I think that traditional IDEs will increasingly become a system for managing agent developers. A big part of this is making it easier to run more agents in parallel while still giving the human developer clear oversight of what’s happening.
I think there’s really two layers to it. The first is to expand how many agents you can comfortably run at the same time, which is what Baton is trying to do. The second thing is that, at some point, the real bottleneck becomes the developer having to manually direct each agent and constantly write prompts for them. That is not the scale.
My guess is that the next step could be to move toward higher-level supervisory agents who sit above individual workers. Rather than managing each agent directly, the developer does more of the work through systems that summarize important information, suggest strategies, and help direct the overall flow of work. I think the cloud and OpenAI will probably make it their own, but I can’t see it happening until coding agents become so strong that you no longer need humans to review code.
<a href