Intermediary Layer Shift: When AI Agents Become the New Internet Gatekeepers

The Event

As AI capabilities advance, traditional search engines are being replaced by AI chatbots and generative engines. Content publishers face a new dilemma: the SEO optimization strategies perfected over the past 20 years are no longer sufficient. Brian Alvey, CTO of WordPress VIP, points out that brands must now serve both human readers and AI crawlers simultaneously to survive in this new era. A new term has emerged: GEO (Generative Engine Optimization), though the industry remains unclear on its definition and implementation methods.

Core Observation

The internet architecture is undergoing a transition through three eras of intermediary layers:

First Generation (1990-2010): Humans actively search → Google intermediates ranking → Humans click and read. Google controls "the probability a webpage gets seen." Publishers optimize for Google's algorithm (SEO)—essentially adjusting content format for search engines.

Second Generation (2010-2023): Humans passively engage on social media → Meta/TikTok algorithms decide what to feed them → Humans consume pushed content. Facebook controls "who sees whose content." News media and content creators begin optimizing for social algorithms (click-through rates, shares, dwell time).

Third Generation (2024-present): AI chatbots autonomously browse the web → AI excerpts and generates answers → Humans read AI's summary instead of clicking through to the original page. ChatGPT, Claude, and Gemini become the new gatekeepers. Publishers lose their last direct access to humans—your article may be crawled, excerpted, and condensed by AI; humans may never see your original content.

Why This Is an Intermediary Layer Shift

The power to define boundaries is shifting.

• In the Google era, search terms determined the conditions for matching your content. Keyword research dictated traffic.
• In the social media era, algorithmic weights determined your content's exposure frequency. Content format (short videos, high-engagement posts) determined reach.
• In the AI era, AI decides whether your content is worth crawling, excerpting, and recommending to humans. SEO becomes "how to make AI think your content is credible," not "how to get Google's crawler to find you."

Publishers in the Google era could say "if the content is high-quality and matches search intent, it will be seen." The social media era changed this to "quality content + algorithm-friendly formatting gets seen." The AI era demands "content must serve both human understanding and machine understanding"—because AI will analyze your article's structure, fact sources, and logical clarity to decide whether to use it in generating answers.

New Concept: GEO vs SEO

SEO's assumption: Humans search → Look at rankings → Click webpage → Read. Publishers optimize for "ranking."

GEO's assumption: AI crawls content → Evaluates trustworthiness and relevance → Excerpts or cites → Generates for humans. Publishers optimize for "probability of being selected for excerpts" and "context when mentioned by AI."

GEO is harder because you cannot see the algorithm or reverse-engineer clear optimization rules. Google at least published a *Search Engine Optimization Starter Guide*. But how does ChatGPT's crawler evaluate credibility? How does Claude decide which sources merit citation? Nobody knows.

Dangers of Intermediary Layer Shift

Direct traffic becomes a slow trickle. Publishers are accustomed to passive traffic influx (Google sends searchers, Meta sends scrollers). In the AI era, publishers may never know how many humans read their content through AI summaries, nor can they track conversions (purchases, subscriptions, ad clicks).

Brands and publishers lose status. In the Google era, brands could at least use SEO to attract humans to their sites. In the AI era, humans don't need to visit your site at all—they get condensed versions directly in chat. This threatens content creators, e-commerce, and news media.

Intermediaries control how content is presented. Google's search results page (SERP) still displays your brand name and link. AI responses might only say "based on web information," leaving users unaware of the source. Publishers become "ghost suppliers"—your content has value, but you get no credit.

Practical Difficulties in Shifting to GEO

1. Rules lack transparency: Google publishes updates and algorithm changes. AI model training is a black box. 2. Costs are asymmetrical: SEO can largely be achieved through content quality and structured data (schema markup). GEO demands every claim be sourced, logic be clear, and content be unambiguous for AI parsing—a heavier burden for news, academia, and analysis. 3. Serving multiple formats creates contradictions: The format best for human reading (narrative, sidebars, visual design) conflicts with the format best for AI parsing (structured markup, redundant annotations, explicit citations).

Possible Futures

If AI agents truly become the dominant entry point, the internet ecosystem could evolve into:

• Content publishers forced to optimize for two competing logics simultaneously, raising costs.
• Small publishers and independent creators marginalized because they lack resources for both SEO and GEO.
• New intermediary businesses emerging, such as "AI-friendliness verification tools" or "GEO consulting services," helping publishers understand how to attract AI favor.
• Further internet centralization, where large publishers and brands have resources to adapt to new rules while small sites become peripheral information sources.

Counterargument

GEO anxiety might be overblown. AI models need high-quality training data; if all publishers optimize for AI at the expense of human readability, they'd poison AI's future training sets. AI companies (OpenAI, Anthropic, Google) themselves have incentives to maintain a healthy internet ecosystem ensuring continuous access to quality content.

Moreover, users will eventually discover that if AI summaries decline in quality, they'll return to search engines for original sources. The market might reach new equilibrium: AI agents serve as quick entry points, but humans will occasionally deep-dive into original sources.

---

Conclusion: Intermediary layer shifts are an eternal law of the internet. Every 10-15 years, new technology changes "who decides what gets seen," forcing publishers to relearn. WordPress VIP's advice—serving both humans and AI—isn't sinister; it's reality. The question is who gets left behind during the intermediary transition.