Flexible Studios: Rethinking Org Design for the AI Era (Part Four)

Over the past three articles, I've argued that AI is changing how organizations should operate. Teams need to become more flexible, organize around outcomes instead of functions, and rethink accountability around shared results rather than individual deliverables. Taken together, those ideas point to a different kind of organization, one that is more adaptable, more collaborative, and better equipped to respond to change.

There is one more shift I believe organizations need to make.

AI is changing how decisions get made.

Access to information is no longer the advantage

For a long time, organizations were built on a simple assumption. Information was scarce, so decisions naturally flowed through hierarchy.

AI is changing that. Information is becoming widely accessible, but access to information is not the same as good judgment.

That is where organizations need to evolve.

What used to make leaders valuable was often access. Access to strategy. Access to context. Access to expertise. Access to the right rooms and the right conversations. But AI changes that equation. It makes information easier to find, synthesize, and package into something that looks complete.

And that is where judgment becomes more important than ever.

A conversation that changed my thinking

Recently, I had a conversation with a team experimenting with AI to assess the accessibility of a product experience. They had built a custom skill using publicly available accessibility standards, along with some additional guidance they had created internally. The results looked incredibly promising.

The AI generated a polished report, highlighted accessibility issues, calculated color contrast, and even assigned percentage scores to different parts of the experience. It looked like exactly the kind of output you would want to share with stakeholders to demonstrate progress.

Then we started asking questions.

One section of the report included a confidence score for an accessibility assessment that was not represented anywhere in the skill's underlying knowledge. We traced it back through the prompts, the documentation, and the source material. There was nothing there. The model had simply made it up.

What struck me was not that it hallucinated. We know AI does that. What struck me was how convincing it was. The report looked polished. The language sounded authoritative. The numbers appeared precise. Unless someone had taken the time to validate the underlying evidence, most people would have accepted it as fact.

That conversation reinforced something I have been thinking about throughout this series. AI does not just change how work gets done. It changes where judgment needs to happen.

Moving faster than our operating models

Right now, AI is one of the hottest topics in every organization. Teams are experimenting. People are building custom GPTs, creating AI skills, generating reports, and demonstrating workflows that look remarkably sophisticated. Many of those experiments produce outputs that feel finished.

That is where the danger begins.

Too often, polished outputs are shared up the chain as evidence that a particular solution works. Decisions begin forming around them before anyone has validated the assumptions, verified the evidence, or understood the limitations of the underlying model.

We are still building the systems, governance, checks and balances, and validation frameworks needed to create AI solutions that people can truly trust. That does not mean we should stop experimenting.

Experimentation is exactly what we should be doing. But we should also recognize that we are still early.

Flexibility is about more than teams

This brings me back to the central idea of this series.

When I argued that organizations need to become more flexible, I was not simply suggesting that we should move people between projects more efficiently. The bigger opportunity is to build organizations that can adapt to the different stages of AI-assisted work.

Not every stage requires the same expertise. AI can accelerate exploration and generate possibilities. Designers apply craft to shape meaningful experiences. Researchers validate assumptions. Engineers verify technical feasibility. Domain experts confirm accuracy. Leaders weigh strategic tradeoffs.

The operating model itself needs to flex around decision points, not just project phases.

Organizations designed around static functions and rigid handoffs struggle in this environment. Organizations designed around adaptable teams can respond as confidence changes, as risk increases, and as new evidence emerges by bringing the right expertise into the process at exactly the right time.

From product design to organizational design

I believe this is one of the biggest opportunities for UX in the AI era.

Not because designers will spend all day prompting AI. And not because design craft becomes less important. If anything, craft becomes more valuable. As AI lowers the cost of producing interfaces, the quality bar rises. The ability to create experiences that are intuitive, thoughtful, accessible, and emotionally resonant becomes an even greater differentiator.

But great designers will also need to operate at another level.

We need the judgment to ask whether we are solving the right problem. We need the judgment to recognize when an AI recommendation lacks evidence. We need the judgment to know when automation should stop and human expertise should take over.

And beyond individual judgment, designers have an opportunity to shape how organizations make decisions.

We can design systems that expose evidence instead of hiding it. We can make confidence visible instead of assumed. We can define validation points throughout the process. We can create checks and balances that ensure AI supports human decision-making rather than quietly replacing it.

Designers have spent decades designing interfaces for users. Increasingly, we'll also design decision-making systems for organizations.

That is no longer just product design. It is organizational design.

Designing organizations for judgment

The organizations that succeed with AI will not simply be the ones with the most advanced models. They will be the ones with the most mature operating models.

Organizations where the right people are involved at the right moments. Organizations where AI accelerates work without bypassing accountability. Organizations where evidence matters as much as output. Organizations where confidence is earned, not generated.

AI may automate more work than any technology we've seen before.

But the organizations that outperform won't simply be the ones with the best models or the fastest automation.

They'll be the ones that know exactly where human judgment belongs.

Designing organizations around that judgment may become one of leadership's most important responsibilities.