Ito is an open source voice assistant for Mac and Windows that transforms your intent into smart text in any app. Speak naturally to write emails, messages, or code without typing. Say intent, not just words.
How did you get around the native Mac transcription being terrible and whisper being slow? Did you take the approach most do where you approximate with dictation first and then hot swap text a few seconds later after it’s processed through a better model?
Being open source is very important for this kind of tool, and also collecting user feedback vigorously is the only way you can truly know what’s important and is predestined for success.
I wish I could just voice to text this instead of typing… oh wait!
Huge congrats, Barron & Evan! Been following Ito for a while and it’s awesome to see it finally launch on Product Hunt. Love the VibeType vision—and extra kudos for releasing it open-source. Keep the updates coming! 🚀
Talking product position. How are you positioning your self differently than wispr?
It worked wonderfully when you first launched.
Guess I should've set to auto-start, forgot iTo was on board until today
(Full transparancy-- I've also been taking wisprFlow and talkTastic thru the paces.)
I downloaded at your first launch, and was impressed.
After a PH alert today about ITO, I launched it again, turned off the other transcriber before testing iTo, clicked install update, held shortcut, and then... well, nothing.
Ver. now installed: Version 0.15.17 (0.15.17)
Invoking of Keyboard Shortcut results: no transcription.
Relaunch has no effect so far, any suggestions?
Thanks in advance for your reply,
— Michael
System: MacOS 26.0.1 (25A362) on a macStudio, yeti mike.
@Ito user here! Super cool product + Barron & Evan rock :)
Congrats on the launch team Ito 🔥🔥🔥
Hey everyone 👋, I'm Arjun and I lead product here at Ito.ai.
You know that 3-minute gap between back-to-back meetings?
That’s where my ideas used to die. Each new meeting brings new context, new faces, new problems. And whatever clarity I had from the last one was gone.
That’s what motivated us to build Ito - a voice assistant that understands what you’re trying to say, not just what you said, and gives you leverage in your workflow.
In practice:
Code review thoughts → speak the feedback → Ito formats it with proper markdown
Post-meeting clarity → capture action items while they're fresh → formatted task list ready
Brainstorm mode → speak stream-of-consciousness → get it back as coherent sections
How it works: Intent over words.
Speak the vibe and the goal. Ito handles the formatting, structure, and polish. No micromanaging punctuation. Just capture your intention.
Our principles:
Open by default - Transparent code and permissions so you can audit, extend, and trust the assistant that lives inside your OS
Fast and frictionless - If it takes longer than typing, we've failed. Speed isn't a feature; it's the foundation.
Launch-day asks 🙏
Tell us your top 3 repetitive writing tasks, and we'll prioritize Ito's skills around them
Share edge cases where the tone or formatting isn't quite right
Give us ideas on features and use cases to build - Ito is a community-led effort, after all
Thanks for giving Ito a spin and for helping us build an assistant that actually assists.
Congrats Barron and team. VibeTyping feels like the missing link between dictation and a real assistant. Intent over words, open source, and a hotkey that works anywhere is a killer combo. How do custom vocab and style presets work?
There's a lot of competition in this space — and so I love that Ito is open source!
It's also cross-platform from the jump, and includes a Rust toolchain for building native components, meaning it's faaaaast.
Very curious to learn about the roadmap...
Hey Product Hunt, I'm Barron Caster, co-founder of Ito.ai!
Why Ito
We believe the future of interacting with technology is through voice, which is more natural than typing. With AI exploding, everyone deserves their own Jarvis. We built Ito because we think your voice assistant should be open source for transparency and customization. We only saw closed or dictation-focused tools out there.
Ito is a voice assistant that understands what you actually want (and eventually will take actions on your behalf).
The Problem
Typing remains the biggest bottleneck to human-computer interactions. Most voice tools fall short with limited or clunky approaches:
Basic dictation apps transcribe but don’t handle context, formatting, or real intent. Some nail transcribing but miss the bigger picture of taking work off your plate.
General assistants (Siri, Alexa, Google Assistant) excel at reminders or searches but fumble in professional workflows, ignoring custom vocab or app-specific actions.
Overreaching permissions are required by other closed-source companies while they install keyloggers without visibility into what they are tracking.
With years of speech recognition experience, we built Ito.ai to be a real voice assistant - not just a way to type a little faster.
Ito introduces VibeTyping - type what you want to say.
Ito understands what you actually want and writes like you mean it. It transforms your words into polished content anywhere on Mac or Windows.
Intent over words: Unlike rigid dictation, Ito understands nuance. Draft a full email, crank out React code, craft a PRD, or polish a tweet with context-awareness.
Works everywhere, instantly: Hit a hotkey in any text box (Slack, Notion, Google Docs, Cursor), speak, and watch it generate formatted drafts, code snippets, or messages.
Fully yours to tweak: Open-source and community-powered. Add custom vocabulary, personalize styles from past chats, and even build voice-driven workflows for accessibility. No black-box lock-in.
Fast out of the gate: Ito is fast.
Open: As privacy advocates, we built a product that you can audit yourself, especially given the extensive permissions needed.
Who is this for?
If you type, Ito frees you to speak your mind and save time. Developers, marketers, Product Hunters, and more.
Get started today
Download Ito for free at Ito.ai. It’s open source, so if you don’t like it, fork it
How did you get around the native Mac transcription being terrible and whisper being slow? Did you take the approach most do where you approximate with dictation first and then hot swap text a few seconds later after it’s processed through a better model?