Welcome, and I hope you’ve had a great week! Before diving into today’s topic, here are 2 links from this past week:

What could you share that people would choose to follow, even when they aren’t shopping for what you sell? A new brand content strategy video I filmed in the studio is out.

Should “thought leadership” be used by charities? If you’re interested in the nonprofit space at all, an article I wrote for the Canadian Marketing Association was just published on their new website:

Now, on to this week’s deep-dive.

When websites start looking the same

Lately I’ve been noticing websites looking really similar.

Of course, others have noticed too, and there’s some content about it (I chose not to read any other coverage before writing this, though).

As a designer, I’m noticing things like the small all-caps text above the main headline, content being placed inside boxes with thin borders and light drop shadows, a modern serif font is paired with a clean sans-serif, with certain words italicized and highlighted in an accent colour.

The spacing is generous. The corners are rounded. It’s clean and polished.

First time I’ve been on the website, yet somehow it feels like I’ve already seen it…

From the computer lab to the AI website builder

Last week in this newsletter, you saw 9-year-old me designing on the computer.

My relationship with websites goes back to learning design and basic web development in school computer labs. I also spent time experimenting on that family computer, trying different layouts, changing colours and breaking things along the way.

Back then, building a website required learning HTML and CSS. I remember using Dreamweaver (before it was even part of Adobe), between its visual editor and the code. The process could be slow, but it also taught me how websites were put together.

Then I moved on to WordPress. Themes became more sophisticated, plugins reduced the need for custom development, and content management became easier. Later, Squarespace made it possible to build a polished site from professionally designed templates using a mostly drag-and-drop interface.

Every generation of tools lowered the barrier. That has been a good thing.

More businesses could get online before they had the budget to hire a developer or designer. More artists could create portfolios. More nonprofits could showcase their cause. More people could publish their work without needing to learn to code or hire an entire agency.

Right now, we’re experiencing another major leap. Instead of choosing a template and customizing it, people can basically just describe the website they want and have AI generate it for them.

What’s everyone using to build now?

Founders and entreprecurious people I know are building using Claude Code and other tools like Lovable. It’s one of the growing number of AI platforms that allows people to build websites and apps using natural language.

You describe what you want, and it can generate the interface, code, database, integrations and other functionality required to make it work. Think “vibe coding” — directing software through conversation and prompts rather than writing all the code yourself.

Over 50 million projects had been created using Lovable up until last month, with approximately 1 million more being created each week. They also reported that 80% of its builders identified as working in non-technical roles.

That’s pretty significant.

People who may never have considered building software or launching a website can now turn an idea into something functional. Founders, designers, educators, marketers and people working in industries far beyond technology are participating.

As someone who remembers having to learn code in a school computer lab, democratization of it is exciting because I think everyone should have access to ways to express ideas, build businesses and make useful things.

But accessibility and originality don’t always move in the same direction.

Why do so many of these websites look similar?

Everyone is copying everyone else, right? It’s actually not that simple in all cases.

Many of these websites are being produced from the same “ingredients”.

To get technical, Lovable’s documentation says that it uses Tailwind for styling, and its guidance encourages the use of shadcn/ui components when possible. These are popular tools for creating clean, consistent and responsive digital interfaces. They make it easier to produce buttons, cards, navigation menus, forms, pricing tables and other common website elements without designing every component from scratch.

There is nothing inherently wrong with that. Design systems and component libraries help websites remain consistent, accessible and easier to maintain. But shared components naturally create shared visual patterns.

Rounded cards, thin outlines, clean typography, muted colours, soft shadows and generous spacing are all reliable choices. They are also the choices that show up repeatedly because they are already widely accepted as signals of a “modern” website.

Then there is the way generative AI works.

When someone asks an AI builder to create “a modern consulting website,” “a premium SaaS homepage,” or “a clean personal brand site,” the system has to interpret what those vague instructions mean. It draws from patterns in its training data and from the conventions built into the tools it uses.

The model isn’t necessarily developing an original visual language for each company. It is predicting a design that is likely to satisfy the request.

Ask enough people for something clean, modern and professional, and many of them will arrive at a similar destination.

I read an article in The New Yorker about another increasingly recognizable AI design aesthetic appearing across websites and presentations: muted tones, prominent serif typography, editorial-looking details and recurring interface components. The article also points to standardized open-source libraries as one reason AI tools develop a recognizable “house style.”

In other words, what I’ve been noticing isn’t just in my head.

Maybe you’ve started noticing it?

It’s showing up IRL, too.

While in New York this week, I even saw this very style show up in a subway ad. This ad for a slide deck builder called Genspark includes an example of a sales deck that uses the exact same visual style and elements I’ve been describing here:

Some of the best advertising and design talent in the world in this city, yet this generic AI design is showing up in big media buys.

So yesterday while in New York’s big Barnes & Noble bookstore, I paid a visit the design section.

In a book on the history of design, a chapter on the arts & crafts movement (way back from 1850-1920) started with this quote:

That got me thinking… will we continue to move further and further away from this? Who’s the designer or artist when we shift to AI tools?

I just appreciate how much effort and thought goes into the craft of design. I resonate with the quote on the back of another (really nice) book in the design section — it’s one that I remember saving back in design school:

The web has had recognizable eras before

This kind of convergence isn’t completely new.

• WordPress themes developed a recognizable look.

• Bootstrap created a generation of websites that used similar navigation bars, buttons, grids and layouts.

• Squarespace templates introduced another visual language that became especially common among photographers, restaurants, consultants and creative businesses.

Design has always moved through trends. We’ve had glossy buttons, skeuomorphic interfaces, flat design, parallax scrolling, minimalist landing pages, brutalism and more gradients than anyone reasonably needed.

AI accelerates the process, though.

Years ago, a designer would notice a trend, interpret it and adapt it. Even when people used templates, they still had to select one, write the copy, choose the images and decide how to assemble the page.

Now, the system can make many of those decisions at once. That removes friction, which makes the process faster. It can also remove some of the moments where a person might stop and ask whether the obvious choice is actually the right choice… an actual design decision.

Studying design homogenization

Researchers have actually started examining this more formally.

A paper called Interrogating Design Homogenization in Web Vibe Coding looked at the risk that AI-generated websites may reproduce dominant design conventions and gradually narrow the diversity of expression online.

The researchers are careful not to argue that AI-assisted design will inevitably make every website identical. Instead, they identify points in the process where homogenization can occur, especially when systems are designed to make generation as frictionless as possible.

Their proposed response is something they call productive friction: deliberately introducing opportunities for creators to question, evaluate and change the default output rather than accepting the first polished result.

Interesting idea, right?

Friction is usually described as a problem in digital experiences. We want fewer steps, fewer decisions and less effort.

But some friction is where thinking happens. It’s where you ask why a particular font is appropriate. It’s where you consider whether the photography actually reflects the people you serve. It’s where you decide that your organization shouldn’t sound like every other organization in the category.

Efficiency can get us to a finished website faster. Productive friction can help us determine whether it is the right website.

What this means for design education

During the school year, I spend part of every week teaching — either within the design faculty at OCAD University or at George Brown College’s School of Design. That gives me another perspective on all of this.

The tools available to students have changed dramatically, even over the relatively short period I’ve been teaching. Students can generate concepts, layouts, images, copy and functional prototypes in a fraction of the time that I could (or anyone else in previous years before these tools).

They don’t all want to use AI, though — actually, many of my students over the last year have been explicit about that. I still think learning how to use this matters. Ignoring the tools altogether would be unrealistic and would leave students unprepared for the industries they’re entering.

But teaching the tool is a very small part of design education. It’s actually not even part of some programs.

Students need to understand hierarchy, typography, accessibility, culture, audience, storytelling and the reasoning behind a creative decision. They need to be able to explain why something should look a certain way rather than simply demonstrating that they can prompt a system to produce it.

A tool can generate a website. That doesn’t mean it has created a brand. And a visually polished output isn’t necessarily a thoughtful one.

The role of education becomes less about memorizing which button to click and more about building judgment.

• How do you evaluate an output?

• What assumptions are built into it?

• Whose perspective is represented?

• What has been simplified or excluded?

• What would make this more relevant, distinctive or human?

Those questions become more important as the tools become easier to use.

Don’t underestimate the audience

I don’t think we should assume that buyers and audiences are stupid.

You may not know that a website was built using AI. You may never use the phrase “design homogenization.” But you can feel when you’ve seen something before.

People notice when 5 companies make nearly identical claims. They recognize the same stock photography, the same visual metaphors and the same language about “unlocking potential,” “driving innovation” or “building the future.”

They may not consciously identify every shared design element, but familiarity affects how memorable a brand feels.

We spend hours moving between websites, platforms, apps and media. We’re constantly comparing experiences, even when we aren’t doing so intentionally.

A website can meet every conventional standard of professionalism and still leave no meaningful impression.

That doesn’t mean everything needs to be experimental or difficult to use. Originality without usability isn’t the goal. But there is a difference between using a convention thoughtfully and simply inheriting every default.

When professional becomes the baseline

The average website today may be more functional, visually polished and accessible than at any previous point in the history of the web.

I think that’s a good thing.

• A small organization no longer needs a large budget to create something credible.

• A founder can test an idea quickly. You could also launch a side hustle.

• A nonprofit can launch a campaign page.

• A consultant can create a professional online presence without spending months learning development.

AI is raising the quality floor. As that floor rises, “professional-looking” non-amateur sites become less of a competitive advantage, though.

Differentiation will come from the decisions that aren’t automatically generated: a specific point of view, an original story, recognizable photography, custom illustration, meaningful cultural context, distinctive language and a brand identity grounded in something deeper than current design trends.

This is where strategy matters. (Says the strategist here).

Before asking AI to build a website, an organization should know what it wants the website to communicate. It should understand who it is speaking to, what those people care about, what makes the organization meaningfully different and how the experience should make someone feel.

Otherwise, the prompt may produce a perfectly nice website. It may simply be a perfectly nice website that could belong to almost anyone.

At the intersection of access and originality

Not saying I want to return to a time when publishing online required specialized skills, expensive software or access to a professional development team.

The democratization of web design is real progress. Tools and tech will allow more people to create, test and share ideas, including people who have historically been excluded from technology and design.

But access to the tools doesn’t remove the need for thoughtfulness.

• As execution becomes easier, intention becomes more valuable.

• As professional design becomes more accessible, distinctiveness becomes harder to achieve.

• As AI learns to reproduce what a modern website is “supposed to” look like, brands will have to work harder to determine what their own website should look and sound like.

Perhaps the average website has never looked better.
But it’s also never been easier to forget.

When so many businesses can create something polished almost instantly, what do you think will make a website or a brand feel genuinely memorable?

Let me know what you think.

I’ll leave you with something that’s the opposite of digital design or websites: A video I filmed while away last month: