Changelog

Made in v1.6.3 – v1.6.5

Our latest update focuses on making Willow more reliable, more intuitive, and more tailored to you.

🚀 Smarter Pipeline Control
You can now reset the dictation pipeline while the device is initializing or processing, meaning if Willow ever hangs, you can instantly escape with a simple Esc press. No more force-quits.

🔔 Customizable Notifications
A brand-new Notifications tab in Settings lets you disable activity reminders if you prefer a quieter workspace.

🔑 Authentication, Fixed
We squashed a tricky bug where cached auth data was sometimes pulling old tokens instead of refreshed ones, leading to transcription not working properly.

📦 Previous Updates

v1.6.4 – Improved authentication handling and better UI response to mid-aware request failures.
v1.6.3 – Fixed infinite recursion with Chrome & Slack explorers (causing crashing) and re-enabled tap gestures after expiration.

— Lawrence Liu (CTO)

Made in v1.6.0 – v1.6.2

Our latest update makes Willow much better at multi-lingual support!

🌍 Real-Time Language Intelligence

• Willow now automatically detects and switches languages on the fly. Speak in multiple languages, even mid-sentence, and Willow keeps up seamlessly.

• Language handling is powered by a completely new backend runtime built for speed and adaptability.

✍️ Smarter, More Natural Text Flow

• Spaces are now intelligently inserted when starting new dictations, so you don’t have to backspace or fix formatting between clips.

• Cleaned-up output removes em-dashes, partial phrases, and other artifacts, producing text that’s immediately ready to use.

🔇 Total Audio Isolation

• Audio ducking now silences 100% of background sound when Willow is active. We’ve been hearing speaker sounds have been reducing transcription quality — this should fix it!

See you all soon!

— Lawrence Liu (CTO)

Made in v1.5 – v1.5.4

This was one of our most significant infrastructure upgrades.

⚡ New Speed Engine (Now Default)

• Our re-engineered dictation core is now officially stable. Short-form transcription is near-instant, and streaming output is significantly faster and smoother.

• Removed random filler expressions caused by streaming artifacts.

🏢 Enterprise-Ready Features

• SSO support for seamless enterprise authentication.

• Sync with enterprise-required privacy settings, including daily transcript clears and privacy lock enforcement. Live now (teams.willowvoice.com)

🔐 Reliability & Startup Improvements

• Willow now launches automatically at login — no UI interaction required.

• Authentication persists across restarts correctly (many edge cases covered)

• Pasteboard data no longer leaks into third-party apps like Raycast.

✨ Cleaner Output by Default

• Partial output tags (etc.) and em-dashes are now removed.

• Notifications for AirPod connections are clearer and more consistent.

This was all about taking Willow from “a tool you open” to “a system that’s just there.” Always on, always fast, and ready for enterprise scale.

— Lawrence Liu (CTO)