Yogurt: A Private, Open-Source Alternative to Granola for Meeting Notes

Overview

Yogurt is an open-source, privacy-first desktop application designed to capture and summarize meeting notes locally. Unlike many existing meeting assistants that require a third-party bot to join a call, Yogurt operates entirely on the user's machine, ensuring that sensitive audio and transcript data remain private. It is built as a cross-platform tool, supporting macOS, Windows, and Linux.

Core Functionality and Privacy

The primary value proposition of Yogurt is the elimination of "meeting bots." By recording audio directly from the system, it avoids the common friction of needing to invite an external participant to a call, which often triggers privacy concerns or meeting recording notifications. Because the application processes audio and generates summaries locally, it addresses the security risks associated with cloud-based transcription services where data may be stored or processed on third-party servers.

Technical Implementation

Built using Electron, Yogurt leverages local AI models to perform its tasks. The application utilizes llama-cpp and whisper for speech-to-text transcription and summarization. This local-first approach allows users to maintain control over their data, though it does trade off some system resources, as local model processing can be memory-intensive compared to cloud-based alternatives.

Community Feedback and Landscape

The release of Yogurt has sparked significant discussion within the developer community regarding the crowded landscape of local AI meeting tools. Users have noted that while the space is becoming saturated with similar projects, there is a clear demand for alternatives to proprietary services like Granola.

Notable Insights from the Community

  • Resource Management: Some users expressed concern regarding the use of Electron, suggesting that native frameworks might offer better performance. Others requested the ability to offload processing to services like OpenRouter to save local compute resources.
  • Diarization: A recurring theme in the feedback is the desire for robust speaker diarization (identifying who said what). While some users have successfully integrated their own solutions, it remains a highly requested feature for these types of tools.
  • Alternative Projects: The community highlighted several other open-source or local-first projects that share similar goals, including:
    • Muesli and OpenOats: Frequently cited as similar alternatives in the open-source space.
    • Yapsnap: Noted for its ability to perform diarization on CPU.
    • Biscotti: Mentioned as a native macOS alternative that avoids Electron.
    • OpenWhispr: Recognized as one of the earlier projects in this domain.

User Perspectives

Many contributors expressed appreciation for the project's transparency and privacy stance. As one user noted:

"I live in a two party consent state and have yet to pull the trigger on an AI transcription service. Yogurt is the first appropriate implementation of the service in my opinion, both for personal privacy and to limit potential issues if Granola or similar service gets hacked and transcriptions go public."

Conversely, some developers emphasized the difficulty of maintaining such projects, noting that the "Mac-only" ecosystem is currently saturated with tools, and that building for the non-Apple Silicon space is a significant, albeit necessary, challenge.

Sources

Related

  • Project
  • Project
  • Dispatch
  • Project
  • Project