As an experienced developer working on a startup prototype, I’d like to be able to add more agent commands while my last request is still running, either to be able to add it to a queue (so that I can walk away and have everything built when I come back) or even to be able to alter the agent while it’s still doing work (so provide feedback and guide it, ideally).
I often have another piece of work come to mind while the agent is working, based on what it’s doing and what I’m reading as it “thinks” about it’s work.
For context, I’m letting the agent run with max autonomy and max speed.
Vibe coder here, but I was actually thinking about something on this topic when trying to work on different parts of a large app at the same time (like a branch if you were in a large Enterprise). Would concurrency make sense to allow multiple agents to work on different parts of the app at the same time?
A lot of my prompts have anywhere from 30m-1h 15m run times, it’d be awesome to work on multiple things at once while I wait.
Would that make sense or would the agents bump into each other?
This would be great for user testing for sure though.