I used to juggle a bunch of terminal windows. But even then the problem wasn't solved: I wanted my agents talking with each other to advance their work:
For example, I used to do market research with one session, then to branding with another and manually bridge the gaps and fill in the context. Now, my market researcher autonomously pings the brand designer and say: "Hey! I finished the presentation, it looks like this, what do you think?"
It virtually enables any one shotting capability by 10x. With AI agents working both independently as well as together, they practically work on autopilot while you just browse their previews to nudge them.
It took a couple of months to develop this tool and I (we, me and my cofounder) are really hopeful this will unlock great capabilities for non-technical folk especially.
What do you think?