I built a website for an internet that might stop visiting websites.

I recently did something that, depending on whom you ask, is either charming or spectacularly poor timing.

I built a personal website.

In 2026.

At precisely the moment publishers are wondering how much longer people will need to visit websites at all.

Excellent timing, Bhavesh.

Mine doesn't sell anything. There's no newsletter pop-up, no course to enrol in, and no downloadable guide promising seven secrets to becoming a thought leader.

It does, however, have a link that says “Who's this guy?”

Which is a fair question to ask of someone who has put this much thought into a website that might receive fewer visitors than the average neighbourhood tea shop.

When the clicks stop coming

For years, the arrangement was relatively straightforward. You published something useful, a search engine discovered it, and someone searching for a related question followed a link to your page.

Getting that arrangement to work in your favour eventually became an entire industry involving keywords, backlinks, consultants, and invoices that could occasionally make you question your life choices.

But the underlying bargain was understandable. Search engines needed websites to point at, and websites needed the visitors those links delivered.

Now AI is changing the bargain.

In its 2026 report on journalism and technology, the Reuters Institute cited Chartbeat data showing that Google organic search referrals to more than 2,500 news websites fell by 33% globally between November 2024 and November 2025.

That doesn't mean every website is losing a third of its visitors, or that AI caused the entire decline.

A Pew Research Center study found that people clicked a traditional search result in just 8% of observed Google searches that displayed an AI summary, compared with 15% of searches without one.

Those searches weren't identical, so the numbers don't prove AI caused the entire difference. And to be fair, Google says its overall organic click volume has remained relatively stable.

The evidence isn't one-sided, and the web certainly isn't disappearing. But the path from searching for information to visiting its source is becoming less predictable.

After spending two decades encouraging people to make their websites attractive to algorithms, the internet now has algorithms capable of reading those websites on our behalf.

There's a certain irony in that.

For anyone whose business depends on visitors, that raises an uncomfortable possibility.

What happens when people get what they need from a website without ever visiting it?

I'll come back to that. My own reasons for building one start somewhere else entirely.

They start with LinkedIn.

A place beyond the job title

I work in enterprise technology and sales, so LinkedIn is an important part of my professional life. But having most of your public writing attached to a professional profile comes with an odd complication.

Your job title accompanies your thoughts everywhere.

A comment about AI becomes something people interpret through your industry. An observation about management might sound like commentary on your employer. An unconventional opinion occasionally triggers a small, entirely imaginary compliance department somewhere inside your head.

Should I really post this? Does it have anything to do with my work? How might someone interpret it?

And slowly, almost without noticing, you start editing the range of things you allow yourself to explore publicly, along with the words you use.

There's a quiet expectation on LinkedIn that everything you say should somehow fit the person your professional profile says you are.

One day I'm interested in AI and workforce transformation. Another day it's ancient civilisations, human behaviour, or a peculiar problem in medical science that has absolutely no business appearing between two posts about enterprise sales.

My first essay on the website, for instance, was a letter asking AI researchers to make broken bones heal faster.

I've fractured four of mine, so I consider this a perfectly legitimate research priority. Frankly, after four fractures, I feel I ought to be entitled to some sort of honorary qualification in orthopaedics.

Try explaining that to an algorithm expecting a consistent professional niche.

I wanted somewhere I could write without first asking whether a thought was professionally relevant, and explore an idea without pretending to have reached a conclusion.

LinkedIn is where people can discover what I do. I wanted a place where they could discover how I think.

A personal website doesn't make my professional identity disappear, of course. Anyone can still read what I publish. But it gives me control over the presentation, context and archive of my writing.

I get to decide which ideas belong together, revisit an unfinished argument months later, and change my mind without treating it as a personal branding emergency.

What experience adds to an explanation

AI is making explanations extraordinarily cheap.

A capable model can assemble a coherent, well-organised explanation of almost any widely understood subject in seconds. Ask it about bone healing and it can explain inflammation, cell regeneration, vascularisation and mineralisation with considerably more confidence than I could.

But that isn't really what my essay was about.

I'd broken four bones, spent far too long waiting for them to heal, and wondered why, after so much progress in medicine, recovery still required quite so much patience.

Given what AI is beginning to make possible in scientific research, could we find a way to make broken bones heal dramatically faster?

A researcher might have a perfectly good explanation for why that's harder than I imagine. There may be biological constraints I haven't considered, complications I don't understand, and decades of work on precisely this problem. I'd genuinely love to know.

The reason I wrote about fracture healing wasn't that medicine needed another general explanation of bone regeneration. It was that I'd lived through the waiting and started wondering whether new scientific tools could change it. Someone else, with different frustrations, might have noticed an entirely different problem.

An experience-shaped observation can expose a gap that a general-purpose explanation misses. AI can assemble established knowledge remarkably well, but it doesn't share the particular frustrations, constraints or lived circumstances that prompt a person to notice a neglected problem. My four fractures made the speed of bone healing unusually important to me; someone else's life might reveal an entirely different blind spot. None of that makes the observation automatically valuable, and originality is no guarantee of quality. It does, however, give me a more meaningful starting point for writing than manufacturing articles around whatever an algorithm might reward.

Perhaps, in a world overflowing with perfectly adequate explanations, the things worth finding will also include observations that couldn't have emerged in quite the same way from anyone else.

And it leads naturally to my second reason for building the website.

This one is considerably more ridiculous.

Being found by a machine

I want AI to discover my writing.

Yes. Really.

I'm aware of the irony of spending several paragraphs explaining how AI might reduce visits to websites before confessing that I'm hoping one will visit mine.

I'm not a famous researcher, a bestselling author or a public intellectual. I don't have an institution whose reputation automatically makes my opinions worth quoting.

I'm just someone with a domain name, an unreasonable number of browser tabs, and an increasingly ambitious relationship with search-engine crawlers.

Yet I find the possibility fascinating.

Imagine someone, somewhere, consulting an AI about a subject I've spent time thinking about.

Perhaps they're wondering why physical healing seems so slow compared with the astonishing pace of progress elsewhere in science. Or perhaps they're exploring some obscure corner of science I haven't even thought to write about yet.

And the AI finds an essay I wrote.

It might link to the essay, cite an observation, or introduce someone to an idea without their having the slightest clue who I am.

Modern retrieval systems can sometimes connect a user's enquiry with documents through their meaning, rather than relying entirely on matching keywords. That creates the possibility of relevant ideas surfacing even when someone doesn't know the author's name or the exact words they used.

There's something delightful about that.

A machine's attention wouldn't make my ideas correct or important. I simply like the idea that someone without institutional standing could contribute something discoverable to a much larger conversation.

Before this starts sounding like a plan to achieve immortality through a sitemap, I should clarify something.

Being discovered by a crawler, indexed by a search engine, included in AI training data, and cited in an AI-generated answer are four different things.

OpenAI, for example, documents separate crawlers for search discovery and content that may be used in training. Making a website technically accessible guarantees nothing about whether an AI system will ever retrieve or reference it.

I could do everything right and spend the next twenty years being comprehensively ignored by every artificial intelligence on the planet.

A humbling prospect.

Which brings us back to the uncomfortable economic problem from earlier.

AI companies are developing increasingly sophisticated ways to consume information from the open web. Meanwhile, the people and organisations producing that information may receive fewer visits in return.

Cloudflare's research has documented enormous differences between how often some AI crawlers request pages and how often their services send visitors back. The precise ratios are difficult to measure, particularly when referrals from native apps aren't fully visible.

But the underlying economic question is serious.

If AI becomes very good at answering questions using other people's work, what keeps those people producing the work?

The economics of publishing depend on more than information being available. Reporting, research, editing and original expertise cost money, while many publishers rely on readers reaching their own sites, subscribing or paying for access. If assistants increasingly deliver answers without sending visitors back, the people funding the source material may lose revenue even as their work remains useful. That tension is not resolved by making content easier for AI to retrieve.

An independent journalist cannot pay the bills with the satisfaction of being included in a chatbot's answer. A specialist publisher needs readers, subscribers or some other sustainable source of income.

My website solves none of that.

I have the luxury of writing without needing each essay to generate revenue. A publisher employing dozens of journalists doesn't, and that difference matters.

A company might evaluate a website through visitor numbers, conversions, acquisition costs and revenue. I spend enough of my professional life thinking about those things to have no interest in subjecting every strange idea that enters my head to a quarterly performance review.

My metric is considerably less sophisticated.

Did I finally give that thought somewhere to live?

I don't know which direction the internet will take. AI assistants might become generous gateways to original sources, or destinations that rarely send anyone elsewhere. Most likely, different kinds of websites will experience very different outcomes.

What I do know is that I don't want the value of writing something to depend entirely on how many people clicked on it.

If twenty people read an essay and one of them thinks about something differently, that's worthwhile. A conversation sparked years later would be better still.

And if one day an AI happens to introduce one of my obscure observations to someone who would otherwise never have encountered it?

Well, that would be rather satisfying.

For now, I have a small corner of the internet that belongs to me, outside the expectations of a professional feed and with no business case required.

Just a growing collection of things I know, things I think I know, and things I would very much like to understand.

There's a link on the homepage that asks, “Who's this guy?”

Maybe one day an AI will have an answer.