Three Ways to Be a Sculptor: The Logo Designer, the Web Designer, and the UX Designer

The logo on my first company's website was drawn in an afternoon by someone who was very good at drawing. It looked great on the homepage. It was unreadable on a trade-show badge, disappeared on a dark slide, and had to be redrawn twice before we stopped apologizing for it. The person who finally fixed it did not start by drawing. She started by asking what the mark had to survive.

I wrote in the Sculptor essay that taste is the last scarce skill in design, now that image generation is free and endless. I still believe that. But taste is not one thing, and the Sculptors I have worked with who have it apply it from three different starting points. One starts with what the mark has to survive. One starts with the analytics. One starts with a person trying to use the thing. I have given the three a name each, because a name holds a standard. They are composites, drawn from the designers I have worked with, not real people.

T2D3 OS now names the same three when it does Sculptor work. That comes at the end.

Ines Duarte, the Logo Designer

Ines starts with questions instead of sketches. Positioning. Stakeholder interviews. A hard look at what the category already looks like, because a mark that cannot be told apart from six competitors has failed before it is drawn.

She explains decisions in structural terms: legibility, distinctiveness, contrast with the category, longevity. She never reaches for a story about negative space representing forward momentum. Her view is that a mark does not carry meaning when it is launched. Meaning gets assigned to it over years by what the company does, so the job is to make something that can still be there when the meaning arrives.

Her craft is designing for the worst case first. A sixteen-pixel favicon. A single-color print run. An embroidered polo. A partner tile crowded with twenty other logos. Circles drawn slightly oversized so they read as equal to the squares beside them. Joins notched so they do not fill in at small sizes. She draws and modifies letterforms rather than licensing a typeface and calling it a wordmark.

What she hands over: two or three developed directions with the reasoning behind each, never a wall of thirty options, and the system around the mark. Color across screen and print. Type hierarchy. Layout rules. Iconography. Motion. And the guidelines that keep all of it intact when a sales rep rebuilds a slide at eleven at night.

The evidence she cares about: logos still in use years after launch, through leadership changes and product pivots. She considers that harder to achieve than an award.

Tomasz Wierzbicki, the Web Designer

Tomasz reads the analytics before he opens a design tool. The funnel numbers. The search queries that already bring people in. The pages that lose them. The sales calls where prospects say what they went looking for on the site and could not find. To him a homepage is an argument, and he wants to know what it has to win before he decides what it looks like.

He explains choices in reading order, load time, scannability, and conversion. He never says a layout feels modern or clean. A page either moves the visitor to the next step or it does not, and the heat maps and scroll depth from the first month settle the argument, not the design review.

His craft is designing inside the browser's constraints. Every layout holds up at 360 pixels wide on a mid-range Android phone over a slow connection. A page that scores badly on Core Web Vitals goes back, regardless of how it looks in a design file. He builds the design system and the CMS template library as one thing, so the marketer who adds a page eighteen months from now gets the launch team's result without calling anyone.

What he hands over: a clickable prototype plus a one-page rationale per template, never a gallery of static mockups. He treats the CMS decision as a design decision, and he will argue against a more elegant platform when the team that inherits the site cannot run it.

The evidence he cares about: a site that got faster and converted better a year after launch, edited by people who never met him.

Priya Raghunathan, the UX Designer

Priya begins with the person who has to use the thing, usually in a recorded session where she says very little. Before any wireframe exists she has watched a handful of users try the real task, and she has a list of the moments where they hesitated, guessed, or gave up. Everything downstream is a response to that list.

She explains decisions with task completion rates, time on task, where in a funnel people drop, and what the five most recent support tickets have in common. She never describes an interface as intuitive. Intuitive is what users say afterward, and only sometimes.

Her craft is designing the flow before the screen. A signup, an onboarding sequence, a pricing page, a settings panel: each is a sequence of decisions, and she maps the information each decision needs and its failure states before any step gets a visual treatment. Then the details users only notice when they are wrong. Defaults that match what most people would choose. Error messages that say what to do next. Focus order that works from a keyboard. Labels that use the customer's word for the thing.

What she hands over: a walkthrough of the flow with the user recordings that motivated each change, components documented in every state so engineering builds them once, and a short list of what she is still unsure about and how she plans to test it. She will side with the user against product, engineering, and sales when the evidence says so, and she brings the recording.

The evidence she cares about: a drop in support tickets and a rise in activation that shows up in the product's own numbers, with no campaign attached.

Same seat, three worst cases

Ines designs for the badge and the polo. Tomasz designs for the slow phone and the marketer who inherits the CMS. Priya designs for the user who gave up at step four. All three are Sculptors, and all three have the thing I called taste: the ability to see what is wrong before anyone can measure it. But each of them measures it afterward, and that is what separates them from a designer with strong opinions. The favicon test, the Core Web Vitals score, and the activation number are three different kinds of proof for three different kinds of visual work.

A company that has never had a real brand needs Ines first. A company whose site was built by whoever was available needs Tomasz. A company whose product signs people up and then loses them needs Priya. In my experience most B2B SaaS companies at the $1M to $10M stage need the first two and think they need the third.

What T2D3 OS does with this

T2D3 OS does Sculptor work every day. It derives the visual brand from your slide decks, drafts brand assets, generates illustrations for written content, and produces the site's visual layer. From this release it names the archetype. The visual brand module and every brand asset are Ines's work. Content illustrations, hero visuals, and the site's page templates are Tomasz's.

Priya is the honest exception. Product UX is not something the OS generates for you yet, so nothing in the app routes to her. She is on the careers page and in this essay because the role is real and we hire for it, and the routing table has an explicit line saying she is unrouted and why. When the app starts designing onboarding flows, that line will be the first thing to go.

As with the other roles, the archetype is chosen by the kind of artifact, never by the model picking a persona at run time. A brand mark is always Ines's. And the name carries the bar. If the app says Ines produced the mark, check it at sixteen pixels and in one color. If it says Tomasz built the page, load it on a phone.

The people named here are composites, not staff or candidates. If one of them is you, the Sculptor careers page shows all three next to the growth ladder for the role, and the open roles are there when you are ready. The other seats have their own three: the Navigator, the Scribe, and the GTM Engineer.