James LePage on SEO for AI Agents: Insights from the WordPress AI Team
James LePage, Director of Engineering AI at Automattic and co-lead of the WordPress AI Team, shared his perspective on how publishers should approach SEO in the age of AI. As the founder and co-lead of the WordPress Core AI Team, LePage coordinates AI-related projects within WordPress, including how AI agents interact with the platform. He outlined emerging trends for the web, AI agents, and their SEO implications.
AI Agents and Existing Infrastructure
LePage noted that AI agents rely on the same web infrastructure as traditional search engines. Their data primarily comes from classic search indexes. He explained:
“Agents will use the same infrastructure the web already has. Search discovers relevant entities. Domain authority and trust signals evaluate sources. Links traverse between entities. Content explains what each entity offers.
I find it interesting how much money flows into AIO and GEO startups when agents still retrieve information from existing search indexes. ChatGPT uses Bing. Anthropic uses Brave. Google uses Google. The mechanics of the web don’t change—only who is traversing it does.”
AI SEO = Longtail Optimization
LePage emphasized that optimizing for AI agents relies on structured data, semantic density, and interlinking between pages. He pointed out that much of what AI optimization companies do is essentially longtail keyword optimization:
“AI intermediaries doing synthesis need structured, accessible content—clear schemas, semantic density, good interlinking. That’s the challenge publishers face today. There’s a bit of FUD in this industry. Billions of dollars flow into AIO and GEO when AI optimization really comes down to long-tail keyword search optimization.”
What Optimized Content Looks Like for AI Agents
According to LePage, content should be intentionally structured for AI consumption, using semantic markup, structured markdown, and clear organization. He describes it as a difference between a “pile of documents” and a “well-organized briefing”:
“Content should prioritize what matters most. Rankings should signal which information is authoritative versus supplementary. Representations should progressively disclose detail, giving agents the summary first with clear paths to depth. This is still static content, shaped with agent traversal in mind. One is far more useful to someone trying to quickly understand what you offer.”
LePage also predicted that in the future, agentic AI may not even need traditional websites—just the content itself. Nevertheless, he recommends keeping content structured at both the page and site level, with interlinking that clarifies relationships between documents and communicates their purpose.
“Think of this as a progression. Today, it’s essentially Perplexity-style web search: gather content, generate synthesis, present to the user. Soon, users will delegate tasks to agents with explicit specifications. Further out, agents may operate autonomously, following guidelines and becoming economic actors in their own right. Humans remain in the loop, but the loop expands. Structured content—clear markdown and semantic markup—helps agents work efficiently even in static formats.”
LePage’s article, Agents & The New Internet (3/5), offers actionable guidance for preparing content for an increasingly agent-driven web.