Article
People using Claude and ChatGPT on their own is one thing. Trying to bring that work together across a team can become a massive mess.
The more I work with companies implementing second brains, the more I notice how much discipline the shared version takes. By a second brain, I mean a place to keep knowledge that AI can help people work with. Once a group starts using it, someone has to think through how that knowledge is organised and how everyone connects to it.
Training employees is a big job in its own right. But it leaves some decisions unanswered.
The engineering work is still there
We used to write software with code that followed defined instructions. AI can now do some of that work for us. We still have to think through the problem we're asking it to solve.
With a shared second brain, that includes decisions such as:
- How should we structure the data?
- Where should we store it?
- How do we keep it secure?
- How does the second brain connect to everyone else?
Those are engineering problems. They remain engineering problems when the people using the system aren't used to making those decisions.
It's easy to focus on getting people comfortable with a tool while leaving the shared structure unsettled. I think that's part of why bringing individual AI work together gets messy. You have to decide how the pieces fit, as well as teach people how to use them.
Give someone the job of driving it
From the cases I've seen, appointing a single person to drive the initiative goes a long way.
That person needs to be able to make decisions about the organisation of the shared system. Otherwise, there are too many chefs in the kitchen, each trying to shape how it works.
If you're bringing your team's AI work together, name the person who will drive those decisions. Give them the scope to work through the data structure, storage, security and connections, alongside the employee training.