OpenFlow
OpenFlow is a free, MIT-licensed voice dictation app for macOS and Windows, forked from Handy by CJ Pais. It is written by Avijeet Dubey. Press a shortcut or say a wake word, and your words appear in the focused field. Speech-to-text and AI cleanup each pick a backend, local, self-hosted or remote. One caveat: builds are unsigned and contributions are not being accepted.
What does OpenFlow do?
The README calls OpenFlow "a local-first, cross-platform (macOS + Windows) voice-dictation desktop app, a privacy-respecting alternative to hosted tools like Wispr Flow. It's built on Tauri 2 + Rust + React." A hotkey or a wake word starts a recording; the words appear where your cursor is.
The feature that sets it apart from its parent, Handy, is independent backend modes. You pick one mode for speech-to-text and a different one for cleanup if you want. For each, the choices are local on-device models, a self-hosted endpoint you run yourself, or a remote hosted provider. On top of that, OpenFlow adds a local analytics dashboard, per-app AI cleanup tone, OS-keychain-backed API keys, and a wake-word hands-free mode.
Key facts
- Licence: MIT on the code. The README notes OpenFlow is a fork of Handy by CJ Pais, with the original copyright and licence preserved.
- Platforms: macOS (Apple Silicon arm64 and Intel x64) and Windows x64. Linux is not listed.
- Three backend modes, set independently for speech-to-text and cleanup: local on-device, a self-hosted endpoint, or a remote provider.
- Model Setup has live Test buttons to validate a self-hosted or remote endpoint before you rely on it mid-dictation.
- API keys are stored in the OS keychain, not in plaintext settings files.
- Push-to-talk, tap-to-toggle, or a wake-word hands-free mode.
- Analytics dashboard: words per minute, time saved, activity by app, top keywords, computed locally, with adjustable privacy modes.
- Built with Tauri 2, Rust and React. Buildable from source with Rust (via rustup) and Bun.
What does it replace, and where does it fall short?
OpenFlow replaces Wispr Flow for the dictation loop, with the choice of where speech-to-text and cleanup run. The README names Wispr Flow directly as the tool it is "a privacy-respecting alternative to."
Where it falls short: the builds are unsigned, because the project does not yet hold an Apple Developer certificate or a Windows Authenticode certificate. The first launch requires a one-time unblock step on each OS, which the README documents. macOS auto-updates may re-prompt for Accessibility, Input Monitoring and Microphone permissions because the signature changes between versions. Contributions are not being accepted right now; the maintainer runs OpenFlow as a solo project alongside a full-time job. Builds before v0.13.0 cannot auto-update to the first updater-enabled release and have to be reinstalled by hand.
How does OpenFlow run?
Download the installer from openflow.computer or the GitHub Releases page. On macOS, open the DMG and drag OpenFlow to Applications; because the build is unsigned, Gatekeeper blocks the first launch, so right-click Open or run xattr -dr com.apple.quarantine /Applications/OpenFlow.app. On Windows, SmartScreen requires one More Info, Run anyway click. Grant Microphone, Accessibility and Input Monitoring permissions on macOS.
Building from source needs Rust and Bun. Apple Silicon uses a prebuilt ONNX Runtime; Intel macOS needs ONNX Runtime installed through Homebrew and linked dynamically. Windows needs long paths enabled first because transcribe-cpp's CMake build produces paths that exceed MAX_PATH. Windows speech-to-text builds are CPU only.
Who is OpenFlow for?
A dictation user who wants backend modes picked independently for speech-to-text and cleanup, with API keys in the OS keychain and an analytics dashboard that stays on the machine. Someone who wants a signed, hand-polished macOS experience should look at Handy or voiceink on the open-source alternatives to Wispr Flow list. The project is a solo build; a team deployment needs plans for the maintenance gap.
What limits does the README state?
"Contributions are not being accepted right now." Builds are unsigned, so the OS warns on first launch and macOS may re-prompt for permissions after an update. Builds before v0.13.0 do not auto-update and must be reinstalled by hand. Windows speech-to-text builds are CPU only. Linux is not listed as a platform. The CMake build on Windows produces paths that exceed MAX_PATH, so long paths must be enabled in Windows first.
Questions people ask
Is OpenFlow open source?
Yes. The code is MIT, an OSI approved licence, and the README preserves the original Handy copyright. There are no additional conditions on the code. Contributions are not being accepted at the moment, but the licence does not restrict forking, reading or modifying the source.
Can OpenFlow be used commercially?
Yes. MIT allows commercial use, modification and redistribution, provided the licence travels with the code. OpenFlow itself is free. If you connect a remote provider for speech-to-text or cleanup, that provider's pricing and retention apply. Local backends keep the audio on the machine.
Does OpenFlow work offline?
Yes with a local backend for both speech-to-text and cleanup. OpenFlow's backend-mode design lets each one be set independently, so a user can run speech-to-text locally and leave cleanup off, or run cleanup locally too. A self-hosted endpoint also works offline from the public internet.
How does OpenFlow compare with Handy?
OpenFlow is a fork of Handy that adds independent backend modes for speech-to-text and cleanup, a wake-word hands-free mode, API-key storage in the OS keychain, an analytics dashboard and per-app cleanup tone. Handy is the older, more conservative project that stays with the one-job dictation loop on all three desktops; OpenFlow drops Linux and adds the extras. The open-source alternatives to Wispr Flow page sets them side by side.
Is OpenFlow signed?
No. The README says code signing and notarisation (Apple Developer ID, Windows Authenticode) are deferred to a future public-distribution phase. The first launch on each OS needs a one-time unblock step, and macOS may re-prompt for Accessibility, Input Monitoring and Microphone after an update because the signature changes between versions.
Sources
- OpenFlow README and LICENSE: github.com/avijeett007/openflow, read 2026-10-26. The README is the source for every feature, platform note, backend mode and limit above, and for the Handy fork lineage.
- Project website: openflow.computer, read 2026-10-26.
More voice dictation apps
A free, offline push-to-talk dictation app for macOS, Windows and Linux that runs Whisper or NVIDIA Parakeet on your own machine and pastes the text at your cursor, under MIT.
A GPL-3.0 macOS dictation app that runs Nemotron, Parakeet, Cohere Transcribe, Apple Speech or Whisper on the device, with command and write modes, per-app prompts, and an optional closed local enhancement model.
A dictation and meeting-notes app under MIT that runs Whisper or Parakeet locally or cloud models with your own key, detects Zoom, Teams and FaceTime calls, and labels speakers on device.
A macOS dictation app under GPL-3.0 whose README names Superwhisper and Wispr Flow as the tools it replaces, with local models, per-app modes, a personal dictionary and an AI assistant mode.
A speech-to-text app inside the Epicenter monorepo under AGPL-3.0 that records, transcribes with a provider you choose, cloud or local, optionally polishes the text, and pastes it at the cursor.
An Apache-2.0 macOS dictation app that calls Gemini 3.5 Transcribe with your own key, writes audio to disk from the first millisecond for crash safety, and inserts polished text at your cursor without an account.
Added October 26, 2026. Every claim above comes from the project's README, LICENSE or model card, read on October 26, 2026, or from the GitHub API on the date shown in the panel. Found an error? Write to reveneau@licheo.com and it is fixed in the next weekly pass. Repository: github.com/avijeett007/openflow.
Want OpenFlow running on your own servers?
Reveneau sets it up on your hardware, connects it to your product, and builds the software around it.