Go Back

AI is making websites age faster: what 54 sites changed in a year

Published date:

October 31, 2026

Author

Nick Hardy

Senior Product Designer

I've been collecting inspiring websites for over a year and saving many of them as PDF snapshots. When I came back to all 54 in October 2026, the surprise wasn't that they got simpler: websites have been moving towards pragmatism for a decade. The surprise was the speed. In one year, more sites aged than I'd normally see in several, and the ones that leaned on AI-generated visuals aged first.

Grid of 54 website previews from Nick Hardy's Inspiration list, saved in 2025 and revisited in 2026
The 54 websites from my Inspiration list, saved in 2025 and revisited in October 2026.

Why I keep an inspiration list

I started the list as a folder of sites I wanted to visit again and again. I photographed them because I knew they would change. Over time it turned into an analytical tool: a way to see how real companies treat their websites over time. I came back to it while working on my own site.

‍

How I compared the 2025 and 2026 versions

I saved most sites as full-page PDF snapshots with a script an early AI model wrote for me. I lost that script, so the newest sites have no PDF yet, and rebuilding it is on my list, because without snapshots you can't really see what changed.

By simplification I mean one thing: removing complex visual elements. If a team added a strong AI-made element, I don't count that as simplification. That's enrichment. Each site landed in one of five groups: simplified, unchanged, improved, gone or renamed, or too new to compare.

Stripe Sessions website: the 2025 snapshot with a purple and orange 3D wave next to the 2026 live site with a larger static ribbon
Stripe Sessions: the 2025 snapshot next to the live site in 2026.

‍

What happened to 54 websites in one year

Of the 54 sites, 14 were added too recently to compare, and one I couldn't check. Of the 39 I could compare, 17 stayed the same, 11 got simpler, 8 improved, and 3 are gone or renamed. Simplification wasn't the majority. The bigger surprise is that many sites that didn't change at all still look a year older. The market around them moved, mostly because of AI.

Bar chart of 39 websites compared between 2025 and 2026: 17 unchanged, 11 simplified, 8 improved, 3 gone or renamed, with site names under each bar
39 websites compared: 17 unchanged, 11 simplified, 8 improved, 3 gone or renamed.

The least surprising part: many SaaS projects moved to pragmatism. When you launch, you want to impress. When you actually have to build and grow the site, you don't have enough hands, and nobody on the team can draw graphics that complex.

‍

Why do websites get simpler?

In my experience, two things trigger it. Either the product team changes, or the company gets enough money and users that marketing moves its effort elsewhere, sometimes into the physical space. Optimisation plays a part, but it isn't the main reason: most of these sites were already well optimised.

Medusa website in 2025 with layered product cards, and in 2026 with an infrastructure dashboard layout
Medusa: an isometric animation in 2025, a product-led layout in 2026.

‍

Which design elements disappear first

3D, illustrations and big full-screen animations go first. The reason is simple. A new product appears and needs a new visual, and there's nobody to make it. One person was paid a lot to create the original, and no one else can repeat it.

Glu website in 2025 with a custom 3D card illustration, and in 2026 with product mockups on a dark background
Glu: custom 3D art in 2025, product mockups in 2026.
Volet website in 2025 with 3D cards and silver coins, and in 2026 with a flat The Wallet headline and no 3D
Volet: 3D compositions in 2025, a flat layout without 3D in 2026.

‍

AI-generated visuals age the fastest

The companies that were first to put AI imagery on their sites made them look cheaper. AI moves so fast that its styles age month by month. Anything generated that appears at three, four or five big companies ages instantly, and filters on top don't help. The teams that added the most generated graphics aged first, because unique AI 3D barely exists, and making raster 3D look unnoticeable is something only a handful of people can do.

AI works well where you can't tell it's there. UI animations built into the site are a good example. You can still spot AI immediately in drawn people and in 3D, but UI elements it makes well. Most of the strong sites in my list don't look AI-made at all. They probably used it in the process, not on the surface.

Five AI-generated visuals from websites: a cat in a hoodie, two women in lime sweaters, a man pointing, and a grid of AI avatars
AI visuals in the wild, 2025 and 2026: recognisable in seconds.
Two Xapo Bank lifestyle images from 2026: a couple looking at a phone and a man in an airplane seat
Xapo Bank in 2026: imagery that is hard to tell from real photography.

‍

Good simplification vs bad simplification

Simplification makes a site better when there's a brand, a consistent style and one strong element to hold everything together. If you have a huge user base, you can simplify however you like and build pages at scale. That's fine too. Today content matters more than ever and deserves more time than before. And content isn't only text: it's also the structure and architecture of the site.

Bad simplification is when there's no direction. Buttons get bigger, elements float, the style is smeared across the site, a random gradient here, something creative there. It looks like a junior made it. Look at premium luxury brands: their style is always very simple and very hard to copy, because many people stand behind it, photographers and designers who spend a week on a single corner radius. When a junior thinks they can do the same, they usually get 10% of it. One payment platform in my list shows this well: its blocks stayed strong because they're easy to repeat, while the graphics started to vary from page to page.

Amie website in 2025 with a 3D calendar dashboard, and in 2026 with a clean layout about AI agents
Amie: simpler in 2026, and stronger for it.

‍

Why some websites never change

Some sites keep one style for years because they're a showcase. They already made one powerful element that's very hard for anyone to repeat, and the real conversations and sales happen offline, not through the website. There are few niches like this: mostly technical products and personal portfolios. Portfolios rarely change for another reason: they're hard to maintain and take time.

Haven website with an illustrated mountain landscape next to Museumnacht Enschede with bold red event type on black
Haven and Museumnacht Enschede: the same style, year after year.

‍

Design patterns that survive simplification

Layouts survive. UI animations will survive too, now that AI makes them fast to produce: anything quick and easy to make survives. Small, elegant animations, for example on cards, survive. Photography survives if you invested in a real photo base, because you can reuse it again and again. Generic stock graphics don't. But there are exceptions: Clay's rebrand uses stock so well that you forget it's stock.

Clay website in 2025 with a serif headline and app interface, and Mesh in 2026 with a dark hero reading Every relationship, remembered
Clay in 2025 and its rebrand as Mesh in 2026.
Mesh network graphic: glowing nodes with stock portraits connected by thin lines on a dark background
Mesh: stock used so well you forget it's stock.

‍

What to take from this

Founders. You need an experienced designer to tell you whether you really need to redo the whole site in a new style right now. Many teams spend too much time on the website instead of the product at the start. A good designer gives direction on layouts that are simple and maintainable, so your own team can build them. If you run a digital business, look at the market broadly and study your competitors. Up close, AI can look like a shortcut. Seen from above, the AI look is mass market, and it doesn't suit everyone.

Marketers. If you commission a unique, ultra-complex 3D solution, expect it to live briefly. It will most likely be a burst for awards rather than a lasting part of the brand.

Designers, especially juniors. Invest in uniqueness and pragmatism. Everyone will build sites with AI, but animating an AI-built site in an interesting way isn't one prompt. It takes weeks of testing and fixing on one element until you find something unique, and that part hasn't changed. Don't rush to redo every site someone generated: they don't need a designer yet, they're testing whether the product works. Designers are needed for support, consistent content, proofreading, ideas and knowledge of a specific business. Adjacent skills like 3D, code or Unity make you much faster. In this era you're limited only by your imagination.

Takeaways by role. Founders: ask an experienced designer before redoing the site, build layouts your team can maintain, study competitors because the AI look is mass market. Marketers: ultra-complex 3D lives briefly. Designers: unique work takes weeks, grow adjacent skills, you are needed for support, content and business knowledge.
What to take from this, by role.

‍

Has this changed how I design?

No. I've been watching this for 11 years, and it has always worked this way. Expensive, complex decisions that take long to make usually lead nowhere. Products that try to save themselves through design never do. Uniqueness is for niches, and simplicity is better than most people think.

What is new is the problem AI creates: startup sites now look so alike that it's hard to tell which product is being advertised. Spend a little time looking at competitors and a little time on how to differentiate. Few people do it yet. Most rely on AI, and in a year, specialists will be growing their skills exactly where nobody else is looking.

Outcomes are my own judgement, made by comparing saved snapshots with the live sites in October 2026. The goal is to show a trend, not to rank anyone.

SHARE

Author

Nick Hardy

Principal Product Designer

More articles and tools like this

I share research, tools and design notes on LinkedIn.
Follow along to see the next one.

Follow on LinkedIn