Elevator Pitch

Meeting Memory is an AI that sits in your video calls, takes notes, and links them to your projects automatically. No bot in the room. No "Meeting Memory is now recording." Just a recap afterward that knows what you were working on.

Core Problem

Solo entrepreneurs and small teams lose 30% of what's said in meetings. Action items fall through cracks. Decisions get made and then forgotten. Current solutions (Fireflies, Otter, Fathom) require a bot to join the call, which changes the dynamic — people self-censor, and the bot becomes a character in the room.

Key Insight

The best meeting recorder is the one nobody notices. Meeting Memory works from the transcription pipeline — it processes the audio stream locally, extracts topics, and links them to your project knowledge base. It doesn't just transcribe — it connects.

Target Users

Initially solo founders and 2-5 person teams who have 10-20 meetings a week. They're evaluating user personas like "Alex the solo founder" who has 15 meetings a week and forgets what was agreed.

Business Model

Freemium: free tier (5 meetings/month, basic transcription), pro ($29/mo — unlimited meetings, topic linking, project integration), team ($79/mo — shared knowledge base, team analytics). See pricing model for full breakdown.

Competitive Landscape

Fireflies, Otter, Fathom, and Granola all require a bot. None of them link meeting content to your existing notes and projects. That's the gap. See competitive analysis for the full matrix.

"The best meeting note-taker is the one you don't have to think about."

Architecture

We use a local Whisper model for transcription, then a small LLM for topic extraction and entity linking. The architecture decisions doc covers why we chose local-first over cloud. The transcription pipeline doc covers the audio processing chain.