Screenshots



Problem Statement
Using a locally or cloud-provisioned LLM, we have trained a model to make Defi and other crypto transactions easy to do. We use this LLM via a browser extension and monitor when a user interacts with a supported web3 app, if the user doesn't have the asset required to interact with the project or doesn't even have the wallet, the agent will guide the user through the correct steps to successfully download the correct wallets and help move the asset to the chain.
Solution
We've leveraged lots of AI in building this demo, the initial version will start as a tamper-monkey script and eventually be packaged into a full chrome extension. On the LLM side we are using GAIAnet a decentralized LLM for privacy purposes, we can support open-ai or Claude api keys later. This interacts with a provisioned llm in the cloud that has been trained on optimal routes and proposes the easiest and cheapest way to achieve the desired chain interaction.
Hackathon
ETHGlobal San Francisco
2024
Contributors
- chews
2 contributions