# Meetily: Local AI Meeting Assistant with Whisper Transcription > Meetily is a privacy-first AI meeting assistant that transcribes and summarizes meetings entirely on the user's device, using Whisper or Parakeet for transcription and Ollama or other configurable backends for summarization. No audio or transcript data leaves the machine. **Zackriya-Solutions/meetily** — Privacy first, AI meeting assistant with 4x faster Parakeet/Whisper live transcription, speaker diarization, and Ollama summarization built on Rust. 100% local processing. no cloud required. Meetily (Meetly Ai - is the #1 Self-hosted, Open-source Ai meeting note taker for macOS & Windows. Understand How to write meeting minutes. - Repository: https://github.com/Zackriya-Solutions/meetily - Website: https://meetily.ai - Stars: 31,155 · Forks: 3,398 - Language: Rust - License: MIT - Published: 2026-08-04 · Updated: 2026-08-18 · Language: en - Canonical page: https://hysenlabs.com/projects/zackriya-solutions-meetily ## The Privacy Gap Meetily Addresses Cloud-based meeting transcription tools process audio on external servers. For healthcare providers, legal professionals, defense contractors, and enterprises under GDPR or similar regulations, this creates data residency and confidentiality problems. The Meetily README opens with this framing directly, citing data breach costs and GDPR enforcement as evidence that cloud meeting tools carry real risk. Meetily addresses this by running every component of the transcription and summarization pipeline on the user's own machine. Audio is captured locally, transcription runs locally via Whisper or Parakeet models, and summaries are generated locally via Ollama or optionally routed to a remote provider the user controls. The README states that no data ever leaves the computer in the default configuration. The target audience is professionals in any regulated sector, self-hosted infrastructure teams, and individuals who prefer data sovereignty over the convenience of cloud services. ## Architecture: Rust, Tauri, and a Python Backend Meetily is built on Tauri, a framework for building desktop applications with a Rust backend and a web-based frontend. The repository is a Cargo workspace with two members: `frontend/src-tauri` (the Tauri application) and `llama-helper` (a Rust crate for interacting with local language models). A separate Python backend handles the transcription workload. The `backend/` directory in the repository contains this component. The top-level directory also includes a `scripts/` directory and a `docs/` directory with platform-specific build guides. The Cargo workspace dependencies include Tokio for async I/O, Serde for serialization, and Anyhow for error handling. The minimum Rust version for the workspace is 1.77. The frontend is web-based, running inside the Tauri shell. This design means the user interface is HTML and JavaScript rendered locally by a WebView, while the heavy processing happens in the Rust and Python layers. The repository's top-level `CLAUDE.md` file indicates the project has integrated AI coding assistant configuration, a pattern visible in projects that use automated development tooling. Custom OpenAI-compatible endpoint support allows organizations with private AI infrastructure to plug in their own summary backend without using a public provider. The README describes this as suitable for organizations that have their own AI infrastructure or preferred providers, keeping the summary step within the organization's own network boundary. ## Installation by Platform On macOS, download the aarch64 DMG from the releases page and drag the app to the Applications folder: ```bash # After downloading meetily_0.4.1_aarch64.dmg: # Open the DMG, drag Meetily to Applications, then launch from Applications ``` On Windows, download `x64-setup.exe` from the releases page and run the installer. The README notes that the packaged Windows build uses a Vulkan-enabled Whisper build and requires an AVX2-capable x64 CPU. CUDA acceleration requires a source build configured with a compatible NVIDIA CUDA toolchain. On Linux, build from source: ```bash git clone https://github.com/Zackriya-Solutions/meeting-minutes cd meeting-minutes/frontend pnpm install --frozen-lockfile ./build-gpu.sh ``` Detailed build instructions for Linux are in `docs/building_in_linux.md` and `docs/BUILDING.md` in the repository. ## Transcription Models and AI Summary Providers Meetily supports two local transcription backends: Whisper and Parakeet. Both run entirely on the local machine. The README describes Parakeet as producing transcription at four times the speed of Whisper, which is the figure in the project description. No cloud transcription option is available; the local-only constraint is by design. For meeting summaries, Meetily supports multiple AI providers: Ollama (recommended for full local operation), Claude, Groq, OpenRouter, and any custom OpenAI-compatible endpoint. Ollama runs local language models, completing the local-only pipeline. The other providers send the transcript text (not the audio) to external APIs, which represents a reduced privacy footprint compared to cloud transcription but is not fully local. Professional audio mixing captures both microphone input and system audio simultaneously, with ducking and clipping prevention built in. The README also describes an Import and Enhance feature (in beta) for re-processing existing audio files with a different model. ## Speaker Diarization and the PRO Edition Speaker diarization, the process of attributing transcript segments to individual speakers, is available in the community edition as of v0.4.1 based on the repository description. The README's PRO upgrade notice lists speaker diarization as "planned for PRO in mid-June," which corresponds to an earlier version of the README. The community edition feature set should be verified against the current releases page. Meetily PRO is a paid tier advertised in the README. It adds enhanced accuracy, advanced export formats, custom summary workflows, and team-ready features. A coupon code (LAUNCH20) for twenty percent off is listed in the README, valid until the next community edition release. The community edition remains MIT licensed and free. The PRIVACY_POLICY.md file in the repository provides the formal data handling documentation. The repository also includes a CONTRIBUTING.md and a CODE_OF_CONDUCT.md, indicating an open contribution model. The `docs/` directory contains platform-specific build guides including `docs/building_in_linux.md` and `docs/BUILDING.md`. For regulated-use adoption, reviewing the privacy policy and the build flags for local model integration is the appropriate verification step. ## Meetily Versus Granola Granola is a meeting notes application that integrates with calendar systems and uses cloud AI to generate summaries. It is designed as a polished consumer product with minimal setup and cloud infrastructure handling the processing. Meetily differs on the fundamental architecture: all processing is local. Granola requires sending audio or transcript data to an external service by design. The trade-off is setup complexity: Granola requires no model provisioning, while Meetily requires the user to download and configure Whisper or Parakeet and optionally set up Ollama for local summarization. For teams inside a compliance boundary, Meetily's local-only model is a hard requirement that Granola cannot meet. For individuals who prioritize ease of use over data sovereignty, Granola's cloud architecture eliminates the model setup step. The two tools serve different constraints rather than competing directly on features. ## Conclusion Meetily is the right choice for professionals and teams who need meeting transcription without sending audio to an external service, particularly in regulated industries where cloud data handling creates compliance risk. It is not the right choice for users who need a polished no-setup experience: Linux requires building from source, and the AI summary quality depends on the local model the user provisions. Before adopting it, verify that your hardware can run a local Whisper or Parakeet model at acceptable speed, and decide whether Ollama local summarization or a remote provider like Claude or Groq fits your workflow. The last push was on 2026-09-15. ## FAQ ### Is Meetily free? The community edition of Meetily is MIT licensed and free. A PRO tier with enhanced accuracy, advanced exports, and custom summary workflows is available as a paid upgrade. ### Is Meetily open source? Yes. The community edition is MIT licensed and the source code is on GitHub at Zackriya-Solutions/meeting-minutes. The build instructions for all platforms are documented in the repository's docs/ directory. ### How to install Meetily? On macOS, download the aarch64 DMG from the releases page and drag it to Applications. On Windows, run the x64-setup.exe installer. On Linux, clone the repository and run `./build-gpu.sh` from the frontend directory after installing pnpm dependencies. ### Is Meetily safe to use? In its default configuration, Meetily processes all audio and transcription locally with no data sent to external servers. The PRIVACY_POLICY.md file in the repository documents the formal data handling policy. Using a non-Ollama summary provider sends transcript text (not audio) to that provider's API. ## Sources - [Official documentation](https://meetily.ai) - [Official README](https://github.com/Zackriya-Solutions/meetily#readme) - [Project repository](https://github.com/Zackriya-Solutions/meetily) - [Release notes](https://github.com/Zackriya-Solutions/meetily/releases) --- Hysen Labs editorial analysis, written from the project's own repository and release notes. Cite the canonical page: https://hysenlabs.com/projects/zackriya-solutions-meetily