How AI search engines read your press releases

When I started my public relations career, over 25 years ago, press releases were written with one target in mind: people. Then, as online visibility became more important, the writing had to adapt so that Google could crawl and index the news stories that were distributed and the content that was published.

But the audience has quietly changed again. Now, news releases are also being read, indexed and summarised by something else entirely; Artificial Intelligence. 

Google’s AI Overviews, ChatGPT’s browsing tools, Microsoft’s Copilot and countless industry-specific models are all scanning, summarising and learning from what’s published online. And while they don’t replace journalists or other target audiences, they increasingly shape what people see first when they search.

In reality, if your press release isn’t written clearly, it risks being ignored by both humans and machines.

Why AI visibility now matters

Here's an example - when someone asks an AI tool “who are the main industrial agents in Hampshire?” or “what’s the retail vacancy rate in Farnham?”, the system doesn’t guess. It pulls from structured, trusted information across the web.

If your stories are written in a consistent, factual way and appear on credible sites, those mentions help the engine connect your name, your company, your sector and your location. That’s what GEO (generative-engine optimisation) is really about. Making sure AI can understand who you are and what you do.

What AI looks for

AI tools don’t see your press release as a layout with quotes and logos. They see text, structure and context. The following things help them “read” your content correctly:

• Clear entities: People, places, companies and property names written consistently.

e.g. Curchod & Co rather than ‘Curchods’ or ‘the company’.

• Defined facts: Dates, locations and sizes expressed precisely.

e.g. 13,954 sq ft industrial property at West Portway Business Park, Andover, Hampshire.

• Contextual cues: Explain what something is, not just what happened.

e.g. “multi-let office property” tells the AI what kind of asset it’s reading about.

• Readable quotes: Keep them short and specific. Avoid empty enthusiasm and say something meaningful that reinforces the subject.
• Metadata and links: Titles, captions, alt text and site structure all help confirm what the story relates to.

When those signals are in place, AI doesn’t just see words, it understands relationships between entities. And that’s what gets surfaced in summaries, overviews and recommendations.

The overlap with good journalism

The irony is that this isn’t really new. The same clarity that helps AI also helps journalists. Both prefer straightforward information, plain structure and a clear “who, what, where, why” order.

The difference is that AI never forgets. Once your release is online, those connections stay in the web’s data layer, quietly influencing how future searches interpret your brand and market position.

What this means for agents, developers and their publicity & marketing teams

If you already use Realiser, most of this is built in. Every release we create is designed to be clear, consistent and factually rich, which naturally supports GEO and semantic optimisation.

But the next step is awareness. Treat each release not just as a story, but as a piece of structured data about your business. Make sure it’s posted in a visible place, with correct names, captions and internal links. That’s how your content gets picked up, credited and re-surfaced when people – or AI – go looking.

Final thought

Press releases still start with people. But they now travel through a world read mostly by machines.

The best stories do both: they sound natural to humans, and they make sense to algorithms.

If you get that balance right, your next release could show up not just in the paper, but in the next AI search result that really matters.