When Intent Becomes the New UI

What tools like Figma Make or Replit are really changing in design practice

When tools like Figma Make or Replit enter the workflow, the real shift in design work is not just speed or automation, but a move from drawing screens to defining intent. AI tools make interfaces cheap to produce, which forces teams to confront a harder question: what is this system actually for, under which conditions does it make sense, and what trade‑offs are we willing to encode?

I was reading Tina Singh’s article, ‘What actually changed in my design work once AI tools entered the workflow,’ and this framing stuck with me:

“Most articles about tools like Figma Make or Replit circle around the same two themes: speed and ‘democratising‘ design. Screens appear almost instantly. Variations multiply. But those metrics, impressive as they are, miss the more interesting shift that’s actually happening in our practice…”

The bottleneck has quietly moved from execution to articulation.

What Tina surfaces is exactly right. And this shift — from execution to articulation — is where intent becomes the new UI.

The real change happens earlier, in that quiet moment when you click a button, watch a full screen materialise, and realise that making it is no longer the hard part. The hard part is answering why it should exist in the first place: what it enables, what decisions it supports, and under which conditions it genuinely makes sense. AI has made outputs easy, but in doing so, it shines a light on where clarity, purpose, and alignment are still missing.

This is the moment where intent stops being a nice-sounding principle on a slide and becomes the actual work of design. In my book, The Anticipatory Design Playbook, I validate this shift: intent sits at the core of autonomous and anticipatory systems, shaping both workflows and algorithms rather than being reverse‑engineered from whatever UI already exists.

The Anticipatory Design Playbook by Joana Cerejo

From drawing screens to articulating purpose

Taking value judgements out of it, for years, many design processes started with the interface. Even when the problem was unclear, the reflex was to open Figma, drop in a frame, and hope that clarity would emerge as the pixels took shape.

Tools like Figma Make or Replit disrupt that reflex. When a system can generate layouts, components, and flows almost instantly from a prompt, starting with the interface alone starts to feel oddly hollow. Screens can now appear faster than teams can explain why they matter.

This is exactly where The Anticipatory Design Playbook starts its practical work: with the question of what a screen or system is meant to enable, what decisions it supports, and which assumptions about users, data, context, and time are baked into it. AI cannot do that thinking for us; it simply exposes the gaps when that thinking has never happened.

Intent as a practical design tool

Intent is not an abstract value statement; AI has made that impossibly clear. To design responsibly with AI, intent has to become operational: a tool that shapes flows, automation rules, and system behaviour from the inside out.

Why intent can’t stay abstract? When intent stays at the level of principles or taglines, optimisation defaults to whatever is easiest to measure — clicks, time-on-page, or short‑term conversion. AI systems expose every gap between what teams say they value and what their systems actually optimise for, because behaviour is driven by data, objectives, and reward functions.

To design responsibly, intent must be translated into explicit levers that shape the system from the inside:

Objectives and constraints: Defining which outcomes are optimized (e.g., long‑term user success, safety, fairness) and which are explicitly off‑limits or capped (e.g., attention extraction, risky recommendations). Reward and error trade‑offs: Encoding what “good” looks like in metrics such as precision/recall balances, acceptable error types, or thresholds for intervention. Guardrails and policies: Specifying when the system should slow down, ask for confirmation, hand control back, or refuse to act. In the book’s three-layer model, intent sits at the centre, workflows translate that intent into actions, and algorithms power those workflows with prediction and adaptation. When intent is explicit, assumptions surface early, system behaviour can be discussed and challenged, and teams move from “producing more screens” to “making better decisions.”

How does intent shape flows and rules? Once operationalized, intent becomes a design material that informs:

Flows: What the system surfaces by default, when it anticipates, and when it waits, including how much friction is added or removed at critical moments. Automation rules: When to trigger automation, when to escalate to humans, and how to adapt behavior as confidence and context shift over time. System behavior: How models personalize, what they filter out, how they explain themselves, and how they allow users to correct or override outcomes. In this sense, intent has to be treated like a system primitive — defined, parameterized, and monitored. Only when intent is made operational in data choices, model objectives, and interaction patterns can AI systems reliably behave in ways that match what teams claim to stand for.

AI tools like Figma Make amplify this dynamic because they do exactly what you ask, not what you meant. Any fuzziness in intent shows up immediately as awkward flows, misplaced emphasis, or interactions that look polished but feel empty once you put them in context. That discomfort is not a bug; it is feedback on your intent.

The Anticipatory Design Playbook leans into that feedback loop, offering methods for clarifying user intent, aligning workflows with long‑term goals, and using foresight techniques like forecasting and backcasting to ensure systems don’t just respond to current behaviour but also support where users are trying to go next.

The tensions AI refuses to hide

Working at the level of intent is demanding work. It requires conversations many teams have been able to avoid: about what “helpful” really means, where automation should stop, and which outcomes the organisation is actually optimising for.

AI tools speed up production, which means they also speed up misalignment when intent is unclear. Designers suddenly find themselves navigating organisational resistance, fragmented tooling, anxious stakeholders, and the emotional labour of explaining why thinking still matters when outputs are almost free. Tools like Figma Make won’t solve this; if anything, they make it impossible to ignore.

But visibility is not harmful: It’s a starting point. When misaligned intents and unspoken assumptions are visible, they become testable. When system behaviour is generated quickly, teams can use that speed to prototype not just UIs but “possible futures” and then course‑correct. The playbook frames this as moving from reactive interfaces to automated and anticipatory systems: systems that understand why behaviour happens, not just what happens next.

What actually changes in the designer’s role

So what truly changes when tools like Figma Make become part of everyday practice is not that designers become less necessary, but that the centre of gravity of their work shifts. Less time goes into manually producing artefacts; more time moves into defining, communicating, and stress‑testing intent.

The book is both a response and a companion to this shift. It gives designers, product leaders, and technologists a shared language for working at the level of intent, a framework (anticipate, imagine, shape) for structuring that work, and concrete methods for turning clarified intent into workflows and AI behaviours that remain transparent, adaptive, and human‑centred over time.

The promise of these new tools is not just that systems get built faster, but that they can be built more clearly: aligned with what people are actually trying to achieve, not just with what the interface makes easy to click.