Texture Is Free Now. What a Brand Still Owns

On September 1, Figma quietly closed a chapter that had been open since the first drop shadow. Progressive blur, the effect that used to cost a front-end team weeks of shader code and a designer a lot of patience, now takes one or two prompts to the Figma agent, and the result ships with its own sliders. That is not a marketing line. It is the phrasing Rogie King, a product designer at Figma, used in the company's own post the same day, and it describes a change in the physics of our trade: texture, the thing studios used to sell as craft, is now free.

I have been making brands for long enough to remember when a well-executed grain was a signature. It told a client, without a word, that someone had spent time. This week that signal died, and I think it is the most interesting design news of the year, because it forces a question most studios would rather avoid. When the surface costs nothing, what does a brand still own?

The week gave me three answers, from three unrelated places: a Figma release note, the winners list of a European UX award and a run of eight typeface releases in four days. None of them was about branding. All of them were.

What did Figma actually ship on September 1?

The features themselves arrived at Config in June: shader fills and effects that you describe in words or with a reference image, and generative plugins that you build by telling the agent what tool you need, its behaviour, its controls, its parameters, with no development environment and no plugin API to learn. What shipped on September 1 is the part that matters for brands. Shaders can now be animated and interactive, reacting to the cursor. Plugins and shaders can be published to the Figma Community, or privately to an organisation on the Organization and Enterprise plans. The generated code is visible, and third-party agents can read and modify these tools through Figma's MCP server. Two days later, on September 3, opacity joined the list of things you can alias to variables without detaching a component.

Read that list again as a brand owner rather than as a designer. A visual effect is no longer a file someone made once. It is a small program with controls, a name, a publisher and a distribution channel. Figma's own post is candid about the ambition: the team built a web-components system called PropsKit so that the agent could generate consistent, usable controls for whatever it invents, and it chose prototyping over documentation to get there. The stated vision is a canvas where design and code are the same gesture.

Here is the consequence nobody at Config said out loud. If a riso grain, a gradient map or a field of luminous particles is two prompts away for every account on the plan, then within a quarter every third landing page in your category will carry one. The 2026 trend the design press has been calling the tactile rebellion, the deliberate roughness that reads as human against AI polish, just became a preset. I wrote in April that imperfection had become the biggest trend of the year. Five months later, imperfection has a publish button.

When every studio can make grain, grain stops meaning craft

This is the part I want to be blunt about, because it concerns my own trade. A lot of what agencies have been billing as art direction over the past three years was, honestly, surface: a texture pass, a grain overlay, a bespoke gradient, a motion flourish. Good surface, often. But surface was defensible only as long as it was scarce, and scarcity was a function of labour. Figma removed the labour.

Think about what happened to photography when every phone shipped with a portrait mode. Shallow depth of field did not become worthless, it became table stakes, and the photographers who survived were the ones who had something to say with it. The same curve is starting for visual effects. The effect will be everywhere. The reason for the effect will not.

So the question for a brand is no longer "can we afford this texture" but "why this texture, and only this one". A brand that can answer that in a sentence has something a plugin cannot generate: a decision. A brand that cannot will look, by Christmas, like the Community tab.

What won the UX Design Awards on September 2?

On September 2 at five in the afternoon, the UX Design Awards announced the winners of their Autumn 2026 season: more than 500 entries from 42 countries, 195 nominated projects, 13 awards and three Golds. I read the whole list looking for surface, and I did not find any.

The Gold in the Concept category went to Wayix, an AI navigation system for wheelchair users built by Siyuan Teng, Yiran Zheng and Zejun Wu, with adaptive routing and a layered guidance interface. The New Talent Gold went to Hi:Stories next door, a master's thesis by Adrian Jans, Sarah Müller and Marc Schade on digital formats for teaching the Nazi era to fourteen-to-nineteen-year-olds. The Product Gold went to LINQ, Automata Technologies' lab automation platform, which the jury praised for letting scientists design workflows around scientific intent instead of the physical constraints of the machines.

Look further down and the pattern holds. Comcast won for a plug-and-play WiFi activation flow with a 93 percent customer satisfaction rate. KTM won for motorcycle dashboards designed around glance time, the fraction of a second a rider can afford to look away from the road. Siemens won for a bedside terminal that takes patient fatigue and staff time pressure as its brief. Opus Studio won for a collaborative music-making system that borrows its logic from board games.

Every one of these is a decision about what matters to a specific person in a specific situation. Intent over constraint. Glance time over feature count. None of them could have been improved by a better shader. That is what a jury of 2026 rewards, and I would bet it is what a client of 2027 will pay for.

Why is type the last material a brand can still own?

Now the counter-movement, and it happened the same week. Between September 1 and September 4, Typecache logged eight typeface releases. Kante, by Jakob Fangmeier, graduated from the Future Fonts incubator to a proper foundry, Grim Type, on the 1st. On the 2nd came Epure by Khorash Type, Sleight by Plattner Type and Neurath X with its mono companion by Studio René Bieder, a name that, to my reading, points straight at Otto Neurath and the Isotype pictogram tradition. On the 3rd, HAL Typefaces released HAL Casino as a matched trio of serif, sans and mono, and Formagari released Metra. On the 4th, Miracle Type released Herakles and Outline Online extended Arketa to Georgian and Greek.

Eight releases in four days is not a slow week. It is a market that keeps growing while the tools around it get automated, and I do not think that is a coincidence. A typeface is the one part of a visual identity that an agent cannot conjure in two prompts, because the value is not in the shapes, it is in the thousand decisions behind the shapes, and in the licence that makes them yours. When I argued in June that type had become the new logo, the case was recognisability. This week added a second case: type is the last scarce material. A brand that commissions or licenses a specific face owns something that the account next door cannot download from the Community tab.

Notice also what the releases have in common. Matched families across serif, sans and mono. Script coverage extended to Georgian and Greek. Mono companions for interfaces. These are system decisions, not style decisions. The foundries are selling infrastructure, and the smart brands are buying it.

Should you let the agent build your brand's tools?

Yes, and for a reason that has nothing to do with speed. The most useful line in Figma's September 1 note is the one about private publishing: an organisation can build generative plugins and keep them to itself. That turns a brand system into a set of tools rather than a PDF.

Picture it concretely. Your brand has three sanctioned effects, one grain, one gradient logic, one way of treating photography. Instead of a page in the guidelines that nobody opens, they become three private plugins with locked parameters and a handful of controls, published to your organisation. A freelancer joining the project on Monday cannot produce an off-brand texture, because the only texture tool in their canvas is yours. The guideline stops being a document and becomes a constraint. That is the moment brand consistency stops depending on goodwill.

This is where a studio earns its fee now: not by producing the grain, but by deciding which grain, building the tool that produces only that one, and writing the rule that says when to use it. It is closer to product design than to decoration, which is exactly why the UX Design Awards list reads the way it does. If you want to see what that looks like as a deliverable, it is the logic behind how we scope brand systems: rules first, screens second.

The risk: a canvas that looks like everyone's Community tab

Let me argue the other side properly, because there is one. Publishing is the feature that makes this powerful and the feature that makes it dangerous. The first popular riso shader on the Community will be installed by tens of thousands of designers, and it will look identical on all of their work. Effects propagate faster than taste. We have seen this film: the same three Google Fonts on every startup in 2016, the same gradient mesh on every fintech in 2021. Generative effects will do it in weeks rather than years.

There is a second risk, quieter. Code access means the generated effect can be edited by other agents, which is excellent for engineers and murky for ownership. Who owns a shader an agent wrote from your prompt, refined by a colleague's agent, published to a Community with its own terms? I do not know, and I suspect nobody does yet. A brand that builds its identity on effects it cannot legally fence has built on sand. A brand that builds on a licensed typeface, a written rule and a private tool has built on something it can defend.

And there is fatigue. Prompting a texture into existence is fun the first twenty times. The value of a studio has never been access to the effect. It has been the willingness to say no to nineteen of them.

What to do next week

Dezeen published its September calendar on the 1st, and it is a useful reminder that the physical world has not been automated. Paris Design Week with Maison&Objet runs September 10 to 14, London Design Festival September 12 to 20, Lake Como on the theme "Border" the same days. Go and touch materials for an afternoon. Your sense of what a surface should feel like is the input the agent does not have.

Then, back at the desk, do three things. Write down the three effects your brand is allowed to use and why each exists. Build them as private plugins, or have your studio do it, and retire everything else from the canvas. And look at your typeface licence as an asset on the balance sheet, because as of this week it is one of the few visual assets that still is.

Texture became free on September 1. Judgment did not.

Figma Design tools Brand systems Typography