Inspiration

In a world full of passive "ping" notifications that are too easy to ignore, We wanted to build a reminder system that demands attention. The goal was simple: bridge the gap between a digital to-do list and a persistent human assistant. If I miss a important notification, call me.

What it does

Nudge is a productivity app with a multi-stage escalation protocol:

  1. Smart Notifications: Starts with a standard push notification.
  2. Voice Escalation: If the task remains undone, the system triggers a phone call via Twilio.
  3. AI Integration: The call isn't pre-recorded. It uses Google Gemini to generate a dynamic, context-aware script based on the specific task details, acting like a real personal assistant.
  4. Interactive: Users can speak back to snooze or complete tasks directly on the call.

How we built it

  • Frontend: Built with React Native (Expo) for a seamless cross-platform experience on iOS and Android. We used react-native-reanimated for smooth, premium UI transitions.
  • Backend: Powered by Firebase (Firestore & Cloud Functions). Firestore handles real-time data syncing, while Cloud Functions orchestrate the serverless logic for scheduling and triggering calls.
  • AI & Telephony: We integrated Google Gemini 2.5 Flash to generate natural language scripts instantly and Twilio Programmable Voice to execute the calls globally with low latency.
  • Security: Implemented Firebase App Check and phone number verification to ensure system integrity and prevent abuse.

Challenges we ran into

  • Native Module Customization: The default Android date picker didn't match our specific "Dark Mode" aesthetic. We had to write a custom Expo Config Plugin to modify the native styles.xml during the build process to force a black theme.
  • Latency Optimization: Connecting Gemini's text generation with Twilio's voice output required fine-tuning our cloud functions to achieve near-instant response times (±0.05s delay) so the call felt natural.
  • Verification Flow: Designing a secure OTP verification flow that satisfied Twilio's trial restrictions while keeping the user experience frictionless was a delicate balance.

Accomplishments that we're proud of

  • Real-time Architecture: Leveraging our 3+ years of Firebase experience to build a highly scalable, secure backend that handles data ownership and fast queries efficiently.
  • Unit Economics: Designing a cost-effective model where the low marginal cost of Twilio calls (~$0.01/min) allows for a profitable subscription model.
  • Cross-Platform Consistency: Achieving a unified, high-performance UI across both mobile platforms using a single React Native codebase.

What's next for Nudge

  • RevenueCat A/B Testing: Implementing A/B testing for paywalls and localized pricing to optimize conversion rates globally.
  • Advanced AI Models: Testing newer Gemini iterations to improve voice realism and conversational capabilities.
  • Credit System: Introducing a "Power User" tier with call credits to further monetize heavy usage while maintaining fair access for all users.

Built With

Share this project:

Updates