Autonomous Cross-Chain Intent Solver

Demo

About

The problem it solves The system brings academic idea into real world by improving it further. The idea was published in Proceedings of the 2024 IEEE International Conference on Blockchain (Blockchain): "From CoWs to Multi-Chain AMMs: A Strategic Optimization Model for Enhancing Solvers” Zeshun Shi∗ , Sydney Sweck† , and Omar Zaki† ∗ Cyber Security Group, Delft University of Technology, the Netherlands † Composable Foundation, Zug, Switzerland. The current DeFi stack is missing a true **foundational layer** for intelligent, cross-chain liquidity orchestration. a) Today, liquidity is split across dozens of chains and isolated pools. b) There’s no unified “brain” that decides when to match directly (CoWs), when to fallback to automated liquidity (CFMMs), or how to bridge leftover orders securely. - 🔄 Users or protocols have to stitch this logic together manually — wasting gas, risking failed swaps, and suffering slippage. The system solves this by acting as an intelligent *foundational layer*:** - It unifies CoWs, CFMMs, and Chainlink CCIP into one system. - It dynamically runs Linear Programming optimization to maximize match rates and minimize costs. - It learns and adapts — so the base layer keeps improving by using ElizaOs. **It turns theory of *strategic optimization* into a real, practical, adaptive *foundational layer* for the next generation of cross-chain DeFi.**** It solves following problems: 1) Fragmented liquidity: Liquidity is split across many blockchains — no single chain has enough to always match intents directly. 2) High cost of failed transactions: If there’s no direct match, users lose gas fees or suffer high slippage. 3) Manual bridging is risky: Users have to manually bridge assets, risking exploits or bridge hacks. 4) No unified solver: Current DEXs and CoW protocols don’t dynamically decide when to match CoWs vs when to fallback to CFMMs across chains. 5) No strategic learning: Solvers today don’t adapt — they run the same static strategy, wasting gas or liquidity if conditions change. 6) End users doing swaps: better prices, lower fees, no manual bridging. 7 ) Intent Batching - Multiple intents optimized together for better efficiency

How it's made

Challenges we ran into 1) Running time is quite high when intent settlement was to be done through ElizaOS techniques. 2) Optimization technique algorithm Mixed Integer Linear Programming ((Java Script LP solver) was taking lot of time to process intents. 3) Integrating with Avalanche Fuji network was having issues such as ENS issues. 4) A lot of time was consumed to get intent matched due to multihop path finding through AMM in MultihopRouter.js 5) Brute force pairwise matching in cowOptimizer.js is challenging because it has high time complexity .