locus

Inspiration

Better performance starts with understanding yourself, and understanding yourself starts with data you've never had. The people who work well and keep improving have a clearer sense of how they actually function, and they make decisions from that. But sometimes we guess at when we focus best, borrow habits from strangers, and judge our days by feel. Locus is built for people who want to understand their own focus honestly: to see how their attention actually behaves, test what genuinely helps, and catch the quiet declines that usually go unnoticed until they've already cost something. For students who can't afford a coach or don't have an advisor paying close attention, that cost can be heavy. Locus is the check-in no one else is giving them.

How we built it:

The app runs in the browser and uses the user's webcam during a focus session. Every few seconds, we capture a frame and send it to Claude with a prompt framing it as an attention analyzer, which reads posture, gaze, and body language from the image and returns a focus score from 0 to 100 in structured JSON. Those per-frame scores are smoothed into a live attention signal for the session, and then aggregated across sessions to show longer-term trends in how the user's focus changes over time. Everything talks to the Anthropic API; no video ever leaves the session.

Problem and safeguards:

A tool that uses a webcam to analyze focus has to take privacy seriously, and we designed Locus around that from the start. No video is ever stored or sent anywhere beyond the real-time analysis call, only the derived focus score is kept, and session data is never shared with employers, institutions, or anyone else. The camera is only active during a focus session the user explicitly starts, and users can permanently delete their history at any time.

Project empowerment:

Locus is built on the idea that the best person to improve your life is you, as long as you can actually see what's happening in it. Instead of prescribing routines or ranking users against each other, we give each person their own data and their own decisions. Users choose which habits to test, define what a good session looks like for them, and interpret the patterns in a way that fits their life. The AI's job is to make the invisible visible and the user's job is to decide what to do with what they see.

Built With

• javascript
• next.js
• node.js
• python