For most of its history, product design has been organized around the artifact. A screen, a flow, a button, a dashboard, a form, an app. The craft was in shaping a material — pixels, components, copy, motion — until the thing felt right.
Behavior, emotion, and perception always mattered, but they entered through the designer's own judgment. The artifact was the anchor. Everything else flowed toward it. Call this artifact-driven product design.
AI changes the center of gravity. The fundamentals still hold — research, behavior, interaction, taste — but the hard part is no longer choosing what screen to design or what flow to build. Across the stack, production is getting faster, cheaper, and more fluid. Making the artifact is becoming easy. Clarifying the intent behind the artifact is becoming hard.
That is the shift: from artifact-driven product design to intent-driven product design.
Intent becomes the design material
In an intent-driven process, the first design object is not the screen. It is the intent.
The question moves from what we are trying to make to what we are trying to make happen.
Are we trying to increase trust? Reduce hesitation? Help someone take a first step? Make a complex decision feel manageable? Move someone from passive interest to active commitment? Prevent confusion before it becomes churn?
This is where behavior design becomes central.
The most important design question is no longer only “what should the user see?” It becomes “what should the user understand, feel, decide, or do next?”
That question has always lived inside good product design. What's new is that it moves from being one input among many to being the thing the whole process depends on.
The artifact becomes a possible expression of intent. The intent becomes the source.
What we actually mean by intent
Intent, in the way we mean it here, is not the same as imagination. Imagination conjures a thing. Intent commits to a consequence. The first is generative; the second is responsible.
It's tempting to read “intent” as a softer word for vision, idea, or product wish. It isn't. A wish floats above the system it would enter. Intent is grounded inside it.
Intent is systemic intentionality.
Systemic intentionality means being explicit, before you make anything, about who changes, what they do differently, what surrounding systems get touched, what dependencies move, what second-order behaviors get triggered, what the thing displaces, what it normalizes, and what it forecloses.
Because no artifact lives in isolation. Anything you ship enters a larger field — connected products, workflows, agents that read or act on your surface, documentation, metadata, information architecture, data pipelines, customer expectations, organizational norms, the habits of the people downstream of the people you're designing for. Each of those gets nudged, even when you didn't plan to nudge it. An intent that ignores those downstream effects isn't really intent. It's a hope.
A useful working definition: intent is a precise, defensible answer to the question — what change are we causing, for whom, in what context, and with what downstream consequences across the system this enters? If you can't answer that, you don't yet have intent. You have an artifact looking for a justification.
The medium of product design shifts from visual to verbal
Here's the part most teams haven't caught up to yet.
If intent is the new design material, then language is how that material gets shaped and transmitted. Language is how you brief an AI tool. Language is how you align a team. Language is how you specify a behavioral outcome precisely enough that an AI — or another human — can act on it without inventing the missing half.
The key competence of designing is shifting, in part, from visual to verbal.
This is unsettling for a generation of designers whose skill was hands-on shaping of the artifact. A designer who is brilliant with the material but imprecise with words can suddenly feel lost, because the part of the job that used to be implicit — the intent inside the designer's head — now has to be made explicit, in sentences, before the work can move.
It's also the most democratizing shift design has had in a long time. Millions of people who never had the visual training, the tools, or the cognitive habits of a designer have language. If intent is the material, then far more people can now participate in design than ever could before. The studio, the trained eye, the years with the pen tool — these stop being the gate. The gate becomes: can you say clearly what should change, for whom, and why?
That is a much bigger door.
The design process changes in three important ways
In artifact-driven design, teams usually start with a format. We need a homepage. We need a dashboard. We need an onboarding flow. We need a feature. The work then becomes about making that artifact effective.
In intent-driven design, the process starts earlier and deeper. The team has to clarify the desired behavioral outcome before committing too quickly to the form. Three things shift.
The brief becomes more important. A weak brief used to produce weak design. Now a weak brief produces a lot of weak design, very quickly. AI generates endlessly from vague inputs, which makes vague intent dangerous: it creates the appearance of progress while multiplying noise.
The designer's role moves from artifact-maker to intent-clarifier. The question is less “what should this look like?” and more “what are we actually trying to cause?” That requires sharper thinking about motivation, context, timing, friction, defaults, and emotion.
Iteration moves upstream. Teams will still iterate on screens. But more and more, they'll iterate on the intent itself. The first version of the intent is too broad. The second is more specific. The third reveals a hidden conflict. The fourth becomes actionable. “We want users to engage more” is not yet a useful intent. Engage with what, at what moment, after what trigger, toward what next behavior? Intent-driven design forces that resolution earlier.
Judgment is the new scarcity
There's a line Elon Musk used to repeat: prototypes are easy, production is hard. It no longer captures the shift. Neither prototyping nor production is simply easy or hard anymore. Production is getting faster, cheaper, more abundant. Prototyping is too. The real difficulty has moved somewhere else: deciding what deserves to be made, tested, and shipped at all.
There's still an obsession with shipping. But shipping velocity is a sub-optimal goalpost when you don't yet know the impact of what was shipped. Impact is the more sensible metric.
Shipping more is easy to count. Impact is harder to judge.
Because the ability to ship code will accelerate far beyond the rate at which human behavior changes — and that changes the bottleneck.
The constraint is no longer just production. It is adoption. It is validation. It is whether anything you put into the world actually changes behavior in a way that matters. Users don't change at the speed code ships. People stay anchored in existing habits unless something genuinely moves them — a stronger motivation, a new pressure, a clearer reason to switch. The hard part is no longer making the thing. It's creating the conditions for change on the other side of the thing.
Constraints, interestingly, are part of the answer. Squarespace won, in part, by constraining the design space — fewer options, fewer styles, fewer steps, a tighter range of choices — and the result was a higher average quality bar across what got produced. AI will raise the floor on visual quality the same way. But visual quality isn't the next frontier. Sense-making is. Does this make sense to build? Does it make sense to ship? Does it fit the system it enters? Does it improve the ecosystem around it, or just add noise, breakage, and unintended consequences?
When production was expensive, scarcity did the editing for us. Now production is abundant, and judgment is the new scarcity. Intent is how that judgment becomes legible — to a team, to a tool, to the people who'll live with what gets shipped.
The new levers
This shift creates new bottlenecks, and they're not the ones teams are used to managing.
Ambiguity. Teams use words like trust, engagement, activation, confidence, and conversion as if everyone means the same thing. They don't. AI can generate from ambiguity, but it cannot resolve strategic confusion — and it can hide that confusion behind polished output.
Translation. A behavioral intent has to become a product decision. If the intent is to reduce hesitation, does that mean better copy, fewer choices, stronger social proof, a clearer default, a different sequence, or a different interaction model entirely? The same intent can produce many artifacts. The work is choosing the right expression for the context.
Judgment. When production was expensive, teams were naturally constrained. Now the constraint weakens, and the cost of poor judgment goes up. Users don't experience your internal productivity; they experience the things you put in front of them. Every unnecessary experiment taxes attention. Every poorly timed nudge adds noise. Every new surface enters an existing system of habits and expectations.
Validation. If we move faster from intent to artifact, we have to ask harder questions about whether the artifact actually produced the intended behavior. Not whether it shipped. Not whether it looked good. Not even whether people clicked. Did people understand faster? Did they hesitate less? Did they decide with more confidence? Did the intervention improve the system, or just add more activity to it?
Ownership. When the artifact was made by a designer, with a team, inside an organization, it was clear whose name was on it. AI blurs that line. If a flow was generated from a prompt, refined by a tool, approved by a PM, and shipped by an engineer — who has skin in the game when it fails? Who gets the credit when it works? Intent-driven design only holds together if someone owns the intent. Without that, AI-assisted work drifts into a kind of authorless output: fast, plausible, and accountable to no one. The teams that will do this well are the ones that decide, early and explicitly, who is on the hook for the behavioral outcome — not just the artifact.
The designer's work becomes more strategic
Intent-driven design does not make designers less important. It makes their judgment more important.
If AI can produce artifacts quickly, the designer's value moves into the quality of the question, the precision of the intent, the framing of the behavioral outcome, and the ability to judge whether a proposed artifact actually serves the system it enters. The designer becomes less someone who shapes the visible surface, and more someone who shapes the conditions for behavior. That means understanding the user, yes — and also the organization, the market, the business model, the surrounding workflows, the moment of use, the incentives, and the downstream effects.
The artifact still matters. It is no longer the starting point. The starting point is the intended change, with its consequences taken seriously.
Why this matters for behavior design
This is the space we work in.
Behavior design has always lived between intention and action. It asks what people are doing now, what they need to do differently, and what conditions would make that change more likely. In an AI-driven product world, that question becomes more important, not less.
When tools can produce faster, teams need better ways to decide. When prototypes are cheap, teams need better ways to choose what deserves to be prototyped. When shipping is easier, teams need sharper ways to know what is worth putting into the world. When demos seduce decision rooms, teams need someone to keep asking what it would actually take to ship and live with the thing. When artifacts multiply, teams need sharper intent.
The next frontier of design is better sense-making — sharper than the interfaces it produces. What should change? For whom? In what context? Through what interaction? With what evidence? And with what consequences for the larger system?
That is the move from artifact-driven design to intent-driven design.
This in our view is the most important design shift of the AI era.