The redesign brief that says "make it modern" is lying to you

The AI headlines keep pulling your attention up and away from the work. But the places your team actually wins or loses this quarter are older and plainer than that. A redesign brief. A toggle in the header. The confidence a visitor feels on a page. Let me catch you up on where the real fights are hiding.

The deep cut

• Craft basics decide the product, not AI features. Across these pieces, the wins live in positioning, typography, and trust, not model count.
• A redesign brief names a decision, not a look. The Branded Agency starts every project by asking what choice the site should help someone make.
• A toggle has to earn its spot. Lea Verou found persistent dark mode switches crowding developer sites more than user needs.

The brief is lying to you

The words in a redesign request are symptoms, and treating them as instructions is how you ship a prettier version of the same problem. The Branded Agency translates them for you: "make it modern" often means outdated positioning, "simplify it" means no one has named the priority audience, "improve conversion" means the offer has no proof.

Their fix is one question before any wireframe: what decision should this site help the right person make? That shifts the room from preference to purpose. It also surfaces the uncomfortable truth that many design problems are unresolved business problems you inherited.

Then sequence claim, meaning, and proof. "Faster" invites "than what?" "Trusted" invites "by whom?" Put the evidence next to the claim where doubt happens, not on a lonely testimonial page. If an unsupported claim can't be backed, the claim changes.

Trust is a typeface, not a tagline

In biotech and pharma, the audience is often a partner or investor who has never seen the technology. The design job is to make something complex feel navigable, and typography carries more of that than teams admit. Zajno's team calls Helvetica the "little black dress" of type: a neutral base that signals credibility without showing off.

The lesson travels past biotech. A restrained system looks less striking than a bold one, but it reads as expertise. You build character through weight, scale, rhythm, and spacing, not through a weird typeface. Pick clarity over the clever option, and test with real content volumes, not polished mockups.

The button nobody asked for

Here is a smaller fight with the same shape. The sun-and-moon toggle became near-mandatory, sitting in the header next to things people actually came to use. Lea Verou noticed those persistent toggles cluster on developer-facing sites, which raises a fair question: did designers add it because users need it, or because other designers have it?

Browsers have detected system light-or-dark preference for years. You can respect that automatically and tuck an override somewhere less prominent. Every button you add asks for attention. The dark mode debate is really the discipline to ask whether a pattern earns its seat, and to say no when it doesn't.

When AI shows up, visibility is the feature

AI changes the flow of a product. The old path was action, response, next action. The AI-era path runs intent, interpretation, recommendation, review, outcome. That adds plain UX questions, not engineering ones: what shows when the system is unsure, how an action gets explained, when to ask for confirmation.

The rule worth repeating in a review: automation should reduce effort without reducing visibility. Show what the system understood and make corrections easy. A confidence label with Confirm and Edit buttons does more for trust than another feature.

That trust problem has a flip side the data people rarely name. One writer describes the "homo metricus," a user who reads themselves through app scores, slowing before their real limit because a screen told them to. The number tells you what can be measured, not what is good. Design that moves judgment out of the user is a choice, so make it on purpose.

The career fight your team is already having

Under all of this sits the worry your people carry into one-on-ones. One essay argues "no one is coming to retrain you", and backs it with history: handloom weavers watched wages fall from 240 pence a week to under 100 in about a decade while the abundance arrived two generations later.

The useful part for you is that the famous question, name the new jobs, has never had an answer. MIT's David Autor found roughly 60 percent of today's work sits in titles that didn't exist in 1940. Your job is not to predict the roles. It is to keep your team doing the craft that holds value when the tools change: naming decisions, making trust legible, cutting what nobody needs.

Three questions for your team

• For our next redesign, what decision should the site help the user make, and can we state it in one sentence before anyone opens Figma?
• Where in our product does a claim sit without proof next to it, and what evidence closes that gap?
• Which UI patterns are in our product because users need them, and which are there because other designers have them?