Think bankrs iterations here are interesting. We started with pure “abi -> tx” back in 2024 but found that the variance in tx outcomes over 100 runs was too high to trust with large sums of money (especially for more complex functions than deposit/withdraw). We’ve ended up giving the agent a transaction building language it could self-define and iterate over, and have found great success here. But its still interesting to think about when you need a skill vs a language/code, and what the minimal version of each is.
deployer
@0xDeployer
03-13
skill files changed the game in terms of what LLMs are capable of out of the box. one could argue that builders won't need to fine tune these models, but can just use skills as frontier models continue to become more powerful.
i think this same pattern could be applied for self
From Twitter
Disclaimer: The content above is only the author's opinion which does not represent any position of Followin, and is not intended as, and shall not be understood or construed as, investment advice from Followin.
Like
Add to Favorites
Comments
Share
Relevant content
