Hey PH community! I’m Rogério Bayer, the developer behind Focus Patrol.

I’m excited to share a side project I’ve been building over the past few months. I created it because I couldn’t find a privacy-respecting tool that solved a very specific issue of mine: I needed a focus tracker that didn’t require sending my entire browsing history to a server.

The objective was simple: a distraction monitor that runs entirely on the user’s machine, with full control over the data.

Focus Patrol uses on-device AI — Gemini Nano when available, or a WebLLM/Llama fallback for broader compatibility — to classify navigation patterns directly inside the browser. No external calls, no data leaving your machine. The extension generates focus reports locally and wipes all logs every night.

One of the key challenges was dealing with the uneven availability of Gemini Nano across devices. That limitation pushed me to implement a local fallback using WebLLM so the experience remains consistent regardless of hardware or platform. Moving forward, I want to refine this fallback model further, exploring options like Gemma 3 or upcoming Llama generations. Better prompt customization and a more dynamic scoring system are also on the roadmap, so the tool adapts more precisely to each user’s daily routine.

Building this made me appreciate how powerful local AI can be — especially inside the constraints of a browser environment. It feels like just the beginning of what can be done when we bring meaningful AI capabilities directly to the device, without compromising privacy or requiring constant connectivity.

Thanks for taking a look! 🚀

https://focuspatrol.bayer.ooo/