Focus, earned — not surveilled.

Fragmented focus is a modern crisis.

A student sits down to study, opens their laptop, and an hour later the page is still blank. Short-form video and endless feeds are engineered to fracture attention — and the apps meant to fight back don't work.

Every person deserves to focus when they want to and step away from their devices when they need to. Modern technology makes that genuinely hard: addictive short-form content and infinite scroll are designed to capture attention, and people lose real hours of their lives to it. Reclaiming that cognitive sovereignty — the ability to decide where your own attention goes — is the heart of the problem we set out to solve.

We narrowed a sprawling issue down to something a team could actually design for:

The direct stakeholders are students and student friend groups; indirect stakeholders include the roommates who share their study space, universities, and platform providers. We built two personas from our research to keep both in view.

To stay close to the emotional reality, we mapped what a struggling student says, thinks, does, and feels. The "feels" quadrant shaped Locus the most — our goal is accountability that never tips into shame.

We studied five tools students already use. Each gets one piece right and misses the combination Locus needs.

The gap is the intersection: no tool combines verified focus, private friend-group accountability, and a privacy-first architecture. Closing that gap is what Locus is for.

This is an adaptive challenge, not a technical one. App blockers already exist, yet they fall short because the real problem is behavioral — the habit of doomscrolling itself. A purely technical limit can't read whether a person is actually being productive; doing that well requires genuinely understanding intent, and that understanding is different for every person. A clean, off-the-shelf solution doesn't exist, which is exactly why the problem is worth solving.

Locus: a private study room with a quiet, honest AI.

Locus recreates the "quiet accountability" of a physical library — you're present and visible to your friends, but no one is watching the content of your screen. On-device AI confirms focus, a shared score keeps the room locked in, and optional stakes add a gentle nudge.

You record a time-lapse of yourself while you study. Each focused minute adds to your study time on a private, friend-group leaderboard. Local AI watches your camera and screen to confirm you're on task — and because it runs entirely on your device, no video, screen, or biometric data ever leaves your laptop. At the end of the week, the leaderboard, optional wager, and optional punishment reset, so a bad week never sinks you.

The flow stays short and recognizable: onboarding introduces the app and its privacy model, you create or join a group with a code, and the dashboard becomes home base. From there a single tap starts a session.

We started with low-fidelity paper prototypes to test the structure before committing to visuals. The red annotations are our own testing notes — combining the group header, moving the leaderboard, and reconsidering the wager controls.

The high-fidelity prototype carries an "academic + playful" identity — clean blue and cream surfaces with a dot-matrix display font that nods to a focus timer. Onboarding now states the privacy model up front, and the dashboard leads with a dominant Start Study button.

Locus rests on one foundation and two primary engines, with optional stakes layered on top. Everything is built so the pressure to focus stays social and honest — never punitive, never invasive.

The foundation — on-device focus AI. A local camera and screen feed analyzes behavior (looking away, phone use, off-task tabs) entirely on your device. Nothing is uploaded. This is what makes every signal below verified rather than self-reported — the honest engine the rest of the product depends on.

The motivational core — the quiet, ambient accountability of studying beside people you actually know.

The engagement layer that turns focus into visible, rewarding progress — tuned so it never shames anyone.

Optional stakes for friend groups who want a sharper edge — never required, and designed to fail gracefully.

Peer pressure restores the library presence that solo apps and blockers lack. Gamification supplies the motivation timers can't, while comeback boosts and a weekly reset defuse the shame that sinks ordinary leaderboards. The on-device AI verifies the focus that timers only assume — and because it runs locally, it removes the surveillance anxiety that makes camera-on apps feel invasive. The optional wager serves students who want real stakes without forcing them on anyone else. Each pillar answers a specific failure we found in the tools students already use.

Three motivational ideas, each chosen on purpose.

Locus stands on three ideas — peer pressure and gamification at its core, with wagers as an optional edge. Each one below says why we chose it over the alternatives, how our research shaped it, and the trade-off we accepted.

Design turned out to be about trust, not pixels.

Working through the full arc — defining the problem, researching, brainstorming, testing, revising, and prototyping — taught us that a product lives or dies on how people interpret its rules, not on how the screens look.

We started imagining an intent-verification gate that interrupts you before you open a distracting app. Research pushed us somewhere more human: social accountability with verified focus, modeled on a library rather than a lockout. Two specific beliefs flipped along the way. We assumed competitive stakes would motivate everyone — testing showed they had to be optional, since some students want to rely on discipline alone. And we assumed an intro screen could carry the privacy story — a single skipped onboarding proved it had to be persistent in the UI.

Defining a fuzzy, adaptive problem precisely enough to design for was the first real challenge. The second was the constant tension between accountability and anxiety: the same leaderboard that motivates one person shames another. The third was making an AI feel trustworthy without over-promising — the moment a wager entered, every gap in the AI's logic became a reason to distrust the whole system.

We'd move from interface polish into the harder engineering: the privacy pipeline around recording, screen analysis, on-device processing, and video sharing. We'd make the AI's reasoning legible, ship the flagged-distraction review tools, and make the wager system demonstrably fair. The biggest lesson is one we'll carry forward — good design comes from iterating on real feedback, and every decision is a trade-off you have to own.