There's No I in Team, But There's a Big One in AI

Everyone using AI right now is getting smarter, faster, and more capable. The question nobody has answered yet is: so what?

That's not a knock on the technology. It's a structural problem. AI is making individuals dramatically more productive — and it's doing almost nothing to make teams more coherent.

Chris and Toby dug into this on a recent episode, and the core tension is worth sitting with for a minute.

The Individual Productivity Trap

When the internet arrived, it created scalable, consistent infrastructure. You built a website, everyone went to the same URL, the experience was predictable. Software did something similar — it let teams hand work across to each other in a structured, repeatable way and capture shared outputs in a common place.

AI doesn't work like that. By design, it generates a unique output every time. That's what makes it feel powerful in the moment. It's also what makes it nearly impossible to operationalize as a team.

What we have right now is a lot of individuals running faster — but not necessarily in the same direction, on the same track.

The analogy Chris used: inventory management. You can have infinite inventory and never miss a sale, but your carrying costs explode. You can minimize inventory and cut costs, but now you're stocking out. The sweet spot requires optimizing the system, not just one variable. AI right now is giving everyone a lever to pull without a shared view of what the system actually needs.

What Actually Makes Teams Work

Pull AI out of the picture for a second. What makes teams effective?

The boring answer: shared visibility. Most teams don't actually know what their teammates are doing. Not in a surveillance sense — in a coordination sense. The Gemba concept from lean manufacturing (go to the place, observe what's actually happening) almost can't exist in knowledge work. You can't watch someone write an email. You can't observe the moment a decision gets made or a shortcut gets taken.

That invisibility is where process debt accumulates. And AI, in its current form, makes individuals more capable of working in that invisible layer — it doesn't surface it.

The Confident Intern Problem

Toby put it well: AI is a confident intern. Enormous information, zero applied experience. It will tell you something is true with the same tone it uses when it's making something up.

Toby ran into this recently while using AI to organize content into a structured series. The output looked clean. The logic tracked. And several of the cited sources simply didn't exist — plausible authors, plausible titles, completely fictional. Not in the early GPT days. Four weeks ago.

That's not a reason to stop using it. It is a reason to stay in the author's chair rather than handing over the pen. The AI gives you a rough cut. The fine-grain work is still yours.

Where It's Actually Working

The use cases that seem to be generating real leverage share a common structure: one person with deep, specialized expertise uses AI to compress and extend that knowledge — and then shares the output with a team.

That's different from everyone individually prompting their way through their workday. It's someone taking months of accumulated knowledge, running it through a model, and producing something the team can actually use together. The AI amplifies existing expertise. It doesn't substitute for it.

The other honest observation from this conversation: if you're using AI to do more, you're probably taking on more. And the system downstream from you hasn't necessarily gotten faster to match.

The Reckoning Hasn't Come Yet

There's a company that recently spent half a billion dollars on tokens in a month without realizing it. Nobody accidentally employs 30,000 people for a month without noticing. The governance question — who owns AI usage, what are we actually buying, what's the ROI — is coming. It just hasn't arrived at most organizations yet.

The individuals closest to the technology are moving into increasingly speculative territory. The people writing the checks are going to need to see the line on the P&L eventually.

What's the practical takeaway? Share how you're using it. Not the hype — the actual use cases, the prompts, the workflows. The leverage from AI compounds when teams start using it in aligned ways. Right now most teams are a collection of people who've each figured out their own workarounds.

That's not a team. That's just a bunch of faster individuals working in parallel.

Process Debt is a podcast about the gap between buying software and making it work. New episodes weekly.