[codingdave]

The idea is solid. The challenge I foresee is giving the AI agents access to your systems to make changes. Most enterprises want their WIP behind firewalls, so how do you anticipate securely providing the AIs access to private networks?

[shahahmed]

thanks for checking it out. i imagine 2 things, 1 - any edit actions require a "human in the loop" approval. 2 - for sensitive info, internal instances of llms or only local models accessing info