Built at TELUS Digital · WillowTree

Never say
"I'll follow up on that"
again.

Cue listens to your meetings, answers your questions in real time using your knowledge base, and writes your summary when the call ends. All on your Mac. No audio ever leaves your machine.

macOS 13 Ventura or later · WillowTree internal

Cue in action — live transcript with AI chat panel
How it works

Three note-takers, one meeting

The AI, the transcript, and you — all in the room together.

🎙

Listens

Cue captures your mic and system audio separately, transcribes everything on-device using macOS Speech Recognition, and labels each speaker in real time.

On-device · No cloud API · No audio uploads

Answers

Ask anything in the chat panel mid-meeting. Cue searches your live transcript and connected knowledge base — docs, proposals, past meetings — and responds instantly.

FuelIX-powered · Ollama supported · Knowledge base integration
📝

Summarizes

When the call ends, Cue generates a structured summary: key decisions, action items with owners, open questions, and the full interleaved transcript and chat.

Editable · Exportable · Stored locally
Privacy & Security

Local by design

Your data stays on your machine.

Every component — speech, storage, and AI — is built so sensitive meeting data never has to leave your Mac.

🎙

On-device transcription

Speech recognition runs on macOS's native SFSpeechRecognizer — the same engine Siri uses, running fully locally. No audio is streamed, no file uploaded, no third-party service called. Your words never leave your Mac.

SFSpeechRecognizer · Zero uploads · No cloud API
🔐

Encrypted at rest

API keys are stored in macOS Keychain via Electron safeStorage — the same mechanism used by 1Password and Chrome. Transcripts, chat history, and meeting summaries live in a local encrypted SQLite database that never syncs to any server.

macOS Keychain · Local SQLite · No cloud sync

Enterprise AI via Fuel iX

When Cue answers a question or generates a summary, it calls Fuel iX — TELUS Digital's enterprise AI orchestration platform. Your data stays within TELUS infrastructure, governed by the system that serves 50,000+ users and holds an ISO 31700‑1 Privacy by Design certification. No third-party AI. No model training on your data.

ISO 31700-1 certified · 50K+ users · No training
🤖

Full local model support

Connect Ollama and run any open-source model entirely on your machine — zero API key, zero network traffic. Cue falls back to Ollama automatically when Fuel iX is unavailable. Every AI slot (chat, interjections, summaries) is assignable independently to any local or cloud provider.

Ollama · Zero API key · Fully air-gappable
Already using Granola? Cue is built for client-sensitive meetings — different in five concrete ways: where the audio goes, speaker separation, sanctioned AI infrastructure, mid-meeting answers, and where meetings live. Why not Granola? →

Get Cue

Download the latest release and start your next meeting differently.

macOS 13 Ventura or later · All releases

Get involved

Built by WillowTree at TELUS Digital. Come build with us.

Cue is an open internal project. Whether you want to use it, improve it, design for it, or just share opinions — you're welcome here. The best way to start is the Slack channel.

#cue-notetaker Join on TELUS Digital Slack
🎙

Use it in meetings

The most valuable contribution is using Cue and sharing what you notice. Real meetings reveal what tests can't.

💬

Share feedback

Post in #cue-notetaker. Reactions, friction, ideas — all useful.

🛠

Build features

React, TypeScript, Electron, Swift. Check GitHub Issues for what's open. Small PRs preferred.

🎨

Design & UX

We have a design system and strong opinions. If you have stronger ones, bring them.