Inspiration

We forget meaningful details from daily meetings. We wanted a "SuperHuman Memory" that doesn't just record, but understands who you meet and how they connect to your professional network.

What it does

SuperHuman is like a second brain for your meetings. It sits quietly in the background, automatically detecting and recording your calls. Instead of just giving you a transcript, it intelligently analyzes the conversation to figure out who you met, what they're good at, and how they're connected to everyone else you know. It turns forgotten conversations into a searchable map of your professional life.

How we built it

MacOS: Swift & ScreenCaptureKit for high-fidelity audio (system + mic). AI: ElevenLabs for transcription; GPT-4o for entity extraction. Backend: Python & FastAPI. Database: Neo4j Graph Database. Challenges we ran into Merging microphone and system audio streams on macOS without drift was technically difficult. Also, performing entity resolution (knowing that "Sarah" and "Sarah Chen" are the same person) within the graph required complex logic.

Accomplishments that we're proud of

Building a truly "zero-configuration" experience—you simply join a call, click "start recording" and the agent handles the rest.

What's next for SuperHuman

Real-time "flashcards" that appear during meetings (e.g., "You last spoke to Sarah about Python 3 months ago"), and a voice interface to query your digital brain

Built With

Share this project:

Updates