Rival

sign in
add integrations
challenge history
create new challenge
challenge result
profile: stats

HackCWRU Software Track

Inspiration

Most productivity and accountability tools fail because there is no real consequence for failure. Users can ignore reminders or abandon goals with no downside. At the same time, competition and incentives are proven motivators.

Rival is built on a simple but underused idea: accountability works when incentives are real. The concept emerged from informal productivity bets during late-night study sessions. We saw an opportunity to formalize this behavior into a scalable system that enforces accountability automatically, without social friction or manual oversight.

What it does

Rival is a competitive productivity platform that lets users challenge friends to 1v1 goals with real money stakes.

Users select a measurable objective (e.g., GitHub commits, study activity via Notion, or custom goals), stake equal amounts ($1–$500), and compete over a fixed time window. Progress is tracked automatically using verified third-party data sources.

At the end of the challenge, an AI referee evaluates the collected evidence and determines the winner, who receives the full pot. The system removes subjective judgment, manual verification, and disputes—turning productivity into a clear, enforceable competition.

How we built it

The mobile frontend is built with React Native and Expo, using expo-router for navigation and Zustand for state management.

The backend is implemented using FastAPI with PostgreSQL, modeling users, challenges, stakes, and outcomes. We integrated GitHub OAuth to fetch real commit data and Notion OAuth to track workspace activity. Authentication is handled with JWT tokens and secure password hashing.

The AI referee uses Gemini 2.0 Flash to evaluate challenge completion based on structured inputs from integrated data sources. It produces deterministic verdicts with explicit reasoning, prioritizing transparency and consistency. All components communicate via REST APIs.

Challenges we ran into

OAuth integration across mobile platforms was a significant technical challenge. GitHub and Notion have different redirect URI constraints, which required careful handling with Expo’s custom URL schemes.

Defining and measuring progress for subjective goals was the hardest product problem. Metrics like “studying effectively” are inherently ambiguous. We addressed this by iterating on AI evaluation logic to balance fairness, consistency, and resistance to gaming.

We also encountered mobile networking limitations—localhost access does not work on iOS simulators—requiring explicit IP configuration. Designing a payout system that discourages manipulation while remaining simple and fair required careful trade-offs.

Accomplishments that we’re proud of

We built a fully functional end-to-end system within the hackathon timeframe. Users can create challenges, receive notifications, view real-time progress, and have outcomes resolved automatically by the AI referee.

All integrations pull real user data, and AI verdicts are delivered with structured reasoning. The UI is intentionally competitive and game-like, reinforcing engagement. Most importantly, we demonstrated that AI can act as a trusted arbiter in incentive-driven systems.

What we learned

We learned how to design and debug OAuth-based integrations across mobile platforms and how to deploy AI as a core decision-making component rather than a surface-level feature.

We also learned that transparency is critical when AI is responsible for enforcing outcomes. From a technical standpoint, we improved our understanding of React Native architecture, FastAPI backend design, and building systems that must be fair, explainable, and robust.

What’s next for Rival

We will add push notifications to improve engagement and retention.

We plan to expand beyond coding and studying into fitness, meditation, sleep, and habit-based challenges. Leaderboards and team-based competitions will enable longer-term engagement.

Long-term, we aim to introduce appeal mechanisms for AI decisions and machine-learning-based detection of cheating or adversarial behavior. Our goal is to make Rival a general-purpose accountability layer that can scale beyond productivity.