For years, we built websites for people.
A potential customer lands on the homepage.
- They click Services.
- They read About Us.
- They look for proof.
- They compare options.
- They fill out a form.
That basic assumption has shaped almost everything about digital marketing for the past 25 years.
But something fundamental has changed.
The visitor to your website may no longer be a person.
Cloudflare CEO Matthew Prince recently revealed that automated traffic crossed a major threshold earlier this year. In May, nearly 60% of requests to websites across Cloudflare's network were coming from bots.
Bots had overtaken humans.
And Prince had expected that crossover to happen much later.
That headline is interesting.
But the number isn't actually the most important part.
The important question is:
What are all those machines doing there?
Not every bot is doing the same thing
When people hear "AI bot," they often assume a company like OpenAI or Google is crawling the web to train its next model.
Some crawlers are doing exactly that.
But that's only one part of what's happening.
Cloudflare is increasingly distinguishing machine traffic by its purpose.
Some machines are discovering and indexing information for search.
Some are collecting information for AI training.
And some are agents that are accessing information in real time to answer a question or complete a task for a person. Cloudflare says 52% of crawler requests it classified by purpose, were for AI training as of June, while mixed-use crawlers accounted for more than 36%.
That third category is where things get especially interesting for businesses.
Because we're moving from a web that looks like this:
Website → Search Engine → Human
to one that increasingly looks like this:
Website → AI → Human
And eventually:
Website → AI Agent → Action
That changes what it means to have a website.
The machine may meet your company before the customer does
Imagine someone asks an AI assistant:
"Who are the best commercial painting companies in Seattle?"
"Which accounting platforms specialize in manufacturing?"
"Who can help my company become visible in ChatGPT?"
"Compare these three providers."
"Which one would you recommend for a company like mine?"
The person asking may never visit ten websites.
The machine can.
It can retrieve information, compare companies, look for evidence, reconcile different sources and construct an answer.
Cloudflare describes essentially this same shift. It says the moment that determines whether someone chooses a company can increasingly happen inside the model's response, before that person ever sees the company's homepage.
That sentence should get the attention of every business owner.
Because it means your website isn't only persuading people anymore.
It is supplying evidence to machines that may influence what people discover.
So what is the machine looking for?
This is where I think businesses are getting AI visibility wrong.
The answer isn't "more content."
A machine needs to understand the business.
That means finding clear, consistent information about things like:
• What you do
• Who you serve
• Where you operate
• What problems you solve
• What you're particularly good at
• Your products and services
• Your founder and subject matter expertise
• Customer experiences
• Comparisons and alternatives
• Frequently asked questions
• Independent reviews
• Press and third-party authority
• Evidence supporting the claims you make
And critically, it needs to understand how those pieces relate to one another.
This is why I've become increasingly interested in disambiguation.
I've seen this firsthand with Monic AI Systems.
A company can be cited and still be misunderstood.
A model can find your content and still associate you with the wrong category.
It can understand individual pages without clearly understanding the entity behind them.
It can know that you exist without understanding when you are relevant to the question being asked.
That's an entirely different problem from traditional website traffic.
Crawlable does not mean understandable
This distinction matters.
A website can be technically accessible to an AI crawler and still be difficult for a machine to interpret.
So I think businesses need to start thinking about AI visibility as a progression:
Crawlable → Understandable → Retrievable → Citable → Recommendable
Those are not the same thing.
- Being crawled gets information into the pipeline.
- Being understood helps machines establish what the information means.
- Being retrievable means the right information can surface for the right question.
- Being cited means your information becomes evidence in an answer.
- Being recommended means the system has enough relevant evidence to include your business when someone is deciding what to do or whom to trust.
That last step is ultimately where the commercial value lives.
Cloudflare itself is moving in this direction
This is what makes Cloudflare's recent announcements particularly interesting to me.
This isn't simply an SEO industry inventing another acronym.
Cloudflare now has an AEO Visibility Dashboard designed to show businesses whether AI assistants are mentioning, citing and recommending them.
It measures citation rate, mention rate, prominence and share of voice across the questions customers ask AI systems.
Cloudflare has also said explicitly that as discovery shifts from search engines toward answer engines, businesses need to understand not only who is crawling their websites, but how their content is being used in AI-generated answers.
And there's another development I think matters enormously.
Cloudflare is separating AI discovery from AI training.
Its new controls are designed to let websites remain discoverable while preventing their content from being used for model training, with Apple, Google and Microsoft honoring or committing to honor the new mechanism.
That distinction tells us something important about where the web is going.
Being available to an AI system does not have to mean giving a model unlimited rights to your content.
Discovery, retrieval, training and agent access are becoming separate decisions.
The website isn't dying. Its job is changing.
We've heard variations of "websites are dead" every few years.
I don't think that's what's happening.
I think the website is becoming more important, but for a different reason.
It remains the place where people experience your brand.
But increasingly, it is also becoming the structured source of truth machines use to understand your business.
- Your expertise.
- Your products.
- Your evidence.
- Your relationships.
- Your authority.
- Your knowledge.
- Your website is becoming both a destination and a data source.
And Cloudflare is already working with OpenAI on ways to make AI search better at discovering and indexing relevant, fresh information from the open web.
That should tell us something about what comes next.
We're building for two audiences now
The human audience still matters.
We need compelling design.
Good storytelling.
Clear calls to action.
Trust.
Emotion.
Experience.
But there's now another audience sitting alongside it.
The machine audience needs something different.
- Clarity.
- Structure.
- Context.
- Relationships.
- Evidence.
- Corroboration.
- Freshness.
And an unambiguous understanding of who you are.
Cloudflare describes the emerging web as needing to be readable, discoverable, callable, and payable for agents.
I would add one more requirement for businesses:
Understandable.
Because the future of digital visibility isn't simply whether a machine can access your website.
It's whether that machine understands enough about your business to know when you belong in the answer.
And now, the machines are showing up in greater numbers than the humans.
That makes this a business problem, not a future problem.