The Internet was built on the assumption that there is a human being sitting on the other side of every screen. That no longer exists. In June 2026, Cloudflare’s radar data showed Agentic AI bots generate 57.4 percent of web requests globally, with humans accounting for 42.6 percent. Cloudflare CEO Matthew Prince, who expected the crossover in 2027, said it came faster than anticipated and called the data “a little messy,” but clearly passed the tipping point, as reported Tom’s Hardware. Ordinary crawlers passed through human traffic a decade ago. What’s New Agented Traffic: The system that browses the web on your behalf when you ask an AI assistant a question.
These readers do not read a page the way a human does. They analyze it, resolve the entities on it, and decide whether they trust it enough to cite it or act on it. That work runs on something old and unnatural: linked data. The most common way to publish it on a webpage is JSON-LDAnd twenty years later it has become the connective tissue of the AI search era.
Where does the linked data come from:
Linked data comes from a 2006 Design Note by Tim Berners-Leewho set out four principles for data machines that could connect to sites: name things with URIs, use HTTP URIs so they can be looked up, return useful information in open standards like RDF, and link to other URIs so machines can find more. The goal was to create a web of data, not just documents. It was sustained in academic circles for years by syntaxes such as RDF/XML and Turtle. Two things changed this. In 2011Bing, Google, Yahoo and Yandex launched schema.orgA shared vocabulary for describing things on the Web. Then JSON-LDProposed by Manu Spoorni around 2010, Linked Data was given a syntax that felt familiar to anyone working in JSON. it became a W3C Recommendation in January 2014with version 1.1 Following in 2020.
Why did JSON-LD win in SEO?
Google starts favoring JSON-LD 2015 And soon new types were first documented in it. It sits in its own script block, separate from the visible HTML, so you can add and maintain it without touching your layout, and it works on JavaScript-rendered content. The crawls have found it again and again most widely deployed structured data FormatBeyond Microdata and RDFA – Google’s Own Structured Data Documentation Confirms. The benefits were rich results, SERP features, and clear entity signals feeding Google’s Knowledge Graph.
Changes in agent search:
The same structured data that yielded rich snippets is now read for layer AI systems to understand and act. AI search does not rank the ten blue links. It extracts the entities, checks their consistency, and collects the answers. Schema markup gives it clear context rather than forcing you to infer meaning from prose. Organization and person markup, added with link to identifiers such as wikidataHelp AI decide who or what your content is about, and sort out the sources it cites with confidence. Results vary, and no one outside the platform knows the exact weights, but the direction is consistent across Google, Bing, and the major engines. (None of this is new thinking: the late Bill Slowski was writing about Entities and Semantic Search on Beach SEO AI answers made this imperative long ago.)
Agents take it further. They don’t just understand the content, they act on it: compare products, check the price, complete a task. This only works if the data is structured, current and consistent. If your schema shows one price and your page shows another price, an agent learns to distrust both. This is why agentive commerce protocols like Google’s are emerging Universal Commerce Protocol and OpenAI agent commerce protocolAssume a machine-readable foundation as below.
Here’s what Google said in Milan:
Google’s Search Central Live Event in Milan in June 2026 Made the direction clear. come from the points below Coverage of the event by Search Engine RoundtableTheir writing primarily of session notes, so consider them as reports rather than official documentation.
Google told “Cross-page @ID Linkage,” Which lets products reference shared organizational data on other URLs through a static identifier instead of repeating it everywhere. This is the linked data principle of reusable identifiers applied to commercial schema. Notes are also described Clean, typed data is being extracted as a context layer that directly powers AI observations and AI modes.The clearest indication yet is that structured data feeds AI answers itself. This was accompanied by renewed investment in schema.org, including plans to publish term-popularity statistics and validation rules in open standards. SHACL And Shakes. A useful myth correction: Google’s parsers don’t reward formal HTML validationAnd forcing artificial chunking is not the goal for AI. Organize content for human readability, because structure that matters is semantics.
Why does this come back to the knowledge graph?
Schema markup is no longer something you slap onto pages. This is infrastructure: A data layer machines read your content to understand, trust, and act on it. It’s a reliable way to maintain scale knowledge graph. Page-by-page markup is brittle. It gets out of sync, contradicts itself in the template, and breaks the moment the AI cross-checks one page against another. A knowledge graph puts your entities and their relationships into a coherent model, then publishes them as linked data that search engines and agents already know how to read.
WordLift does the same thing. we build a Knowledge Graph for your brand and mint a static, unique identifier It contains, for each entity, a referable URI that other pages and external datasets can point to. This is exactly the reusable-identifier approach that Google describes in matching, and we associate those entities with connections to official sources such as wikidata.
Mechanics have moved from rich snippets to AI answers for agents, but the Foundation has not. Name your things, describe them in shared terminology, link them together and keep them precise. Linked Data was the right bet in 2006. now it’s better.