James LePage, Director Engineering AI at Automattic, and the co-lead of the WordPress AI Workforce, shared his insights into issues publishers must be fascinated by when it comes to search engine marketing. He’s the founder and co-lead of the WordPress Core AI Workforce, which is tasked with coordinating AI-related tasks inside WordPress, together with how AI brokers will work together throughout the WordPress ecosystem. He shared insights into what’s coming to the net within the context of AI brokers and a number of the implications for search engine marketing.
AI Brokers And Infrastructure
The primary remark that he made was that AI brokers will use the identical net infrastructure as serps. The primary level he makes is that the info that the brokers are utilizing comes from the common basic search indexes.
He writes, considerably provocatively:
“Brokers will use the identical infrastructure the net already has.
- Search to find related entities.
- “Area authority” and belief indicators to guage sources.
- Hyperlinks to traverse between entities.
- Content material to grasp what every entity presents.
I discover it attention-grabbing how a lot cash is flowing into AIO and GEO startups when the underlying method brokers retrieve data is by utilizing present search indexes. ChatGPT makes use of Bing. Anthropic makes use of Courageous. Google makes use of Google. The mechanics of the net don’t change. What adjustments is who’s doing the traversing.”
AI search engine marketing = Longtail Optimization
LePage additionally stated that schema structured information, semantic density, and interlinking between pages is crucial for optimizing for AI brokers. Notable is that he stated that AI optimization that AIO and GEO corporations are doing is simply primary longtail question optimization.
He defined:
“AI intermediaries doing synthesis want structured, accessible content material. Clear schemas, semantic density, good interlinking. That is the problem most publishers are grappling with now. The truth is there’s a little bit of FUD on this business. Billions of {dollars} flowing into AIO and GEO when a lot of what AI optimization actually is is just long-tail key phrase search optimization.”
What Optimized Content material Seems Like For AI Brokers
LePage, who’s concerned in AI throughout the WordPress ecosystem, stated that content material must be organized in an “intentional” method for agent consumption, by which he means structured markdown, semantic markup, and content material that’s simple to grasp.
Somewhat additional he explains what he believes content material ought to seem like for AI agent consumption:
“Shows of content material that prioritize what issues most. Rankings that sign which data is authoritative versus supplementary. Representations that progressively disclose element, giving brokers the abstract first with clear paths to depth. All of this nonetheless static, not conversational, not dynamic, however formed with agent traversal in thoughts.
Consider it because the distinction between a pile of paperwork and a well-organized briefing. Each comprise the identical data. One is way extra helpful to somebody making an attempt to shortly perceive what you supply.”
Somewhat later within the article he presents a seemingly contradictory prediction of the function of content material in an agentic AI future, reversing at present’s system of a effectively organized briefing over a pile of paperwork, saying that agentic AI won’t want a web site, simply the content material, a pile of paperwork.
Nonetheless, he recommends that content material have construction in order that the data is effectively organized on the web page degree with clear hierarchical construction and on the web site degree as effectively the place interlinking makes the relationships between paperwork clearer. He emphasizes that the content material should talk what it’s for.
He then provides that sooner or later web sites could have AI brokers that talk with exterior AI brokers, which will get into the paradigm he talked about of content material being cut up off from the web site in order that the info will be displayed in ways in which make sense for a person, utterly separated from at present’s idea of visiting a web site.
He writes:
“Consider this as a development. What exists now could be basically Perplexity-style net search with extra steps: collect content material, generate synthesis, current to person. The person nonetheless makes choices and takes actions. Close to-term, customers delegate particular duties with specific specs, and brokers can take actions like purchases or bookings inside bounded authority. Additional out, brokers function extra autonomously primarily based on standing pointers, changing into one thing nearer to financial actors in their very own proper.
The development is towards extra autonomy, however that doesn’t imply people disappear from the loop. It means the loop will get wider. As a substitute of approving each motion, customers set pointers and evaluate outcomes.
…Earlier than full web site delegates exist, there’s a center floor that issues proper now.
The content material an agent has entry to will be introduced in a method that is smart for the way brokers work at present. Presently, meaning structured markdown, clear semantic markup, content material that’s simple to parse and perceive. However even inside static content material, there’s room to be intentional about how data is organized for agent consumption.”
His article, titled Brokers & The New Web (3/5), supplies helpful concepts of the best way to put together for the agentic AI future.
Featured Picture by Shutterstock/Blessed Inventory
