CanHav uses AI agents to analyze on-chain and off-chain DeFi signals, helping users understand protocol risk in real time. The platform integrates blockchain to improve collaboration between agents.
CanHav is an agentic AI research layer for DeFi. Today, most protocols are judged by on-chain metrics like supply, TVL, and price, while the signals that often drive yield-bearing and synthetic stablecoins sit off-chain. These include funding rates, loan repayments, reserve composition, custodian disclosures, and protocol updates.
CanHav uses autonomous AI agents to track these fragmented signals, combine them with on-chain data, and produce typed, confidence scored risk judgments for assets like Ethena’s sUSDe and USD.AI’s sUSDai. The agents are gated by an on-chain SecurityRegistry, operate only as research tools, and publish their conclusions to public dashboards.
CanHav agents can integrate with dashboards like Dune which show what happened on-chain. CanHav explains what it means.
I started exploring the Arbitrum ecosystem at the Founder House in New York with a different idea, which I realized does not have product-market fit. However, I have kept exploring the ecosystem and once the hackathon began, I decided to study the ecosystem more broadly instead of building in isolation. After researching the Arbitrum Portal and the wider DeFi landscape, the gap that stood out to me was that most important DeFi signals, especially for yield-bearing and synthetic stablecoins, are off-chain and untracked.
This platform was built from scratch during the hackathon and shaped heavily by mentor feedback. One of the calls encouraged me to deepen my Alchemy integration for stronger on-chain data.
Feedback Session # 4 expanded the product from an off-chain research layer into a combined on-chain and off-chain intelligence system. On the same call, I was encouraged to research Dune, which resulted in me integrating it into the platform. On the last feedback session, I was encouraged by Ben to focus in on a problem and not try to do everything. That's why I also took out a DEX that I was building within the platform. This allowed me to focus on the real use case that I had been working on.
N/A