Parrot: Open-Source AI Meeting Recorder and Live Copilot for macOS

Parrot is a free, open-source AI meeting recorder for macOS that enables real-time assistance during calls and local transcription without the need for a meeting bot to join the session. Unlike traditional AI note-takers that process audio in the cloud and provide summaries after the call, Parrot focuses on "live" utility by suggesting answers from a user's own documents while a conversation is still in progress.

Local-First Architecture and Privacy

Parrot is designed to keep audio and sensitive data on the local machine. It records what the Mac plays and hears, meaning it does not appear in the participant list of meeting apps like Zoom, Google Meet, or Microsoft Teams.

Data Residency Options

Users can choose between three primary privacy configurations:

  • Fully on Mac: Uses Whisper for transcription and Ollama for the copilot and reports. In this mode, nothing leaves the machine, and the app can function entirely offline.
  • With Own Keys: Users provide their own API keys for providers such as Claude, OpenAI, Gemini, Groq, Deepgram, or TypeSafe. Requests go directly from the Mac to the provider, bypassing Parrot servers.
  • Mixed Mode: A common configuration where transcription remains local (Whisper) while live assistance is powered by a cloud model (e.g., Claude).

Document Handling

Parrot allows users to upload PDFs, text files, or markdown documents to create a local knowledge base. These documents are chunked and embedded on the Mac using Apple's own language models. The system can perform cross-lingual retrieval, such as finding an answer in a Spanish document to answer an English question.

Real-Time "Live Copilot" and Meeting Profiles

The core differentiator of Parrot is its ability to provide "live nudges" and answer suggestions during a call.

Meeting Profiles

Users can select from seven built-in profiles or create custom ones to change how the AI monitors the call and writes the final report. For example:

  • Sales Discovery: Specifically watches for objections and buying signals.
  • Interviews: Looks for red flags and strong candidate answers.

Live Assistance

The copilot provides real-time tips over the call app. It can monitor timing and words to alert the user if a conversation drifts or if a participant has gone quiet after a specific point. These nudges are based on text and timing, not the voice tone of the participants.

Post-Call Analysis and Integration

Once a call ends, Parrot generates a report that includes a summary, action items, and coaching feedback. Every point in the report is linked to the specific line in the transcript for easy verification.

Speaker Detection

Parrot uses a 13 MB local model to distinguish between different voices. Users can name these voices once, and the app can remember voiceprints to suggest names in future calls. Integration with the calendar allows the app to suggest names based on the meeting invitees.

External AI Integration

Parrot integrates with Claude Desktop, Claude Code, Cursor, and Codex. This allows users to query their meeting history using their preferred AI interface (e.g., "What did I promise Northwind last week?"). This integration is read-only and must be explicitly enabled by the user.

Technical Specifications and Requirements

  • OS Requirements: macOS 14 or later on Apple Silicon (M1 or newer).
  • License: Open source under GPL-3.0.
  • Language Support: Auto-detects languages; uses Parakeet v3 for faster free transcription of 25 European languages.
  • Codebase: Approximately 24,000 lines of Swift with three dependencies.

Community Feedback and Insights

Early users on Hacker News have highlighted several points regarding the tool's utility and implementation:

"Tried this out... the audio it records has a echo. When I go to play back a speaker's voice it sounds I can hear what sounds like 2 audio streams, slightly offset."

Some users suggested renaming the "co-pilot" feature to "Coach" or "Assistant" to avoid confusion with Microsoft Copilot. Others requested the ability to attach entire folders to the knowledge base rather than individual files to streamline document management.

Sources

Related

  • Project
    moona3k/macparakeet

    MacParakeet is a native macOS app (and CLI) for offline, local‑first voice dictation, audio/video file transcription, and meeting recording on Apple Silicon. It runs several Core ML‑based speech‑to‑text models (Parakeet v3, v2, Unified) and optional local Nemotron, Cohere, or WhisperKit engines, keeping data private. Optional AI providers can be hooked for summaries or text rewriting. The stable DMG (0.8.7) is downloadable directly or via Homebrew, and the CLI enables automation of all features.

  • Dispatch
    Trace: Offline Mac Meeting Transcripts with Real-Time Flagging

    Trace is a privacy-focused Mac application that provides on-device meeting transcription and real-time moment flagging without using cloud servers or meeting bots.

  • Project
    fastrepl/anarlog

    A privacy-first, local-first AI meeting notepad that captures audio directly on-device to provide transcriptions and summaries without requiring a bot to join the call.

  • Project
    Laxcorp-Research/project-raven

    An open-source meeting copilot that provides real-time AI assistance and automated recaps using local storage and user-provided API keys for transcription and LLMs.

  • Project
    yazinsai/OpenOats

    A real-time meeting assistant for macOS that transcribes calls locally and surfaces relevant talking points from your personal notes using RAG.