Beta · Free · Open source

Browse the web with your hands

hands.hinow.ai turns your webcam into the pointer. Scroll a page, step through its links and click them with a few plain hand poses. No mouse, no keyboard, no touchscreen. Nothing to hold and nothing to aim at.

Add to Chrome, free See the source

Chrome 116+ · needs a webcam · works offline

The problem

Pointing is the hardest thing an interface can ask for

Every mouse asks the same thing: move a cursor onto a target a few millimetres wide, and hold it perfectly still while you click. For most people that is invisible. For someone living with tremor, reduced strength, limited reach or fatigue, it is the barrier that turns an ordinary page into a locked door.

The usual answer is to make the targets bigger. Ours was to stop asking people to aim at all. The selection steps from link to link, the way a keyboard does, and a gesture confirms it. Accuracy stops mattering; intention is enough.

Three scenes of a click that will not land: a man at his desk with a hand on his forehead in front of a button, a speaker pointing at a projected chart while the cursor sits still, and an older woman waiting for a museum kiosk to answer her touch
A speaker on stage with a raised hand towards a giant slide wall, controlling the presentation from a distance

On stage

No clicker, no cable and no asking for help

Anyone who presents knows the scene: the clicker is on the table, the laptop is on the other side of the stage, and changing a slide turns into a walk in the middle of a sentence, or a request for help from someone in the audience.

With the camera framing you, your hand does it from wherever you are. Nothing to hold, nothing to hide in your palm, nothing that can drop or run out of battery.

What the room needs

  • A camera that frames you from the waist up
  • Even light on your hands: strong backlight is what gets in the way most
  • The presentation open in the browser

How to prepare the deck: It works with an HTML presentation: an ordinary page with navigation buttons that can be invisible, because the click gesture triggers the buttons and the deck advances. A PDF opened in the browser and long-scroll decks work too, advancing by scrolling.

The four poses that do all of this

Open handscrolls the page, or moves to the next link
Index finger upscrolls up, or clicks if you hold it for two seconds
Index finger to the sidesteps from one link to the next
Closed fistlocks everything, on either hand
See the full vocabulary

Every day

The browsing you already do, without touching anything

Scroll the news page, open the story, go back, open the next one. It is what people do a hundred times a day, and it is exactly what jams when a hand will not obey, when an arm tires before the end of the article, or when both hands are busy with something else.

An open hand scrolls. A finger held sideways steps from link to link, with the chosen one highlighted on screen. Two seconds with the finger up and it opens. At no point do you have to carry a cursor to a small target and hold it there.

A person at a laptop browsing a news site with hand gestures, without touching the keyboard or the mouse

In the lab

The analysis moves on without anyone removing a glove

Sterile bench, gloves on, sample open. A keyboard and a mouse there are a contaminated surface and a full stop: to zoom into an image, compare two results or open the next test, someone has to stop, decontaminate, touch and come back.

With the camera, whoever is at the bench moves through what is on screen without touching anything. The glove stays clean, the sample stays where it was, and the reading is not interrupted to operate the computer.

In three steps

  1. 1The analysis on screen, ready to be read
  2. 2Both hands take over the interface
  3. 3The result is browsed with no keyboard and no mouse
See the photo

What already works: Scrolling, stepping and clicking in any system open in Chrome, which covers web image viewers, records and results dashboards. Zooming and rotating with both hands, as in frame 2, is the next step we are going to build.

In exhibitions

A wall that responds without anyone touching it

A museum, a gallery, a trade fair, a company lobby. The big screen is already there, and the only way to interact with it is usually to touch: a queue to get close, greasy glass by the end of the day, and plenty of people who simply do not touch.

With the camera, the content responds to whoever is standing in front of it, metres away. Nobody has to learn anything, nobody has to touch, and the same installation serves one person after another with nothing shared but the view.

In three steps

  1. 1A visitor approaches the piece, still at rest
  2. 2Both hands go up and take control
  3. 3The installation responds and the wall bursts into colour
See the photo

What already works: Any content open in Chrome can be browsed and triggered by gesture, metres from the screen: that is enough for a kiosk, a catalogue or a browsable exhibition. Reacting to continuous movement, as in frames 2 and 3, is the next step we are going to build.

In the classroom

The digital board obeys from where the teacher is standing

The interactive screen in a classroom is expensive and almost always ends up used as a projector. Changing content means walking back to it, touching it, hunting for the right button and losing the thread of the explanation on the way, several times a lesson.

With the camera, the teacher drives the material facing the class, from where they already are. The lesson does not stop to operate the equipment, and nothing depends on reaching the top of the screen or on holding one more remote.

In three steps

  1. 1The lesson starts with the content on screen
  2. 2The teacher raises both hands and takes control
  3. 3The material opens up and follows the explanation
See the photo

What already works: Browsing and triggering by gesture any material open in Chrome, including a lesson page, a PDF and video, without leaving the class. Handling a three-dimensional object with both hands, as in frame 3, is the next step we are going to build.

At home

The whole web with no mouse, no keyboard and no help

For anyone living with tremor, low strength or limited reach, the computer in the living room is not hard to understand: it is hard to reach. Carrying a cursor to a link a few millimetres wide and holding still while it clicks is the most demanding task an interface can ask for, and it is what turns an ordinary page into a locked door.

With the webcam the laptop already has, the hand becomes the pointer with no aiming required. The selection steps from link to link, a gesture confirms, and reading the news, opening a video and answering a message stop depending on having someone nearby.

In three steps

  1. 1The page opens and the selection starts on the first link
  2. 2An open hand moves to the next, a closed fist stops everything
  3. 3The index finger up for two seconds clicks

What already works: This is the case that works in full today, with no promise of a next step. Scrolling, stepping through links, clicking and changing pages is exactly what the extension already does, on any site open in Chrome.

With your hands full

The recipe scrolls itself with flour on your hand

A kitchen, a workshop, a repair bench. The screen is a hand's width away, the next step is just below, and the hand that would scroll the page is the one in the dough, in the grease or holding the part. The usual answer is to clean up, touch, and get dirty again, several times per recipe.

With the camera, a gesture in the air scrolls the page, advances the step and plays the video without touching anything. The screen stays clean, and whatever was in your hand stays in your hand.

In three steps

  1. 1The recipe or the manual open on screen
  2. 2An open hand in the air scrolls to the next step
  3. 3A closed fist locks everything while you work

What already works: Scrolling, playing and stepping through links works today on any page open in Chrome, which already covers recipe sites, PDF manuals and video. The real limit here is not the gesture but the camera: a hand against a window, in strong backlight, is what interferes most with recognition.

The four poses that do all of this

Open handscrolls the page, or moves to the next link
Index finger upscrolls up, or clicks if you hold it for two seconds
Index finger to the sidesteps from one link to the next
Closed fistlocks everything, on either hand
See the full vocabulary

Where this fits

Every screen someone has to operate from far away, or without touching

Once the hand becomes the pointer, the list of places where this solves a real problem gets long. These are situations where the mouse is out of reach, where touching the screen is not allowed, or where the hands simply cannot let go of what they are doing.

Presenting and teaching

  • Talks and presentationsadvance slides, point, highlight elements
  • Meeting roomsrun video calls, documents and presentations from a distance
  • Educationa teacher handling content, maps, anatomical models
  • Architecture and engineeringwalk through CAD and BIM models in front of the client

Big screens and distance

  • TVs and giant panelsdashboards, maps, videos and menus from across the room
  • Command centresswitch between cameras, maps, alerts and indicators
  • Projected screensturn a whole wall into an interface
  • Kiosks and digital signageinteraction with no physical contact
  • Interactive shop windowsbrowse products through the glass
  • Museums and exhibitionsvisitors interacting without touching the screens

When hands cannot touch

  • Medicinereview scans and images without leaving the sterile field
  • Laboratoriesoperate interfaces in gloves, without contaminating equipment
  • Professional kitchensrecipes, timers and videos with your hands full
  • Workshops and maintenancethe manual stays open while you work on the machine
  • Industryscreens operated with protective gear on your hands
  • Hazardous environmentscontrol the computer from a distance
  • Receptions and hotelspublic systems operated without contact
  • Gymsvideos, series and music during a workout

On the everyday computer

  • Accessibilityanyone who struggles with a mouse or a keyboard
  • News and readingscroll, go back, go forward, open stories
  • Videoplay, pause, volume and moving along the timeline
  • E-commercebrowse the catalogue, zoom and rotate products
  • Smart homemusic, lights and automation from the computer
  • Design and 3Dmove the canvas, rotate and zoom objects
  • Music and DJmixer, effects and tracks without touching the gear
  • Gamesgestures as extra commands

What is ready today: Today the extension scrolls the page, steps through and clicks links and changes pages, on any site open in Chrome. The rest of the list is where this way of interacting makes sense, and it is what guides the order of what we build first: dragging, zooming and keyboard emulation are next.

How it works

Two hands. Four poses. That is the whole vocabulary.

Each hand has one job, and you choose which is which: there is a left-handed mode. On-screen panels list every gesture and light up the one being recognized, so you always know whether the camera understood you.

One hand scrolls the page

Open handScroll down
Index finger upScroll up
Index finger to the sideNext link
Closed fistStop everything

The other picks and clicks

Open handSelect the next link
Index finger to the sideBack to the previous link
Index finger up, 2 secondsClick the selected link
Closed fistStop everything

Both hands together change page

Hold one hand open and point the other one left or right for a second. Two poses at the same time never happen by accident, which is exactly why the one action that replaces your whole screen is the one that asks for both hands.

A closed fist stops everything

On either hand, at any moment. The easiest pose to make is the one that interrupts the system, and it pauses without losing what you had selected.

In practice

This is how it looks on your screen

The panels in the corners list every gesture and light up the one being recognised, so you never have to wonder whether the camera understood you. The selected link is highlighted, and the strip at the bottom says in words what is happening. All of it can be switched off once you no longer need it.

The extension running over an article: the left hand's gesture panel in the bottom left corner, the right hand's in the opposite corner, and the status strip in the middle reading Link selected

Why it exists

What we are building

01

Make the web usable without a mouse

Not easier to use with a mouse. Usable without one, on any site, without the site having to cooperate or even know we exist.

02

Ask for intention, never precision

Every gesture in the vocabulary is a whole-hand pose that a child can form. No pinching, no aiming, no holding still on a small target.

03

Fail towards doing nothing

When recognition is uncertain, nothing happens. An interface that acts on a guess is worse than one that waits, because undoing a mistake by gesture is expensive.

04

Keep the camera at home

Hand tracking runs entirely on the device. A product that watches you has to earn that trust by never sending the picture anywhere.

05

Stay free, and stay open

Accessibility should not have a toll booth. The code is public so anyone can audit the privacy claims, fork it, or adapt it for a need we did not think of.

Privacy

The video never leaves your computer

Each frame is read into memory, analysed by a hand-tracking model that ships inside the extension, and discarded. No frame is stored: not on disk, not in the browser. No frame is transmitted. What travels between the parts of the extension is the recognized gesture and a handful of coordinates, never the picture.

And this is enforced, not merely promised: the extension declares connect-src 'self', so the browser itself blocks any outbound request from its pages, including from third-party libraries bundled inside it. You do not have to take our word for it, and the code is public so you can check.

Read the full privacy policy →

No accountNo sign-up, no login, no email.
No analyticsNo telemetry, no crash reports, no tracking of any kind.
No networkThe extension makes no requests at all. It works offline.
Camera off by defaultIt only runs while you have gestures switched on.

Questions

Frequently asked

Does it work on every website?

Yes, on any ordinary page. The exceptions are pages Chrome does not allow extensions into at all (its own settings pages and the Web Store), plus PDFs in the built-in viewer and DRM-protected video, which do not expose their content to the page.

Does it send my camera anywhere?

No. Frames are analysed on your machine and discarded frame by frame. The extension declares a policy that makes any outbound request impossible, and the hand-tracking model is bundled inside it, so recognition works with no internet connection.

How much does it cost?

Nothing. It is free and open source, and it will stay that way.

Do I need a good webcam?

No. Any ordinary laptop camera is enough. Even lighting helps more than resolution does, and there are sliders for stability and reach if your setup needs them.

I am left-handed. Does that work?

Yes. A single switch swaps the roles, so the hand that picks and clicks is your dominant one, and the on-screen panels swap sides to match.

Can I use it with only one hand?

Partly, today. One hand can scroll and move the selection between links, but clicking still needs the other hand. A full one-hand mode is the next thing we are building.

Can I type with it?

Not yet. Text entry still needs a keyboard, and it is one of the next things we are building.

Which browsers?

Chrome 116 or newer, on desktop. Other Chromium browsers usually work but are not tested.

We are in open beta. Come and test it.

The extension works, it is free and it installs in a click, but it is still in testing. Use it for a few days and tell us how it went: what you report is what decides which gesture we build next.

Open handscrolls the page, or moves to the next link
Index finger upscrolls up, or clicks if you hold it for two seconds
Index finger to the sidesteps from one link to the next
Closed fistlocks everything, on either hand
Add to Chrome, free See the source