Conversationaly – Privacy-First AI Meeting Assistant

Conversationaly, a privacy-first AI meeting assistant for macOS, Windows, and Linux, records meetings, transcribes them live, and generates summaries entirely on-device by default, with no account, cloud round-trip, or telemetry. The free, open-source desktop app, a fork of Meetily rebuilt around transcribe.cpp and a bundled llama.cpp sidecar, supports ~87 transcription models across 16 families (default: nemotron-3.5-asr-streaming-0.6b-q8) and offers opt-in cloud STT and LLM providers. It runs a local Gemma 4 model for summaries and supports GPU acceleration via Metal, CUDA, Vulkan, and ROCm.

Records your meetings, transcribes them live, and writes the summary — on your machine, with no account and no cloud round-trip unless you deliberately configure one. Table of Contents Conversationaly is a desktop app macOS, Windows, Linux that captures your microphone and system audio, transcribes the meeting as it happens, and generates a summary. Transcription models and the summary LLM both run locally by default — nothing is sent anywhere. Cloud providers are available if you want them, but they are opt-in, per-feature. It is a fork of Meetily https://github.com/Zackriya-Solutions/meeting-minutes , rebuilt around transcribe.cpp https://github.com/handy-computer/transcribe.cpp and a bundled llama.cpp sidecar. It is fully free — there is no paid tier, no license key, no telemetry. Local transcription — ~87 models across 16 families Whisper, Parakeet, Nemotron, Canary, Voxtral, Qwen3-ASR, SenseVoice, Moonshine, GigaAM, … , downloaded on demand. Default: nemotron-3.5-asr-streaming-0.6b-q8 , multilingual across 39 locales. Live transcript — streaming-native models transcribe continuously as you speak; batch-only models are segmented by voice activity and still work live. Built-in AI, no Ollama required — a bundled llama-helper sidecar runs Gemma 4 locally for summaries, and can also transcribe directly as an audio LLM. Bring your own LLM — summaries via Built-in AI, Ollama, Claude, Groq, OpenRouter, OpenAI, or any OpenAI-compatible endpoint. Optional cloud STT — Deepgram, ElevenLabs, Groq, or OpenAI, if you prefer a hosted transcriber. Professional audio mixing — microphone and system audio captured together with RMS-based ducking and clipping prevention. Import & enhance Beta — transcribe existing audio files, or re-transcribe a past meeting with a different model or language. Summary templates — pick or write the structure your summaries follow, and set the summary language independently of the spoken one. GPU acceleration — Metal on Apple Silicon, CUDA NVIDIA , Vulkan AMD/Intel , ROCm AMD on Linux . Local storage — meetings, transcripts, and models live in a SQLite database and a model directory on your disk. Prebuilt installers macOS .dmg , Windows .exe , Linux .deb / .rpm / .AppImage are published on the Releases page https://github.com/bykof/conversationaly/releases when a version is tagged. Requires Rust, Node.js, pnpm, and cmake. See docs/BUILDING.md /bykof/conversationaly/blob/main/docs/BUILDING.md for per-platform prerequisites. git clone https://github.com/bykof/conversationaly cd conversationaly/frontend pnpm install macOS ./clean build.sh Linux auto-detects GPU backend ./build-gpu.sh Windows clean build windows.bat Linux specifics: docs/building in linux.md /bykof/conversationaly/blob/main/docs/building in linux.md . macOS — microphone, plus screen recording for system audio ScreenCaptureKit, macOS 13+ . Windows — microphone; system audio uses WASAPI loopback. On first launch, onboarding downloads one transcription model and one Gemma 4 tier. From then on: - Microphone and system audio are captured, mixed, and written to a recording. - The same mixed audio is resampled to 16 kHz and fed to the transcription engine, which emits transcript lines as the meeting runs. - When you ask for a summary, the transcript goes to whichever LLM provider you configured — the local sidecar by default. Everything above is a local process. Cloud STT and cloud summary providers are the only paths that leave your machine, and only when you select one and supply a key. A single Tauri application: a Rust core audio capture, transcription, storage, summary orchestration and a Next.js frontend, communicating over Tauri commands and events. There is no separate server to run. Details: docs/architecture.md /bykof/conversationaly/blob/main/docs/architecture.md . cd frontend pnpm install ./clean run.sh macOS: build and run info logging ./clean run.sh debug verbose logging clean run windows.bat Windows ./dev-gpu.sh Linux pnpm run tauri:dev plain dev mode pnpm run tauri:dev:metal force a specific GPU backend pnpm run tauri:dev:cuda pnpm run tauri:dev:vulkan pnpm run tauri:dev:cpu Architecture notes and conventions for contributors live in CLAUDE.md /bykof/conversationaly/blob/main/CLAUDE.md ; GPU backend details in docs/GPU ACCELERATION.md /bykof/conversationaly/blob/main/docs/GPU ACCELERATION.md . The backend/ directory is an archived Python/FastAPI service from before the Tauri rewrite. It is unsupported and not needed to build or run the app. Issues and pull requests are welcome. See CONTRIBUTING.md /bykof/conversationaly/blob/main/CONTRIBUTING.md for project structure and guidelines. MIT — see LICENSE.md /bykof/conversationaly/blob/main/LICENSE.md . - Conversationaly is a fork of Meetily https://github.com/Zackriya-Solutions/meeting-minutes by Zackriya Solutions, which it builds on under the MIT license. - Transcription runs on transcribe.cpp https://github.com/handy-computer/transcribe.cpp , built on ggml https://github.com/ggerganov/ggml / whisper.cpp https://github.com/ggerganov/whisper.cpp . - Local LLM inference uses llama.cpp https://github.com/ggerganov/llama.cpp via llama-cpp-2 https://crates.io/crates/llama-cpp-2 . - We borrowed some code from Screenpipe https://github.com/mediar-ai/screenpipe and transcribe-rs https://crates.io/crates/transcribe-rs . - Import & Enhance was contributed by Jeremi Joslin https://github.com/jeremi , improved by Vishnu P S https://github.com/p-s-vishnu and Mohammed Safvan https://github.com/mohammedsafvan . - Thanks to NVIDIA for the Parakeet and Nemotron speech models, and to the teams behind the other model families in the catalog.