Making AI Development a Team Sport

Right now, AI development is a private conversation between one person and one AI agent. The rest of the team can't see it, can't contribute to it, can't learn from it.

The rest of your team doesn't see what happened or why. There's no shared understanding, no opportunity for others to weigh in while it's happening.

I think that needs to change.

The current state

You open your AI coding tool. You have a back-and-forth with the agent. Code gets written, problems get solved. It works, mostly. But it's isolated. Your team-mates don't see the reasoning, the false starts, the course corrections. Knowledge stays trapped in your session.

What makes teams work

Every great team I've worked with had a mix of people. Some dive straight in, break things, iterate fast. Others read the manual first, think it through, move carefully. There are the communicators who translate between technical and non-technical stakeholders. The gardeners who maintain and nurture the codebase over time. The mess makers who prototype rapidly without worrying about cleanup.

When those people work together, they balance each other. The cautious ones prevent the quick movers from creating disasters. The quick movers prevent the cautious ones from overthinking everything. The gardeners clean up after the mess makers. The communicators ensure everyone understands why we're building what we're building.

AI development doesn't have that dynamic yet. It's just you and the agent. One perspective.

Making it collaborative

We need workflows where multiple people can see what's happening. Where one person's conversation with an AI agent is visible to the rest of the team. Where someone else can jump in and say "hang on, have you considered...?" before things go too far down the wrong path.

The technology isn't quite there yet, but this is the shift that's needed. AI development needs to become a team sport, not a solo activity.