Meetily

About Meetily

Free, MIT-licensed desktop meeting note taker for Windows and macOS. It records your microphone and system audio, transcribes on your own machine with Parakeet or Whisper models, and writes summaries with a built-in local model, a local Ollama server or a cloud provider you choose.

Best for: People who want meeting transcripts and summaries without a meeting bot or a cloud notetaker, and who will check a summary against the transcript before acting on it.

Deployment: Desktop installer for Windows (x64) and macOS (Apple silicon) from the project's GitHub releases. The first run downloads the Parakeet transcription model and a built-in summary model, about 3.2 GB together. A local Ollama server can be chosen for summaries instead of the built-in model.

Skill level: Beginner

Tradeoffs

Summaries are the weak point. In our November 2026 test every summary, on the built-in model and on Ollama, had something wrong or missing: an invented date, the wrong owner for an action item, or a quote stamped with the wrong segment time, and nothing in the app warns about it. During live recording the view held text back while speech continued, and the saved transcript of one run lost its opening sentences, which Retranscribe recovered from the saved audio. The privacy promise holds only while summaries stay on a local provider. The Community Edition is MIT; speaker identification, custom templates and integrations are in the paid PRO edition.

Related guides and resources

Explore step-by-step setup guides, comparisons, and stack recipes for this tool category.

Best for

People who want meeting transcripts and summaries without a meeting bot or a cloud notetaker, and who will check a summary against the transcript before acting on it.

Why use it

Meetily's free Community Edition does the whole job on your computer: recording, transcription with NVIDIA's Parakeet model by default, and summaries with a bundled local model or a local Ollama server. We installed version 0.4.1 on a Windows desktop in November 2026 and ran a clean two-minute test recording through it. The imported transcript was accurate apart from how numbers were written, and we saw no outside network connection while it transcribed, summarized or recorded, only during the first-run model download and one connection to a GitHub address after setup. Recordings and imported files are kept in a folder under your Music directory.

Key features

• Records microphone and system audio together, with a reminder to tell participants the meeting is being recorded
• On-device transcription with Parakeet by default, with optional Whisper models
• Imports existing audio files, and Retranscribe re-runs a saved recording with a different model
• Summaries from a built-in local model, a local Ollama server or a custom OpenAI-compatible server, with Claude, Groq, OpenAI and OpenRouter also offered
• Usage analytics off by default

Common AI use cases

• Private notes for internal meetings and calls
• Transcribing recorded interviews and other audio files
• Drafting meeting summaries with a local model through Ollama

Who should use it

• Individuals and small teams who want meeting notes that stay on their own computer
• Ollama users who want a meeting front end for a model they already run
• Anyone replacing a cloud notetaker who is willing to read the transcript

Who should not use it

• Teams that need speaker labels in the free edition: speaker identification is part of the paid Meetily PRO
• Anyone who would act on a generated summary without reading the transcript
• People who need a readable live transcript during a fast, continuous conversation

Evaluation checklist

• Keep summaries on Built-in AI or Ollama if nothing should leave the machine, because the same list offers cloud providers
• Check every action item's owner, date and timestamp against the transcript
• After an important live recording, run Retranscribe on the saved audio
• Check disk space and bandwidth before the first run, which downloads about 3.2 GB of models without asking

Tradeoffs

Summaries are the weak point. In our November 2026 test every summary, on the built-in model and on Ollama, had something wrong or missing: an invented date, the wrong owner for an action item, or a quote stamped with the wrong segment time, and nothing in the app warns about it. During live recording the view held text back while speech continued, and the saved transcript of one run lost its opening sentences, which Retranscribe recovered from the saved audio. The privacy promise holds only while summaries stay on a local provider. The Community Edition is MIT; speaker identification, custom templates and integrations are in the paid PRO edition.

Pros

• Accurate on-device transcription of clean audio, quick even on a CPU
• No outside connections during transcription, summaries or recording in our test, and analytics off by default
• Works with a local Ollama server, which runs summaries on your GPU
• Retranscribe recovers words a live recording missed

Cons

• Generated summaries can state wrong owners, dates and timestamps with no warning
• The live view can show almost nothing for minutes, and a live recording can lose its opening words
• No speaker labels in the free edition
• About 3.2 GB of models downloaded on first run with no choice shown
• Cloud summary providers sit in the same list as the local ones

Alternatives

• MeetGeek