows with the right transcripts instead. The default model is Claude 3.7 Sonnet in reasoning mode, though as of today users can choose their own.
Notion and the competitive frame
Notion launched what many described as a Granola clone the day before this conversation. CEO Chris says the Notion founder DMed him on Twitter ahead of the launch — a gesture he read as mutual respect rather than aggression. His read on the broader competitive picture is that transcription has always been a commodity and incumbents will bolt it on everywhere. The question is who builds the product that actually reasons over that context in ways that feel native to the AI era.
His analogy: the jump from web to mobile forced entirely new product paradigms, and the jump from pre-AI to post-AI is larger still. Companies born in the new medium tend to win it — though he acknowledges large incumbents can add capable AI features too.
The longer vision
Meeting notes were never the destination. The original thesis was a contextually aware, AI-powered tool for thought. Meetings turned out to be an exceptionally strong wedge because conversations contain dense, undercaptured information.
The roadmap extends into agentic action: email, Slack, Google Docs, and eventually agents that execute post-meeting work autonomously, leaving humans to apply judgment rather than process output. Chris frames the vision as AI doing all the busy work so that the only thing left for people is the call.
On the $43M specifically: hiring is the primary use. Transcription costs more than LLM inference at current scale, but both are covered by subscription revenue. The team is based in London and plans to open US offices, though locations are not yet confirmed.