Meetily
Meetily records, transcribes, and summarizes meetings 100% on your device. Audio and data never leave. Private, secure, in your control.
Overview
Meetily is a privacy-first AI meeting assistant that records, transcribes, and summarizes meetings entirely on-device, so audio and data never leave the user's machine. It uses local Whisper/Parakeet models (claimed 4x faster than standard Whisper) for real-time transcription with automatic speaker diarization, then generates customizable AI summaries. It works bot-free across Google Meet, Zoom, Teams, Discord, Slack Huddles, and Webex by capturing system audio rather than joining as a visible bot, and can run fully offline after the model download. Developed by Zackriya Solutions, it's open-source (MIT) under a community tier, with a paid Pro tier adding team features.
The Verdict
Who Should Use Meetily?
Best For
- Privacy-conscious individuals and teams who don't want meeting audio sent to a cloud service
- Organizations with GDPR/HIPAA data-residency requirements
- Users wanting a bot-free recorder that doesn't show up as a visible participant
- Developers wanting MCP, webhook, or CLI integration into their own tooling
Not Ideal For
- Teams wanting zero-setup cloud convenience — local models require a download and local compute
- Non-technical users uncomfortable running a self-hosted/local tool (Community tier)
- Large teams needing a mature admin dashboard — Enterprise tier is custom-priced and less established than incumbents
What's Great
- 100% on-device processing — audio and transcripts never leave the device by default
- Free, open-source Community tier (MIT license) with no per-minute caps, unlike cloud competitors' capped free tiers
- Works across Google Meet, Zoom, Teams, Discord, Slack Huddles, and Webex via system audio capture, no visible bot
- Export to Markdown, PDF, DOCX; integrates via MCP, webhooks, and CLI
- Strong open-source adoption: 3,511 forks and an active contributor community (see badge for current star count)
Watch Out For
- Local processing means transcription speed depends on the user's own hardware, not a cloud server
- Linux build must be compiled from source; no prebuilt package listed
- Little public third-party review coverage (no significant Reddit or Hacker News discussion found)
- Bringing your own API key (e.g. Claude, OpenAI) for cloud-model summaries sends transcripts, not recordings, outside the device — worth confirming exactly what leaves if using that option
Pricing
View all features & details
Key Features
- On-device transcription (Whisper/Parakeet), ~4x faster than standard Whisper
- Real-time transcription with speaker diarization
- Customizable AI summary templates
- 10+ audio/video file import formats
- Multi-language support with auto-detection
Platforms
- macOS (.dmg), Windows (EXE)
- Linux (build from source)
- Works with any video conferencing app via system audio capture
Integrations
- MCP (Model Context Protocol)
- Webhooks
- CLI
- Export to Markdown, PDF, DOCX
Company
- Developed by Zackriya Solutions
- GitHub repo created December 2024
How It Compares
| Feature | Meetily | Otter.ai | Fireflies |
|---|---|---|---|
| Processing Location | 100% on-device | Cloud | Cloud |
| Free Tier Limit | No cap (Community) | 300 min/mo | 800 min/mo |
| Open Source | Yes (MIT) | No | No |
| Bot-Free Recording | Yes | No (visible bot) | No (visible bot) |
| Self-Hosted | Yes | No | No |
| Starting Paid Price | $10/user/mo | Varies | Varies |