Looking at all the latest posts over the past few weeks — and exploring some of the new tools that came out — it feels like designers have been handed a Ferrari. We can whip up UIs, visuals, and wire up flows almost instantly. No surprise that some have already started claiming that design is "solved." But is it really? Are the engine and wheels all that designers, PMs, or engineers ever needed?

Not every part of design was meant to be a race on the track. Speed helps, but it can't replace knowing why you're driving — or where you're going.

When good enough ships, judgment is the only differentiator

Design has always been about the decisions that precede and follow the artefact: what should exist, what shouldn't, what is good enough, and what still needs work based on feedback and judgement.  As production gets easier, these decisions begin to matter more. Reasoning, judgment, and taste are what create the value design was always there to solve for — and more importantly, what will create differentiation in a world that is more than ever converging into a similar-looking mould.

The skills, markdown files while helps us sharing context as much as we can, the AI still (atleast so far) has to pull from everything given in the context scope — every pattern, layout decision, colour call that has ever been documented or you choose to document something that's existing — and surfaces what it understands to be plausibly correct with it. The result is already visible: similar cards, gradients, minimal layouts, and a lot of Instrument Sans. A convergence that isn't entirely AI's fault, but if not used well has potential to be accelerated - resulting in interfaces that lack a judgement call. Interfaces that are functional but forgettable, polished but predictable..

The design phases still exist — and some matter more

Dennis Hambeukers talks about the 'Four layers of Design' in this article which I think augments well with my alignment -

Design is more than aesthetics, functionality, or strategy — it’s a way of solving problems and shaping the world"

While I encourage you to read the full article, I want to draw your attention to what AI is solving for, and what matters more now.

Some of the aspects of interaction design, visual design are already working reasonably well with AI tools, and that's worth being clear-eyed about - especially in the realm of expanding existing tools into known problem spaces. Working within design systems — extending components, maintaining visual consistency, applying repeatable patterns to solve problems — is territory AI handles with growing confidence. The rough mechanics of colour distribution, visual hierarchy, predictable interaction states: decent, improving, will only get better.

But the layers where differentiation actually lives are different in kind, not just degree. What does delight mean for these users — not a generalised persona of today, but a specific generation, context, and set of expectations of tomorrow? What worked for a Gen X enterprise user — density, control, explicit hierarchy — can feel cold and over-engineered to someone who grew up on consumer apps. The inverse is equally true: the lightness that reads as elegant in a consumer product can feel unserious in a high-stakes workflow. That distinction isn't in the training data. It has to be (atleast for now) understood, and then designed through.

There will most likely be still a due process: finding out what is actually worth solving, what shape the solution should take, how to make it genuinely good for people who will spend hours of their working life inside it. Speed will force change but the layers, most of them, will remain. And some become more important precisely because everything else is equalising.

Inventing new may still need a canvas

There is something older than software that designers have always relied on: the hand-eye-mind loop. Throughout history, designers — and artists before them — have externalised ideas on surfaces. Paper sketches, whiteboards, physical models, design canvases, prototypes. When your hand draws something, your eye sees it made concrete, and your mind learns from it and adjusts. This loop is how we discover problems and insights that purely mental thinking misses. A designer's mind is often buzzing with branching ideas — the right medium is what lets them express it while still allowing them to tweak in between, as clarity forms through the act of making.

The tools have changed. The loop hasn't. And the evolution of those tools has been deliberate — each shift extending what the loop could do. We moved from pen and paper to design canvas tools not for convenience but for capability. Canvas tools showed motion. They showed device context. They let you feel how something sat on a phone versus a laptop, how neumorphism implied depth, how glass morphism suggested material. The canvas became a thinking environment, not just a drawing surface.

Now, some of that has gotten dramatically easier with AI tools. If you want to emulate a known interaction style or simulate how a material behaves — glass, metal, fabric — LLMs have strong intuitions about how those things work, because they've learned from everything that already exists. A passable first take is fast.

But "passable" and "invented" are not the same thing.

Designing interactions for environment that doesn't exist yet, new mediums which aren't stress tested with existing patterns yet - requires you to understand what works for users, map the unknowns, ideate on a free canvas if needed, and bring AI in as the force multiplier for implementation either in between or at the end. 

A strong designer mind needs a multi-modal toolkit: sketching, writing, conversation, and yes, prompting AI — all of them will have a place. The mistake is treating generation as the end game. Prompting is one mode of thinking. It is not a replacement for the loop.

The Problem Space Is Expanding, Too

What we are designing is changing too. Products now have two audiences simultaneously: humans who need clarity, trust, and usability, and AI agents who need structure, semantics, and machine-readable logic. Both are becoming intertwined. The patterns that served one won't automatically serve the other.

We will be able to only invent beyond what the model knows if you first think beyond what the model knows. Designers who default entirely to generative tools will find their work bounded by the edges of what already exists. The new problem spaces — spatial computing, agentic interfaces, multimodal experiences — sit beyond those edges.

The Subtlety of Craft

When making gets cheap, functional parity arrives fast. What shows up next — and what users will choose products based on — may well be harder to name but easier to feel. How does it respond? How does it communicate? How delightful is it? What solved a problem without introducing new constraints? Not in a decorative sense — in a "this feels thought through" sense.

Coming up with a coherent tone of voice, new patterns of satisfying micro-interactions, a playful touch of animation — these require taste that was built by spending time with people, a changed environment and products. The taste that knows the first output is rarely the right one. That gap — between the generated and the genuinely considered — is where the future of design lives.

The Monks Who Will Still Design

So who are the Monks who will hold patience in a world of buzz and still design?

To me, it's those deliberate ones - The designers who use AI to move faster but don't confuse speed with arrival. Who asks, before building anything: Does this need a UI at all? Is this the right problem? What are we not questioning? Who trusts the hand-eye-mind loop to surface what prompting can't — and who also knows when to pause the loop, generate something fast, and iterate. Who keeps the full toolkit close: pen, canvas, conversation, code, prompt. Who uses each at the right moment.

The future may well need the patience to find the right canvas; not just where design gets documented and implemented but where design gets discovered.