aZ
Zero-knowledge-backed matchmaking, powered by a purpose-built AI agent.
Demo
About
An AI matchmaker verifies each user’s claims with zero-knowledge proofs, so every profile is trusted without revealing the underlying data. It then pairs people (or agents) by analyzing both verified traits and conversational preferences, learning from every outcome to deliver ever-better matches.
How it's made
Self Protocol turns user credentials into zero-knowledge proofs that feed a LangGraph workflow for compatibility scoring and ranking. Running as a Fetch.ai uAgent, the system inherits a public address, mailbox messaging, and full interoperability across the Agentverse network. In order to create the knowledge graph we used neo4j + mem0 + pinecone. Finally we used copilotkit for bringing GenUI to our frontend. It's worth mentioning that we took some inspiration from PEACOK, a framework for building agents that can simulate human behaviour.