ZKPs and Blind Compute for Mobile Consumer Group Gaming Experiences
DIBS! is a mobile app that uses cryptography to create entertaining consumer experiences--play with friends and at parties by using DIBS! to generate random card prompts that send users into a frenzy to complete their tasks accurate as quickly as possible.
The app is built with React Native where users can create or join group games. Users can start and play games in real-time.
We built DIBS! with Privy, Nillion, and Succinct.
DIBS! is a mobile app for consumers that:
Technical Stack Frontend: React Native iOS app for user interaction. Backend: Rust program for verification logic. Integrations: Privy for authentication, Nillion for private inference, Succinct SP1 for proof generation.
Privy Users are able to log in and bypass crypto wallet complexity for earned tokens. This allows users to link their social identity and enables tracking for proof verification.
Nillion SecretLLM processes both the input and prompt in a blind manner, ensuring that we as developers of the consumer app nor Nillion nodes has access to sensitive data of raw text or image. SecretLLM assesses semantic similarity between prompt and user input. The app then sends text inputs to Rust backend.
Succinct SP1 proves the textmatcher program executed correctly and outputs a zk proof thats verifiable onchain. The program ingests the text or image that the user submitted. Proof can then be verified onchain by peers, thus ensuring trust in the consumer app and protecting user data.