Flare Hackathon showcases Verifiable AI on Blockchain with Google Cloud Confidential Space
One of the key challenges in the blockchain space is achieving data richness and complex computation while maintaining verifiability. Blockchains are optimized for deterministic execution and decentralized consensus, but they lack native support for large-scale parallel processing, in particular high-performance AI workloads.
The Flare x Google Cloud hackathon, held in collaboration with Blockchain at Berkeley from March 7-9, brought together student developers and researchers from institutions including UC Berkeley, UWaterloo, and ETH Zurich to explore how blockchains can support AI-driven computation without sacrificing verifiability.
Participants utilized Google Cloud Confidential Space, a secure enclave using Trusted Execution Environments (TEEs) that provides hardware-enforced confidentiality while generating cryptographic attestations. These attestations are verifiable on the Flare blockchain, ensuring that computations were executed correctly and without external interference. The enshrined data protocols on Flare, Flare Time Series Oracle (FTSO) and Flare Data Connector (FDC) are uniquely suited to ensure data provenance and integrity verification for AI.
The hackathon saw over 460 participants across four tracks, resulting in verifiable AI solutions spanning AI x DeFi (DeFAI) to Consensus Learning. All projects utilized Confidential Space instances running on the AMD Secure Encrypted Virtualization (SEV) platform with virtual Trusted Platform Module (vTPM) attestations. These attestations served as cryptographic proof that computations occurred within an isolated and tamper-proof environment.
”We're thrilled to see developers using our Confidential Space secure enclaves to build trusted AI solutions. Projects like 2DeFi and the Flare AI Kit's launch are exciting steps towards a future where AI and blockchain ensure data integrity and computational assurance." - Rene Kolga, Product Manager, Google Cloud.
2DeFi: Seamless TradFi to DeFi Transition with AI
One standout project, 2DeFi, was developed by Alex and Hitarth, two students from the University of Waterloo. Their project won 1st place in the AI x DeFi (DeFAI) track and 2nd place in the RAG Knowledge track.
2DeFi tackled the challenge of bridging TradFi users into DeFi on Flare using Google Gemini AI. The project allowed users to upload screenshots of their Robinhood portfolios, which were analyzed by Gemini 2.0 to assess their risk tolerance. The AI-generated risk profile was then mapped to an automated DeFi strategy, incorporating staking and liquidity provision on Flare.
To further enhance user experience and onboarding, 2DeFi integrated embedded wallets that enabled Google login-based wallet creation on Flare, eliminating the complexities of traditional Web3 onboarding.
Flare AI Kit: Empowering Verifiable AI Agents on Blockchain
The hackathon also marked the official kickoff of Flare AI Kit – an open-source SDK for building verifiable AI Agents on Flare using Google Cloud Confidential Space. The hackathon submissions leveraged Flare AI Kit, showcasing its potential as a foundation framework for AI-powered blockchain applications. Designed to simplify AI integration while ensuring verifiability, provenance, and AI safety, Flare AI Kit provides developers with secure, attestable AI execution environments and a robust security framework.
The Flare x Google Cloud Hackathon was just the beginning. As developers and researchers continue to explore the intersection of AI and blockchain, Flare AI Kit will evolve to become a powerful, production-ready framework for verifiable AI applications. With ongoing improvements to its functionality, usability, and integrations, the goal is to refine Flare AI Kit into a turnkey solution and bring it to the Google Cloud Marketplace, where it can serve as a standalone, production-ready solution for enterprises and developers alike.
By bridging Confidential Space TEE, AI, and the Flare blockchain, Flare AI Kit aims to redefine how verifiable AI applications are built, deployed, and trusted on a global scale. Ready to start building on Flare? Explore the Flare Developer Hub for in-depth documentation and resources.