Doing | Voice and visual context for AI builders. No subscription. | Product Hunt Doing is for AI builders who use voice and screenshots to bring context to Claude Code, Codex, and other AI agents. Tap a hotkey and Doing listens to your voice and captures screenshots. Optimized over thousands of hours of building with Claude Code. Private, local, no account, no subs. Just a quality tool that you own and works well.

Interested in Voice Transcription and local private AI models?

I'd love your feedback on Doing: www.producthunt.com/products/doing

Hey! We're on Product Hunt today. Say hi www.producthunt.com/products/doing

New today: Hands free mode.

Just tap "shift" while holding your hotkey to lock recording on, talk for as long as you like, and then tap your hotkey again to stop.

Great for longer transcriptions or when you decide you need your hand back.

@talat.app is another one worth checking out.

Useful AI tools like Granola need to be re-thought to be private, on-device, and local. There's just too many risks with sharing your data with a subscription based cloud service that has to provide VC level returns to investors.

This is the way

New today: Hands free mode.

Just tap "shift" while holding your hotkey to lock recording on, talk for as long as you like, and then tap your hotkey again to stop.

Great for longer transcriptions or when you decide you need your hand back.

Terminal session management. I have a cmux workspace for each project I'm working on, which then has multiple Claude Code sessions going, as well as other terminals for building/running tools, and browser testing. cmux also has great notifications for letting you know when Claude needs something.

What’s your current AI engineering / building stack?

We are heavy on Claude Code (Max), cmux, and doing.tools.

#buildinpublic #indiedev

How to Write Better AI Prompts with Voice You give better instructions when you talk than when you type. Here's how to use that for AI prompts.

New guide on how to get better outcomes from Claude Code and AI by using voice instead of typing doing.tools/guides/bette...

Doing vs Other Voice Transcription Apps for Mac How Doing compares to SuperWhisper, Wispr Flow, and other Mac voice transcription tools. Local, fast, private, $49 once.

Curious how Doing compares to other voice transcription tools like Superwhisper and WisprFlow? Doing is focused on AI Builders and those working only with AI tools. Read our comparison to learn more: doing.tools/compare

Friday drop: the Doing app now let's you grab screenshots in addition to voice via your recording hotkey. Great for quickly adding visual context to your voice transcriptions. Try it: doing.tools

And when you're ready for faster transcription and builder-friendly workflows, check out doing.tools

macOS 26 Voice Transcription: Setup Guide for Developers (2026) Apple's foundation model makes Mac dictation actually good. Setup steps, hotkey config, and real developer workflow tips.

Want to use text-to-speech on your mac to talk with apps like Claude Code and Codex? Here's a guide on how to use Apple's built-in voice transcription model on macos 26+ for free: doing.tools/guides/macos...

Favorite feature: YOLO Mode. Turn it on and Doing automatically pastes and submits your words to the target app.

You transcriptions are stored on device as daily markdown files. Review your history from Obsidian or any other text editor. Simple, LLM friendly.

We use NVIDIA's Parakeet TDT for blazing fast on-device local transcription. 10x faster than whisper or other cloud based models. Completely private.

Just hold your hotkey, speak, and your words are pasted automatically where your cursor is.

Doing is fast, private, local voice transcription for mac. No subscription. Works with every app like Claude Code, Codex, Amp, Cursor, Slack, etc. Try it: doing.tools

What doing?