OVRIDERYour voice. Your agents. Your memory.

OVRIDER, founded by Oxford University alumni, starts as hands-free control of your whole computer, and grows into something much bigger: a panel of AI advisers you design yourself, and a memory of you that you can ask questions of.

It takes a little setting up, and the Helpbot is always one prompt away to guide you. The more you put in, choosing your agents, giving them their specialisms and files, and letting Neuron learn your day, the more it gives back.

1. Run your computer by voice and gesture

• Dictation: your words appear live as you talk, punctuated and edited by voice ("full stop", "carriage return", "delete word"), then sent into any app with "over", like on a walkie-talkie.
• Point with your finger: say "control" and your webcam turns your hand into the mouse. Move your finger and the pointer follows, with a targeting sight that locks on; make a fist to pause, say "focus" to steady it on something small, and click by voice. It all runs on your own computer: nothing from your camera leaves it.
• The mouse and keyboard: say "target" for a targeting grid and steer the pointer by compass ("north east 3"), click with "make it so", save the places you click most and jump back to them by name, scroll, press any key, or bring up a glowing on-screen keyboard.
• Tabs, windows and apps: say "window" and every window on your screen is outlined and numbered, then pick one by its number; switch tabs, jump to Chrome or Word, and minimise, maximise or close windows, all by voice.
• Every screen, every computer: OVRIDER works across several screens at once, and you can give each computer its own name ("Engage Bill", "Engage Henry"), so the desktop and the laptop each answer only to you when you mean them.
• Your way: every command can be switched on or off, and OVRIDER comes in nine colour schemes, from HAL's red eye to a starship's warp field and a car launch's spotlights, each with its own animated background, that you can switch between by voice.

2. Build your own panel of AI agents

Create as many agents as you like, each a specialist you design: a value investor, a takeover lawyer, a brand expert, a film buff, a chef. Give each one a name, a voice, a face, a call sign and a specialist focus, then load it with your own notes and files. Say "agents" and your whole panel appears across the screen as a row of cards, ready to be called.

• Call them by name: say "Charlie" and Charlie answers, aloud in his own voice and in a glowing panel on your screen. Bring in another ("Martin, what do you think?") and they build on each other's answers, like a meeting of advisers.
• Speak, type or share: ask by voice, type in the box under an agent, or drop a PDF on it, and the agent reads it and tells you what it makes of it.
• Real research: agents can search the web, open a company's website, read its earnings call transcripts and its official filings, look up encyclopaedias and NASA's data, describe the window you're looking at, and find or draw pictures.
• Summarise anything: say "summarise this conversation" and any agent writes up everything your panel has discussed. One click turns a summary or a transcript into a polished 600-word article, ready to save as a PDF or copy.
• A note taker that never forgets: Scribe keeps one running notebook of your ideas and reminders. Say "Scribe, note that…" at any moment, in any conversation, and it's kept, ready to read back or summarise whenever you ask.
• Keep the conversation: save a discussion by name, load it again later to carry on, and print any of them as a clean PDF.

3. Films, music and video, on command

Your agents aren't only for work. Connect the services you already love and they become a voice-controlled remote for your whole media life.

• YouTube: ask an agent to find the latest videos from a channel or on any subject, and say "play number 3" to watch it right there, in cinema size or popped out beside your work.
• Read any video in seconds: say "bring up the transcript" for every word of a video, or ask "what does he say about interest rates?" and the agent answers from the video itself. Every transcript you open stays in the conversation, so any agent can discuss it.
• Spotify: "play some jazz", "play my favourite playlist", or pick up a podcast episode where you left off, then pause, skip or go back by voice.
• Films and series: ask about any film or show, its cast, its reviews or where it came from, and see its poster and watch the trailer on the spot.
• And dinner: ask for a recipe for whatever you fancy, with the ingredients and every step, read out as you cook.

4. Neuron: a memory of you

Neuron is the part of OVRIDER that gets to know you. It begins with your screen: switch it on and it quietly turns what you read, write and work on into a memory, and writes you a brief of each day. Give an agent access to it and ask what you saw, when, and what it said.

Over time it remembers not just your days but your weeks, months and years, and the ways you can share your world with it will keep growing, from the apps and devices you already use.

• Yours alone: everything Neuron keeps goes into your own Google Drive or Dropbox, never to OVRIDER, and you decide what it learns.

Free, or as powerful as you want

Out of the box, OVRIDER uses free, open-source AI that runs on your own computer: speech recognition, voices, the agents' thinking, reading pictures and Neuron's notes. Nothing to pay, and nothing leaves your computer to be processed.

To make that work, the first time you log in on a computer OVRIDER downloads these free, open-source models onto it by itself, one after another (about 6.5 GB in all, speech recognition first, so you can start talking within minutes). They stay on that computer for next time, and you can choose others, or remove them, in the menus.

When you want more, connect your own accounts, such as OpenRouter for the leading AI models or ElevenLabs for lifelike voices, and pay them directly for what you use. OVRIDER doesn't charge for any of it.

Who it is for

Anyone who thinks for a living: investors, analysts, lawyers, founders and researchers who want advisers on call and a memory that keeps up with them; people with RSI, limited mobility or other accessibility needs who want a computer they can run by voice; and anyone who would rather talk than type.

Questions

Is OVRIDER free?

Yes. Log in with Google or GitHub and everything works with free, open-source AI on your own computer. The premium options use your own accounts with those services, paid to them directly.

What do I need?

A microphone and a modern browser such as Chrome or Edge. Controlling the mouse, keyboard and the on-screen panels needs a small free plugin, for Windows or Mac, which OVRIDER helps you set up. Neuron also needs your Google Drive or Dropbox connected, to keep its memory in.

Does it work on a Mac?

Yes. The free Mac plugin gives the same controls as the Windows one, using the Mac's own shortcuts (for example ⌘ + Tab to switch apps). Macs ask you once to allow keyboard, mouse and screen access in System Settings.

What happens to my voice and my screen?

OVRIDER only listens after you press Engage, and stops when you press or say Disengage. With the free speech recognition, your voice is turned into text on your own computer. Neuron only captures while it's switched on, and keeps its captures in your own Google Drive or Dropbox. The privacy policy has the details.

What does OVRIDER keep about me?

Your login, and your settings and API keys, encrypted, so they follow you to any computer. Your agents' files, saved conversations and Neuron's memory are kept in your own Google Drive or Dropbox. See the privacy policy.

Start using OVRIDER