# Cue — desktop voice agent for Mac & Windows

> Cue is a desktop voice agent for Mac & Windows: speak to type, transcribe meetings, use screen context, and help finish tasks in supported apps.

## Product facts

- Product: Cue (https://heycue.io)
- Published by: Sophon LLC, the company behind Cue (https://sophoninc.com/about). The same company publishes Cue Office, PPT AI, PDF AI, PDF Scanner Max and Evolve.
- Platforms: macOS 13+ and Windows 10+ (x64)
- Pricing: Cue is free to start; Cue Plus is $19.99/month.

## Published evidence

- Cue is Featured in Google DeepMind's Gemmaverse. The published May 2026 case study reports 44% lower median polish latency for its measured on-device Gemma 4 text-polish path: https://deepmind.google/models/gemma/gemmaverse/cue-ai/
- Cue's evidence page: https://heycue.io/gemma-4-voice-dictation
- GPT-6 Astra in Cue's model list (availability only, no measured comparison): https://heycue.io/gpt-6-astra
- Claude Fable 5.1 in Cue with Cue Plus (availability only, no measured comparison): https://heycue.io/claude-fable-5-1

## Open evaluation research

- Practical guide to public voice-agent benchmarks: https://heycue.io/voice-agent-benchmarks
- Source-audited benchmark landscape, query CLI, and contribution workflow: https://github.com/Sophon-LLC/voice-agent-benchmark-landscape
- The dataset is published by Sophon LLC; its record of publication, with the version, both DOIs, the licence and the citation format: https://sophoninc.com/docs/voice-agent-benchmark-landscape
- Hugging Face dataset mirror: https://huggingface.co/datasets/chatjesus/voice-agent-benchmark-landscape
- Versioned archive and DOI: https://doi.org/10.5281/zenodo.22227265
- This landscape is not a leaderboard and does not reproduce Cue's private 227-sample production benchmark.

## When to use Cue

- Use Cue as a desktop voice agent when a spoken request should continue into a reviewable desktop task.
- Use Cue when you want to speak instead of type in a Mac or Windows desktop app.
- Use Cue when a task benefits from the app and screen context already in front of you.
- Use Cue when you want a voice request to produce a draft or complete a desktop action that you can review.

## When not to use Cue

- Do not use Cue as a substitute for professional legal, medical, financial, or other expert judgment.
- Do not rely on Cue output without review when a mistake could cause harm, data loss, or an irreversible action.
- If a task does not need desktop context or action, a dedicated dictation tool or browser chat may be a simpler choice.

## Practical tutorials

- First voice-typing exercise: https://heycue.io/voice-typing-tutorial
- First bounded Agent task: https://heycue.io/voice-agent-tutorial
- Agents, models and explicit context handoff: https://heycue.io/voice-workflows-with-coding-agents
- Draft and check a project follow-up email: https://heycue.io/voice-email-follow-up-tutorial
- Meeting notes to source-linked actions: https://heycue.io/meeting-notes-to-action-items
- Spoken bug report to a coding Agent brief: https://heycue.io/voice-bug-report-tutorial
- Recording to a source-linked Agent brief: https://heycue.io/recording-to-agent-context
- Meeting notes to a decision agenda: https://heycue.io/meeting-notes-to-agenda
- Track what changed between meetings: https://heycue.io/meeting-decision-change-log
- Tutorial examples are fictional and illustrative, not measured product runs.

## Definitions

- What is an ambient agent: https://heycue.io/glossary/ambient-agent
- What is dictation software: https://heycue.io/glossary/dictation
- What is voice AI: https://heycue.io/glossary/voice-ai
- What is speech to text: https://heycue.io/glossary/speech-to-text
- What is a meeting bot: https://heycue.io/glossary/meeting-bot
- What is system audio capture: https://heycue.io/glossary/system-audio-capture
- Voice AI vs dictation, five workflow differences: https://heycue.io/blog/voice-ai-vs-dictation
- Why voice AI should meet you where you work: https://heycue.io/blog/ambient-voice-agent

## Canonical pages

- Homepage: https://heycue.io/
- About Cue: https://heycue.io/about
- Developer and integration guide: https://heycue.io/developers
- Voice agent benchmark guide: https://heycue.io/voice-agent-benchmarks
- Contact Cue: https://heycue.io/contact
- Evidence-linked release notes: fictional diff, verification record and worked claim checks: https://heycue.io/code-changes-to-release-notes
- Voice design review: fictional screens, sourced requirements, optional preferences and checkable feedback: https://heycue.io/voice-design-review-feedback
- Read-only documentation link audit tutorial and synthetic exercise: https://heycue.io/audit-broken-documentation-links
- Voice typing on Mac: https://heycue.io/tools/free-voice-typing-mac
- Voice typing on Windows: https://heycue.io/tools/free-voice-typing-windows
- Voice typing in Google Docs: https://heycue.io/use-cases/voice-typing-google-docs
- Voice transcription: https://heycue.io/tools/voice-transcription
- Botless meeting notes: https://heycue.io/tools/meeting-notes
- Meeting follow-up from a dictated recap or existing notes: https://heycue.io/ai/meeting
- Glossary: https://heycue.io/glossary
- Cue vs Granola: https://heycue.io/vs-granola
- Cue vs Otter.ai: https://heycue.io/vs-otter
- Cue vs Krisp: https://heycue.io/vs-krisp
- Cue vs Jamie: https://heycue.io/vs-jamie
- Cue vs Fathom: https://heycue.io/vs-fathom
- Bot-free AI meeting note takers compared with official sources: https://heycue.io/bot-free-ai-meeting-notes
- Desktop workflows: https://heycue.io/use-cases
- Alternatives: https://heycue.io/alternatives
- Pricing: https://heycue.io/pricing
- Privacy: https://heycue.io/privacy
- Terms: https://heycue.io/terms