The gap between what AI can do and how people and companies use it is where the next wave of value lies. Closing it is a design challenge.

Every major technology shift expands what’s possible. But realizing its value has always required changes well beyond the technology itself. It involves the people, roles, processes, and organizations.

We are in a similar moment with AI. The systems are evolving extraordinarily fast. The harder work is helping people and organizations adapt so that this technology creates meaningful value.

The gap between what AI can do and how people and companies use it is where the next wave of value lies. Closing it is a design challenge.

We are in a similar moment with AI. The systems are evolving extraordinarily fast. The harder work is helping people and organizations adapt so that this technology creates meaningful value.

The gap between what AI can do and how people and companies use it is where the next wave of value lies. Closing it is a design challenge.

We are in a similar moment with AI. The systems are evolving extraordinarily fast. The harder work is helping people and organizations adapt so that this technology creates meaningful value.

Where I see important work for UX

Designing the new relationship between people and technology.

As systems increasingly interpret, create, decide, and act, UX moves beyond designing the interface to shaping the relationship between people and technology — and the transition to new ways of working.

This means understanding what work should move to machines, what remains human, and then designing the experiences that help people express intent, exercise judgment, stay informed, intervene when needed, and remain accountable for outcomes.

Human–machine complementarity · Inspired by Fitts (1951), adapted for human–AI systems

The new experiences we need to design

Intent

How people express goals, constraints, and preferences.

Delegation

How work is assigned between people and systems.

Oversight & verification

How people understand, review, and validate what the system does.

Error & recovery

How people detect problems, intervene, and get back on track.

Accountability

How responsibility is defined and maintained.

How I think we should lead through the transition

There isn’t a playbook for this moment. As I work through it, these are the principles I keep coming back to: learn quickly, stay grounded in the reality of customers, and hold the work accountable to outcomes as the technology continues to evolve.

Learn at the frontier

AI is evolving too quickly to learn from the sidelines. Building is faster now, which means we can turn ideas into prototypes, test what actually works, and bring that learning back into the work. I believe we need to take advantage of that shorter distance between an idea and evidence — learning by doing.

Meet customers where they are

Technology can move much faster than people and organizations can. Understanding how work happens today — the roles, workflows, processes, constraints, and expectations around it — is essential to designing a transition that people and organizations can actually make.

Treat adoption as transformation

Putting AI into a product is the relatively easy part. Realizing its value can require concerted change across workflows, roles, governance, and operating models. That takes the work beyond traditional UX boundaries and into partnership with the teams responsible for implementation and broader transformation.

Hold the bar at outcomes and value

Experimentation matters, but activity isn’t progress. The bar should remain whether we’re actually improving the work and creating meaningful value for customers and end users — in ways consistent with the values and responsibilities the experience demands.

Invest the surplus in quality and governance

If AI allows us to build and iterate faster, I believe we should invest some of that surplus rather than simply using it to ship more — going deeper into quality, reliability, performance, guardrails, and governance. Speed should give us the capacity to make experiences better, not simply make more of them.

Automation can remove humans
from the execution.

It cannot remove humans
from the accountability.

Human accountability · Diabetic retinopathy screening

Automation can remove humans
from the execution.

It cannot remove humans
from the accountability.

Human accountability · Diabetic retinopathy screening

Automation can remove humans
from the execution.

It cannot remove humans
from the accountability.

Human accountability · Diabetic retinopathy screening

What I’m exploring now

Three areas of focus for me and my team.

Preparing customers for agentic transformation

Understanding how roles, workflows, expectations, and organizations need to evolve as systems begin doing more of the work themselves.

Defining human–AI experience patterns

Exploring the interaction models needed when people articulate intent, delegate, supervise, intervene, verify, and remain accountable for work performed with AI.

Building evaluation frameworks

Defining what “good” means for agentic experiences and building the tooling and intelligence to determine whether AI is actually improving the work and outcomes it was intended to transform.

In the Loop with AI

My thoughts on living, working, and designing with AI.

These are perspectives shaped by my own experiences with AI — building with it, living with it, experiencing some of its magic firsthand, and running into its many limitations. A collection of short observations on what I’m learning along the way.

Designing for the doubt we don’t always have

I’ve been paying attention to how I verify AI output. Specifically, what makes me stop and check, what helps me do it well, and when I simply accept an answer and move on. The harder design problem may be the moments when I should be questioning an answer but I don’t. How do we make verification easy when people seek it? More importantly, how do we create the right moments of friction or prompting when the stakes demand more scrutiny?

COMING SOON

AI is moving at software speed. Can organizations?

AI capabilities are advancing extraordinarily fast. But agentic transformation requires much more than access to the technology. I’m looking at what earlier technology adoption curves might tell us about this moment — and the dependencies that could determine whether organizational transformation takes a few years, a decade, or longer.

COMING SOON

When AI can think for us, what should we keep thinking about?

AI can extend our capabilities into areas where we have little expertise — while also reducing how much thinking we need to do ourselves. I’m interested in what happens to judgment and critical thinking when they are no longer exercised through the ordinary act of getting work done. If technology removed physical exertion from everyday life and we invented gyms, what are the climbing gyms for our minds in a world where AI can increasingly think for us?

COMING SOON

The gap between what AI can do and how people and companies use it is where the next wave of value lies. Closing it is a design challenge.

The gap between what AI can do and how people and companies use it is where the next wave of value lies. Closing it is a design challenge.

The gap between what AI can do and how people and companies use it is where the next wave of value lies. Closing it is a design challenge.

The gap between what AI can do and how people and companies use it is where the next wave of value lies. Closing it is a design challenge.

Where I see important work for UX

Where I see important work for UX

Where I see important work for UX

The new experiences we need to design

The new experiences we need to design

The new experiences we need to design

Intent

Delegation

Oversight & verification