Weekend prototype, asking for the truth: Pact puts Claude in the side panel to think and Fo to act, so an idea gets proposed and you approve the next step. Built it in a weekend for the GSD hackathon at withpact.vercel.app. Tear the pitch apart: what's the weakest part of that promise?
The weakest part is the same one Konstantinos flagged — two agents to learn before the value shows up. The approval step is the real product though: an idea becomes a next step you approved is a cleaner promise than the two-agent setup, and it should lead.
commentslikes