AI Discovery: What Do Communicators Actually Need to Know?

I recently wrote that communicators don't need technical expertise, but we increasingly need technical capacity.

For communicators who do not consider themselves technical people, I want to unpack this:

If you're responsible for how an organization creates and distributes content, you should probably understand (at least conceptually) how machines find and interpret webpages, how they connect people and organizations with areas of expertise, what makes information easy to retrieve and cite, and how AI discovery differs from traditional search.

As you plan content for yourself or your organization, I would ask these four questions at the beginning of your process—not when you’re ready to publish.

1. Can machines find your content?

Publishing something on the internet does not necessarily make it discoverable.

Search engines and AI systems use crawlers to find information across the web. Your content needs to be accessible to them, linked in ways they can follow, and structured so they can make sense of what they find.

This is where terms like crawlability, indexing, sitemaps and robots.txt enter the communications conversation. You don't necessarily need to know how to configure them, but you should know enough to ask whether your content can actually be found.

• Crawlability: Whether search engines and other machines can access and move through the pages on your website to discover your content.
• Indexing: Whether a search engine has processed your webpage and added it to its searchable collection of information.
• Sitemaps: Files that give machines a structured list of the pages on your website you want them to know about.
• robots.txt: A file that gives web crawlers instructions about which parts of your website they are or aren't allowed to access.

2. If a machine lands here, how easily can it understand what it's looking at?

Humans can look at a webpage and visually recognize a headline, author, biography, body content, and publication date without much trouble.

Machines benefit from clearer signals, starting with sensible page structure and descriptive headings (things you’re accustomed to if you’ve ever focused on SEO).

A layer down from that are metadata and schema markup, which give machines additional structured information about what they're looking at and how those things relate to one another.

• Metadata: Information about a webpage that helps machines understand and describe its content, such as its title, author, description, and publication date.
• Schema markup: Structured data added to a webpage that explicitly tells machines what the content represents—such as an article, person, organization, or event—and how those things relate to one another.

There are tons of free tools like schema.org’s Validator to audit webpages for these components and produce the necessary HTML to add to the head template of your page.

3. Can machines connect you to what you know?

I write regularly about AI and communications. A few months ago, I was interviewed by Help a Reporter Out for its What is PR in 2026? article, which featured my earned-to-owned framework:

“Serko’s owned ecosystem is circular and extensive, all designed to optimize for AI. The system takes something like a media hit and creates owned content structured for AI discoverability with schema and citations.”

The article goes on to connect my name directly with the framework and the broader idea of building “an ecosystem of credibility.”

Once that earned media hit existed, I treated it the same way I’ve been arguing organizations should treat earned media: I turned it into an owned asset. I published a summary on my own domain, linked back to the original article, connected it to my related thought leadership, and added structured data to help machines understand what the page is, who it’s about, and how those pieces relate.

Now there are multiple signals pointing in the same direction: I write about AI and communications; an independent publication interviewed me about it; that publication associated my name with a specific framework; and my own domain reinforces and connects that third-party validation to the rest of my work.

That’s what I mean by a network of relevance. It isn't one webpage with all your best content or one stellar media hit crowning me a thought leader; it's a collection of connected, corroborating signals that help machines understand who someone is, what they know, and why there’s reason to associate the two.

4. Can machines extract and cite the useful part?

AI discovery isn't just about whether a machine understands the content on a page. It’s much more important to consider if the machine can:

• identify the most relevant information based on a human’s query or prompt;
• extract that information accurately; and
• attribute the helpful response to you.

That changes how I think about content—both what I write and how I write it.

I need to pose questions and provide clear answers; I need descriptive headings that strongly indicate what the reader will find below; I need attributable expertise and circular linking to supporting sources. These aren't just editorial choices; they make information easier for machines to retrieve and surface to human audiences.

So, what does “technical capacity” actually mean?

Technical capacity means understanding enough about the infrastructure carrying your work to make better communications decisions—and to know which questions to ask to reach your very human audience.