A calmer Windows control layer

Don’t reach, just ask

Fluent turns voice, gaze, and text into Windows actions you can inspect before they run.

Open source for Windows. Voice and text work without a camera.

Voice Gaze Text

Move through every app, in your own way.

One attentive layer translates intent into visible, interruptible Windows actions.

One calm layeracross Windows

Works where your work already lives

OutlookVS CodeEdgeTeamsNotionSlackFile ExplorerGitHub

Less control relay

One request. One visible route.

Say the outcome, add a target, and let Fluent prepare the small steps without hiding them.

You

“Draft a clear reply from these notes.”

Speech is transcribed locally

Fluent

Fluent prepared a reply and paused before sending.

Waiting for you

Made for the way you work

Choose a path and see Fluent in motion.

Accessibility

A shorter route through Windows.

Use voice, optional gaze, or text when reaching, clicking, and repeating become the hardest part of the task.

See the accessible path

A professional with limited hand mobility speaking toward a desktop computer in a bright studio

Understanding before action

The control stays with you.

Fluent makes uncertainty visible, keeps consequential actions paused, and leaves recovery close.

Reviewable actions

Important steps wait for you.

Inspect the target, edit the output, keep waiting, or approve one prepared action.

Prepared action Paused before send

Read selected notes Draft reply Send once

Your decision

Send the prepared reply?

Thursday afternoon works well. Please send the updated notes beforehand.

Personal language

Your words stay yours.

Names and terms can be corrected before they become part of a request.

Jason Fluent WinUI DeepSeek PowerShell Suhari

Optional gaze

Attention names the target.

Looking can add context. It never clicks, selects, or acts on its own.

Clear boundaries

Signals stay separate from the plan.

Raw gaze is not stored. Speech is transcribed locally. The plan receives only useful context.

Read the implementation

Built by people, for people

What Fluent is designed to return.

Visible intent

The request stays readable before it becomes a sequence of actions.

Useful pauses

Fluent slows down when the next step needs certainty, consent, or a human correction.

Easy recovery

Cancel, edit, and try again are part of the route, not emergency exits hidden after the fact.

Private signals

Raw audio and gaze samples stay out of the action plan. Only the useful context moves forward.

Personal agency

The computer can help without taking over. You choose the input, the boundary, and the final step.

Windows voice control, explained

Voice and gaze control for Windows, with the final say kept visible.

Fluent is a preview Windows accessibility and productivity app for people who want to use less keyboard and mouse input. Natural voice or text requests express the outcome. Optional gaze adds context. Every route stays visible so you can inspect, change, stop, or approve what happens next.

What is Fluent?: Fluent is an open-source Windows control layer that translates a spoken or typed request into visible, interruptible UI Automation actions. It is being developed as inclusive productivity software with access needs at the center.
Can Fluent control Windows by voice?: Yes. Use the Fluent wake phrase, click the microphone, or hold the push-to-talk shortcut. Fluent transcribes speech locally, prepares the Windows actions, and pauses consequential steps for review.
Does Fluent support hands-free computer control?: Fluent is designed to reduce repetitive reaching, clicking, and typing. Voice can drive a request, while optional gaze can identify the intended on-screen target without acting as a click by itself.
Do I need a camera or eye tracker?: No. Voice and text work without a camera. Gaze is optional, must be enabled separately, and adds target context only when a command is submitted.
Does speech audio leave my computer?: No cloud speech service receives the raw audio. Transcription happens on the device. The resulting text request is then sent to the hosted planning provider the user configures.
Is Fluent ready for everyday use?: Fluent is currently a preview for developers and early testers, not a finished consumer release. Review the public roadmap and current source before trying it.

Ask another AI

Still not sure Fluent fits the way you work?

Open a prepared question and get a second opinion on Fluent's fit, privacy, consent, and current limits.

A professional speaking naturally beside a laptop in a warm workspace

Start with less effort.

Use the signal that costs you least. Keep the final say over what happens next.

Windows. Open source. Built around review and consent.