How it works

Your computer, finally
working for you.

Friday sits on your desktop and acts on your behalf, no tab-switching, no copy-pasting, no learning new tools. Here's exactly how it does that.

The basics

Three steps, every time

Every interaction with Friday follows the same simple pattern, no matter how complex the task.

01

Summon Friday from anywhere

Press Ctrl + Q on your keyboard, from any app, any window, any moment. Friday's HUD overlay appears instantly without interrupting what you were doing.

You don't need to minimize your work, open a browser tab, or find an app icon. Friday is always one keystroke away, running quietly in your system tray until you need it.

⌨️ Global hotkey ⚡ Instant launch
02

Talk to it naturally

Click the microphone or just start speaking. Friday uses real-time voice streaming, it hears you as you talk, not after you finish. No push-to-talk, no waiting.

You can also type if you prefer. Either way, use plain language, "summarise my emails from this morning", "find the invoice from last Tuesday", "book me a meeting with the team for Thursday". No commands to memorise.

🎙️ Live voice 💬 Or type 🌍 Plain English
03

Friday acts, you stay focused

Friday doesn't just answer, it does. It can open apps, click buttons, fill in forms, search the web, read your screen, and run long tasks in the background while you keep working.

When it needs your input, a confirmation, a choice, a correction, it asks. When it's done, it reports back. You stay in the driver's seat the whole time.

🖱️ Takes real actions 🔄 Background tasks ✅ Checks in when needed

Under the hood

What Friday can actually do

Here's a closer look at Friday's core capabilities and how each one works in practice.

🎙️

Real-time voice conversations

Powered by Gemini Live API

Friday uses Google's Gemini Live API over a persistent WebSocket connection for bidirectional audio. It processes your voice as you speak, not after, so responses feel instant, like talking to a person.

  • Low-latency 16-bit audio streaming
  • Understands natural speech, accents, and hesitations
  • Responds in voice or text depending on context
  • No wake word required, just press Ctrl+Q and speak
👁️

Screen vision & computer use

Sees and understands your display

Friday can take a screenshot of your screen and understand what's on it using Gemini's Computer Use capability. Ask "what does this error mean?" or "click the confirm button", it sees exactly what you see.

  • Reads text, buttons, and UI elements on screen
  • Can navigate complex or unfamiliar interfaces
  • Handles web pages, desktop apps, and documents
  • Screenshots are used only for the active task
🖱️

Full desktop control

Windows UI Automation bridge

A lightweight C# bridge uses Windows UI Automation to interact with any native Windows app, not just browsers. Friday can click, type, scroll, and read content in apps like Outlook, Excel, Slack, and more.

  • Works with any standard Windows application
  • Injects real mouse and keyboard events at OS level
  • Reads UI element labels, values, and states
  • Runs as a background sidecar process
⚡

Background sub-agents

Delegate and forget

For longer tasks, research, drafting, data gathering, Friday spawns a background sub-agent that works independently while you continue your conversation or switch back to your work.

  • Multiple tasks can run simultaneously
  • Sub-agents report back when complete
  • You stay in flow while work happens in the background
  • Each agent is isolated and can be cancelled
🌐

Web browsing & automation

Chrome extension bridge

With the Friday Browser Bridge extension installed in Chrome, Friday can navigate websites, fill in forms, extract information, and interact with web apps, going far beyond simple web searches.

  • Navigates to any URL on your behalf
  • Reads and interacts with live web pages
  • Fills forms, clicks links, extracts content
  • Auto-reconnects via heartbeat, always ready
🔒

Local-first & private

Your data stays on your machine

Friday has no backend server. Conversation history is stored locally in a SQLite database on your PC. The only external connection is to Google Cloud Vertex AI to process your AI requests.

  • No account required to use core features
  • Conversation memory stored on-device only
  • Voice audio is streamed, not stored, by Friday
  • Google Cloud handles AI processing securely

Questions

Common questions

Core app functions and local conversation history are available offline, but AI responses require a connection to Google Cloud Vertex AI. Voice and task execution features need internet access to work.
Friday can interact with any standard Windows application that uses Windows UI Automation, which covers the vast majority of desktop apps including Outlook, Excel, Word, Slack, VS Code, browsers, and more. Apps with custom or non-standard UI frameworks may have limited support.
No. Friday streams your audio in real time to Google Cloud Vertex AI for processing, but does not store recordings locally or on any Friday server. Google's data handling is governed by their Cloud Privacy Notice.
Friday only takes actions you explicitly ask for. For consequential actions, like sending a message or deleting a file, it will confirm with you first. You're always in control, and you can stop any task at any time.
Friday is built for Windows 10 and Windows 11. Support for macOS and Linux is on the roadmap but not yet available.
No, the Chrome extension is optional. Without it, Friday can still control the Chrome window through Windows UI Automation, but the extension unlocks deeper web interaction like reading page content, filling forms, and navigating complex web apps more reliably.

Ready to try Friday?

Free to download. No account needed.

Download for Windows →