Everyone on my team uses AI now. Claude, Cursor, ChatGPT. Alone, each of us moves fast. Together, we got slower.
The problems I keep hitting:
- All the context lives inside each person’s private AI chats. Nobody else can see it.
- Handing work to a teammate means explaining everything again to a new AI, from zero.
- We repeat work, because nobody knows what the other person’s AI already did or tried.
Today we fix it with chat messages and copy-paste. It eats time every day.
How does your team handle this? A rule? A shared file? A tool? I want real setups, not theory.
Hi Jazz!
We had some similar issues, but I think there’s some easy things you can implement to help.
First off, make use of Plan mode and have it persist your plans to the repo. This will help build some context right into the repo. Also, if its not already, have your agents document what it builds, almost like a developer guide.
Anything pertinent that needs to be done a certain way should become a skill/rule. Any repeated processes, you might consider a Cursor command. That should help with bullets one and two.
For bullet three, that seems less like an AI problem and more like a PM or management problem. Does your team use any sort of PM tool like Linear to assign work? When I;m planning a new project for the team, I use AIs plan mode with given requirements, then when the team is satisfied with the plan, I have it create linear tickets for me based on the plan. If you told it you want to logically split the work, I bet it would do a pretty good job.
I hope this makes sense.