The Oasis team introduced Huralya, a framework that lets developers create privacy-preserving AI assistants using confidential smart contracts on the Sapphire runtime.
The key innovation is combining secure on-chain logic with off-chain AI computation, ensuring that user data and model interactions stay private while remaining verifiable.
This approach opens up possibilities like:
Personalized AI with encrypted memory
On-chain coordination between multiple private agents
End-user control over both data and AI outputs
For developers, Huralya demonstrates how privacy and verifiability can coexist in decentralized AI applications.
Full article in case you'd like to deep dive more: https://oasis.net/blog/huralya-private-ai-assistants
Would you build an AI assistant this way? How do you see confidential smart contracts shaping AI privacy?
Top comments (3)
It's great to see a concrete example of confidential compute being used for AI not just DeFi. This framework will be huge for other devs building private applications.
Huralya on Oasis Sapphire is a solid step toward privacy-preserving AI assistants combining encrypted memory, on-chain coordination, and user-controlled data makes decentralized AI much safer and more practical.
Really solid read!! Huralya feels like a real step toward AI that actually respects user data. Combining Sapphire’s confidential smart contracts with off-chain AI compute is such a clean way to keep personalization private but still verifiable.