DeFAI with Mantis

In this session hosted by The BLOKC, an ecosystem growth engineer from the Mantis development team unpacks DeFAI, the intersection of decentralized finance and artificial intelligence, with a focus on cross-chain intent execution. The talk draws on patterns from CowSwap, 1inch Fusion, Uniswap X, and the Inter-Blockchain Communication (IBC) protocol.

What the session covers

Standard automated market makers force users to manually route trades across chains. An intent-driven model flips this: users declare the end state they want, and a decentralized network of Solvers competes to deliver the best execution. Mantis is a specialized processing rollup built for intents. Its routing algorithm treats every decentralized action, whether a swap, a loan, or a yield position, as a mathematically provable token exchange. Execution happens cross-chain over the IBC messaging protocol, with settlement landing natively on both Solana and Ethereum.

What you can build

Because Mantis separates how an action is expressed from how it is executed, developers can attach conditional logic to localized LLM agents, for example, buy token Y when influencer X tweets it, then sell at 5x. When the condition fires, the agent translates the command into a programmatic intent and solvers execute it cross-chain, with front-running protection built in. This opens the door to autonomous AI trading agents.

What’s next

The roadmap includes Dice, a user terminal for building customized AI agents that run programmatic limit and stop-loss orders across Solana and Ethereum. Mantis also plans to open-source its localized LLM framework so the community can build financial AI applications directly on the Mantis intent layer.