talat records your microphone and system audio, transcribes both sides of every conversation in real time, and turns meetings into searchable, editable notes. Powered entirely by on-device AI, and stored on your computer.
One time purchase, with every future update included when purchased during pre-release.
Features
- 100% on-device transcription — mic and system audio never leave your machine
- Automatic meeting detection across Zoom, Teams, Google Meet, and anything that grabs your mic
- Real-time speaker diarisation you can edit, split, or reassign on the fly
- On-device LLM summarisation (Qwen3-4B by default) with configurable system prompts
- Bring-your-own-LLM for summaries: OpenAI, Anthropic, OpenRouter, Ollama, or the built-in local model
- Auto-export to any folder — Obsidian, iCloud, Dropbox, a local Git repo, wherever you keep your notes
- Webhook export to any HTTP endpoint, Markdown or JSON, with auth headers and retry logic
- Built-in MCP server so you can chat to your own meeting history from Claude Code or any stdio MCP client
- Click-to-seek audio playback linked to every line of the transcript
- Full local search (SQLite + FTS5) across every transcript you've ever recorded
- 20MB Tauri app, runs on Apple Silicon Macs and Windows 10/11
- Pay once ($49 pre-release, $99 at 1.0), no subscription, no account required, no in-app analytics
Use Cases
- Client-privileged conversations (lawyers, therapists, doctors, coaches) where sending audio to a third party isn't an option
- Freelancers and consultants keeping a searchable archive of every client call
- Regulated or enterprise settings where cloud SaaS doesn't clear procurement
- Remote teams running 1:1s, standups, and retros without adding a meeting bot to the call
- Solo founders capturing user interviews, sales discovery, and investor calls
- Researchers running qualitative interviews with reliable transcripts and proper search
- Journalists and writers taking source interviews entirely locally
- Anyone who wants meeting notes as a durable record they own, rather than a feed they rent
Fazier Deal
Sponsors
Buy
Comments
Hi folks. I'm Nick; I'm building talat with my friend and cofounder Mike. I love Granola and use it every day, but I never loved that everything passes through someone else's servers. Over the last year I pieced together the bits that would let me build a local-only version (system audio capture, AEC, automatic meeting detection), and FluidAudio (Parakeet TDT on Apple's Neural Engine) finally gave me the transcription model to match. Mac and Windows, both entirely on-device, no account, pay once. If you try it and something breaks or annoys you, please tell me. Cheers, Nick
Premium Products
Fazier Deal
Sponsors
Buy
Makers
Makers
Comments
Hi folks. I'm Nick; I'm building talat with my friend and cofounder Mike. I love Granola and use it every day, but I never loved that everything passes through someone else's servers. Over the last year I pieced together the bits that would let me build a local-only version (system audio capture, AEC, automatic meeting detection), and FluidAudio (Parakeet TDT on Apple's Neural Engine) finally gave me the transcription model to match. Mac and Windows, both entirely on-device, no account, pay once. If you try it and something breaks or annoys you, please tell me. Cheers, Nick
Premium Products
New to Fazier?
Find your next favorite product or submit your own. Made by @FalakDigital.
Copyright ©2025. All Rights Reserved