·building·project·shipped

Bonk

A tiny AI life coach: a friendly desktop doge watches your active window, tracks goals, and gently bonks you back toward the work you meant to do.

What It Is

Bonk is an Electron desktop pet for staying aligned with your own goals. A small doge sits in a transparent window while you work. When the app detects that you have drifted into distracting apps or sites for long enough, it swaps into a bonk animation, plays a small sound, and posts a short supportive nudge in chat.

The pitch is deliberately gentle: not a harsh alarm, not a surveillance dashboard, and not a productivity scorecard. It is a lightweight companion that remembers what you said you wanted to do and notices when your attention is sliding away from it.

How It Works

The app has two surfaces:

  • A main window for goals, settings, and coach chat.
  • A transparent desktop pet window that stays visible while you work.

The repo describes the current behavior as polling the active window once per second. If a distracting app or site remains active for about 15 seconds, and the cooldown has passed, Bonk runs a 1.2 second animation, briefly swaps from the idle doge image to the bonk image, plays a beep, and adds a concise nudge to the chat.

AI support is optional. The settings UI can store OpenAI or Anthropic keys locally through Electron storage, and the coach can use those providers for small supportive messages and proactive check-ins. Without that, the core loop still works as a local desktop pet and distraction detector.

Why It Matters

The interesting product choice is tone. Bonk treats focus as a relationship with intent rather than a punishment problem. The transcript frames the core use case clearly: you open the laptop to work, end up 30 minutes into YouTube or Twitter, and need a friendly cue that reconnects you to the proposal, task, or goal you sat down for.

That makes Bonk closer to a tiny ambient coach than a full life-management system. It knows your goals, watches for obvious attention drift, and gives just enough friction to help you remember what mattered before the distraction took over.

Build Notes

The implementation is a compact JavaScript, HTML, and CSS Electron app. The main process owns timers, active-window checks, AI calls, IPC, and the pet window. The preload bridge exposes safe window.coach calls to the renderer. The renderer handles the main app UI, settings, goals, chat, and pet surface.

The remaining product frontier is smarter context classification: distinguishing harmless research from real drift, making nudges more personalized without feeling invasive, and keeping the desktop pet useful over repeated work sessions instead of turning into another ignored notification.

Neighborhood

Related

lifeloggerlifeloggerWorkplace Surveillance SystemWorkplace Surveillance Syst...Bonk