Part 3: Who Owns the System When Nobody's Clicking?

Parts 1 and 2 described what's changing and what the work becomes. This piece is about what happens if design doesn't move to claim the role that's opening up — and what it looks like if it does.

---

The comfortable version of this conversation ends with, "UX will still matter, it'll just look different." That's not enough. The honest version is harder.

Either UX expands into something closer to system governance than product design, or it contracts into a finishing service: polishing the parts of the interface that still exist. That's not a good position to be in. It's also not inevitable.

The question driving my thinking isn't "what will interfaces look like in 2035." It's "who shapes system behavior when there's no interface to point to." The answer determines what happens to this profession.

Two futures, not one

I've been tracking two distinct trajectories for the past year. Both are plausible, and both may happen simultaneously across different domains and timelines.

The first is the Delegated World. Software becomes an execution layer for intent. Users specify goals. Systems decompose those goals, act, and return results. Human work moves up the stack, from operating tools to supervising outcomes. The interface recedes into the background. It still exists for exceptions, for moments when the system genuinely needs human judgment. But it is no longer the primary surface of the software.

In this world, software has effectively disappeared into the plumbing. You do not navigate it. You instruct it. The interaction is less like using a product and more like managing a relationship with something that acts in your name.

The second is the Trusted Mesh. AI capability expands, but so does the accountability apparatus around it. Users demand legibility: why did this happen, what did it use, can I undo it. Regulation enforces transparency. Ambient computing becomes tightly permissioned. Software expands into the physical environment, but every action is auditable and every permission is revocable.

In this world, the interface does not disappear. It proliferates: voice, screens, spatial layers, ambient cues. The UX problem is not designing one interface. It is maintaining a coherent, trustworthy experience across a mesh of modalities and contexts.

These are not competing predictions. Different domains will land in different places. Healthcare will look more like the Trusted Mesh for a long time, because the accountability stakes demand legibility. Some consumer domains may move faster toward the Delegated World. Enterprise productivity is already trending that direction. Both futures are real. Both may be happening in your organization right now, in different product lines.

The dangerous middle

What worries me about both futures is the same thing: the loss of human agency.

In the Delegated World, the risk is that delegation becomes surrender. Users hand off decisions they should own. Systems act in ways that aren't legible. When something goes wrong, nobody knows why. The system did it, and nobody designed accountability into it.

In the Trusted Mesh, the risk is subtler. The architecture of trust becomes a UX feature instead of a real constraint on system behavior. Clean audit logs nobody reads. Provenance indicators that look reassuring but do not actually tell users what they need to know. The appearance of legibility without the substance.

In practice, that is a badge that says "Based on 4 sources" with no way to see what those sources say or what the model pulled from them. The audit trail is there. The understanding is not.

Both failure modes are design failures. Not engineering failures, not policy failures. Design failures, because someone had the opportunity to make human agency real and legible and did not.

Who shapes this if designers don't

Here's the uncomfortable part. If UX doesn't grow into this space, someone else fills it.

Lawyers will define accountability in terms of liability, not lived experience. Engineers will define system behavior in terms of what's technically feasible, not what's humane at scale. Product managers will define outcomes in terms of metrics, not agency. None of those lenses are wrong. They're just incomplete.

The lens missing from most of these conversations is the one UX has always held: what does this mean for the person on the other end. What is it like to interact with a system that acts in your name. Do you understand it. Can you recover when it gets something wrong. Does it leave you more capable or more dependent.

That's not a screen design question. But it's absolutely a design question. And right now there's a vacuum where the answer should be.

The skills that survive

If the tools change, the platforms change, and the deliverables change, what's left?

Judgment. Knowing when to trust the system and when to override it. Knowing what should not be built at all. This is not teachable through frameworks. It comes from accumulated exposure to how systems fail, from strong opinions grounded in evidence, and from the willingness to say "this is wrong" when the business case says otherwise.

Systems thinking. Understanding second- and third-order effects. A product that works well for individual users while creating harm at scale is a design failure, even if the individual interactions are excellent. The orgs that miss this will miss it loudly and publicly.

Taste. This sounds soft until you work in a world of infinite generation. When AI can produce anything, the question becomes what should exist. That is a curatorial judgment. It requires a standard. Designers have historically been the people who hold and defend that standard. In a world of abundance, that function becomes more important, not less.

Narrative and influence. The systems I'm describing are invisible. You cannot show a stakeholder a screenshot of how agent delegation works. The closest thing you have is a decision tree on a whiteboard: here's what the system handles, here's what requires a human, and here's where control transfers. That is the artifact. Most stakeholders have never seen one, and most designers haven't built one. You have to explain it, build alignment around it, and make the invisible legible to the people making decisions about it. That's communication work. It's political work. It's absolutely design work. Designers who can't do this will find their influence bounded by what they can point to on a screen.

Ethical reasoning. Not the performative kind. The kind that holds up under pressure, when the business case points one direction and the right thing points another. The governance frameworks being built now, from the NIST AI RMF to the EU AI Act, formalize accountability in ways that land on product teams. Someone has to bring genuine moral reasoning to those conversations. A compliance checklist is not enough.

The long bet

In 2035, the designers who matter will not be defined by tool proficiency or a portfolio of screens. They will be defined by their ability to shape how systems behave: how they earn trust, how they communicate their limits, and how they preserve human agency in environments where software acts without being asked.

The dangerous scenario is not AI replacing designers. It is designers staying in their lane while the lane becomes irrelevant.

As AI absorbs the execution layer, the interface stops being the default assumption. It becomes one surface among many. UX either evolves into the function that governs how systems behave on behalf of people, or it contracts into finishing work on whatever screen is left.

That is the choice. It is not rhetorical.

I am betting on expansion. Not because it is inevitable — it is not — but because the alternative is a profession that had the right skills at the right moment and chose to stay in its lane anyway. I do not think that is what UX is.

The window is open. It will not stay that way.