coco transcribes recordings, prepares you for the next meeting, compares what was said across meetings, and keeps a memory of people, projects and commitments that you can search. It runs on your computer and uses an AI account you already have.
Record in the browser or upload audio and video. Whisper transcribes on this computer; computers that cannot run it well use Gemini through the Antigravity CLI, the Gemini API, an OpenAI-compatible API or Alibaba Cloud Bailian. Transcripts in txt, docx, pdf, srt, vtt or json go straight in.
Nine templates, from minutes and action items to a follow-up draft. Reports start with the conclusion, state each fact once and keep to a length set by the meeting.
Before a meeting, a brief drawn from past meetings and memory. Across meetings, three modes: commitments and changed positions, signals beneath the surface, and a synthesis of several interviews.
Every meeting adds to a memory of people, projects, commitments and terms. Each entry has its current state and dated facts linked to the meeting they came from. Search it, correct it, or ask about one person across all meetings.
The screens show an invented demo library.
Open Terminal, paste the line and press Return. coco then appears in Applications.
curl -fsSL https://raw.githubusercontent.com/cocohahaha/coco/main/scripts/install.sh | bash
Open PowerShell, paste the line and press Enter. coco then appears in the Start menu and on the desktop.
irm https://raw.githubusercontent.com/cocohahaha/coco/main/scripts/install.ps1 | iex
Download the ZIP from GitHub, unzip it and double-click coco.command on a Mac or coco.bat on Windows. Python and the dependencies are set up on first start.