Context is the whole product
The intervention is only persuasive because it can name the assignment you are avoiding. Pulling Canvas, Calendar, and activity data into one model before generating a nudge is what separates this from a timer.
All projects
Best Use of ElevenLabs — Hacklahoma 2026
The Trail
Won Best Use of ElevenLabs at Hacklahoma 2026 — built in 24 hours.
1st
Best Use of ElevenLabs
24h
concept to demo
3
data sources unified
A behavioral nudging app that fights phone distraction by turning your real obligations into timed interventions.
Why it mattered
Screen-time apps fail because they punish you with a blank wall. The bet here was that a nudge lands better when it knows what you are actually avoiding — a specific assignment due Thursday, not "you've used Instagram for 40 minutes."
The problem
Focus apps block apps. They do not know why you picked up the phone. Without context about your actual workload, a blocker is just an obstacle you learn to route around.
What I built
- Apple Screen Time framework (Family Controls, Device Activity, Managed Settings) to detect usage thresholds and render custom shield overlays over distracting apps.
- Gemini API decomposing an overwhelming workload into prioritized, small action steps delivered as in-app nudges.
- ElevenLabs voice synthesis delivering spoken encouragement on goal completion — the piece the judges recognized.
- Aggregation of Canvas assignments, Google Calendar, and fitness data into a single context model.
Engineering notes
Shipping against Apple's guardrails
Screen Time APIs are deliberately restrictive about what an app may observe and block. Working inside Family Controls and Managed Settings — rather than fighting them — was most of the 24 hours.
Curious about any of this?
I am glad to go deeper on the architecture, the tradeoffs, or the parts that did not work the first time. That conversation is usually more useful than the README.