This week brought a sharp mix of practical CSS upgrades, hard lessons in AI-assisted design workflows, and a standout case study in concept-driven web development. Our team has been pulling apart each development to identify what matters for client projects right now. Here’s what we’re acting on.

Key Takeaways

• CSS gap decorations are now fully supported in Chrome 149 and Edge, giving designers native control over grid and flexbox gap styling without hacks.
• AI co-designers need structured context files — not just colour values — to produce usable output on client projects.
• Chat-based interfaces are increasingly seen as the wrong paradigm for AI agent interactions in product design.
• A design subscription studio built its entire site around a single conceptual metaphor, combining Nuxt, Three.js, and disciplined creative direction.
• AI fatigue is real, and information architecture principles are being repurposed to structure AI-driven design workflows.

CSS Gap Decorations: Native Styling Without the Workarounds

We’ve been waiting for this one. As detailed in CSS-Tricks’ breakdown of the new gap decorations spec, Chrome 149 and Edge now let you style the gaps between grid and flexbox items directly in CSS. No more pseudo-element hacks. No more border tricks that break at certain breakpoints.

For our client builds, this means cleaner code, fewer layout bugs, and faster turnaround on card grids, pricing tables, and dashboard layouts. We’re already updating our internal component library to take advantage of properties like row-rule and column-rule within gap contexts. If your site uses CSS Grid heavily — and it should — this is a meaningful improvement to maintainability.

Why Your AI Design Tool Needs More Than Hex Values

Dropping a brand’s colour palette into Claude or another AI tool and expecting polished design output is a common mistake. A thorough piece on what AI co-designers can’t infer from hex values alone makes the case for creating structured context files at the start of every project.

Our team has adopted a similar approach. Before we involve any AI tooling in a design workflow, we prepare:

• Brand voice and tone documentation
• Target audience profiles with behavioural notes
• Spacing and typography scales with rationale
• Accessibility requirements and WCAG targets

The output quality difference is stark. Context files turn a mediocre AI assistant into a genuinely useful one.

Chat Is the Wrong Interface for AI — So What’s Right?

A growing number of UX practitioners are pushing back on the default chatbot pattern. The argument laid out in UX Collective’s analysis of why chat is the wrong interface for AI is straightforward: chat over-simplifies complex agent capabilities and forces users into a linear, text-heavy interaction model.

We agree. For client-facing AI features we build — booking assistants, product finders, support tools — we’re increasingly designing structured UI patterns with guided inputs, contextual suggestions, and visual feedback rather than open-ended chat boxes. Users complete tasks faster. Satisfaction scores go up.

Concept-Driven web design: The PX PUSH Case Study

PX PUSH, a design subscription studio, built its website around a single metaphor: an old, always-on machine. As explored in Codrops’ detailed build breakdown, the site runs on Nuxt with Three.js handling the visual layer. The result is a site where every interaction reinforces the brand concept.

This is the standard we hold ourselves to. Technical capability means nothing without a governing creative idea. The best client websites we deliver start with a clear concept, then select the stack to serve it — never the other way round.

AI Fatigue and the Return of Information Architecture

Designers are tired. The weekly curation from UX Collective’s roundup on IA for AI and battling AI fatigue highlights a useful counter-trend: applying classic information architecture thinking to organise AI-driven experiences. Structure beats novelty. We’re using IA frameworks to map AI touchpoints across client sites so that each interaction has a clear purpose rather than existing because the technology allows it.

The throughline across all five developments is discipline. New CSS features demand clean implementation. AI tools demand proper onboarding. Creative sites demand a governing idea. The agencies and designers who treat these as craft problems — not hype cycles — will build better products for their clients.

Frequently Asked Questions

What are CSS gap decorations and how do they improve web design?

CSS gap decorations let you style the spaces between grid and flexbox items natively, without pseudo-elements or border workarounds. They shipped in Chrome 149 and Edge, making layout code cleaner and easier to maintain.

How do web designers get better results from AI design tools?

By creating structured context files that include brand voice, typography rationale, audience profiles, and accessibility targets — not just colour palettes. This additional context dramatically improves the relevance and quality of AI-generated design output.

Why is a chatbot interface not always the best choice for AI features on a website?

Chat forces users into a linear, text-heavy interaction that often oversimplifies what an AI agent can do. Structured UI patterns with guided inputs and visual feedback typically help users complete tasks faster and with greater satisfaction.

What is concept-driven web design and why does it matter?

Concept-driven web design starts with a single governing creative idea that shapes every visual and interactive decision on the site. It ensures the technology stack serves the brand story rather than the other way round, resulting in more memorable and cohesive user experiences.