Jarrett Sidaway
CEO & Co-Founder, FetchRight
For more than two decades, digital strategy in publishing has been built around a simple reality: the page was the atomic unit of value. You created pages, optimized them, distributed them, and measured their performance. Success was defined by traffic, engagement, and, increasingly, conversions tied back to those pageviews. The business model, the technology stack, and the organizational structure were all aligned around that unit.
AI upends that logic.
Users are no longer starting with a homepage or even a list of search results. They are starting with a question. And increasingly, they are receiving a synthesized answer rather than a menu of options. The publisher's presence, if it appears at all, shows up downstream: as a citation, a supporting link, a shard of context, or sometimes not at all.
It is understandable that this feels like disintermediation. But viewed differently, it is something else: a re-architecture of discovery. And in that re-architecture, publishers have a choice. They can allow their content to be atomized and reassembled without their input, or they can define how their expertise is represented in this new answer layer.
The publishers who treat this as an architectural problem, rather than a distribution problem, will define what comes next.
In the traditional web era, a page served two roles at once. It was the complete experience for the reader, and it was the core object that search engines discovered, ranked, and recommended. Metadata, headings, URLs, and internal links were signals designed to make that page visible and attractive to search.
AI systems do not consume pages in the same way. When an AI assistant "reads" a page, it does not absorb the layout, design, or even the full narrative arc. Instead, it extracts fragments: key sentences, factual statements, definitions, names, dates, and claims. It slices the page into units that are meaningful to a model, not to a human sitting in a browser.
From the model's perspective, what matters is not the page as a container, but the information embedded within it. That information is then embedded into vector spaces, indexed semantically, and later recombined with other fragments to construct an answer to a user's question.
As a result, publishers now operate in a world where the question is the new starting point, and the answer is the new surface. The page becomes one step removed from the user. It still matters — but as a source object, not the primary interface.
This does not make the page obsolete. It makes it insufficient as the sole unit of strategy.
If the page is no longer the core unit of discovery, what replaces it?
The emerging answer is an architectural layer that sits between publisher content and AI systems: a layer where knowledge is expressed in a form that is legible to machines while remaining grounded in editorial judgment.
Historically, publishers built their own architecture for discovery: sitemaps, taxonomy, internal linking schemes, and landing pages that organized content into coherent journeys. That work was aimed at both humans and search engines. The new layer is more specific. It is designed primarily for machine interpretation.
This layer includes structured summaries, topic-level representations, canonical answers to recurring questions, and explicit signals about authority, freshness, and use rights. It treats an article not just as a page to be read, but as a source of reusable, well-defined insights.
In practical terms, this means asking new questions:
When publishers begin to represent their content in this way, they are effectively designing the answer layer, not leaving it to be inferred.
From a purely traffic-centric perspective, one might ask whether any of this matters if AI systems send fewer clicks back to publisher sites. But that framing ignores two realities.
First, brand influence does not vanish in an answer-centric world. It changes form. If AI relies on a publisher's analysis, reporting, or interpretation to construct an answer, there is strategic value in ensuring that the publisher's name, perspective, and authority are present in that answer. Being invisible while powering the system is not a sustainable position.
Second, the economic relationship between publishers and AI platforms is still being defined. If publishers can demonstrate that their content is not simply one of many sources, but a structured, high-integrity backbone for key domains of knowledge, they gain leverage in negotiations over licensing, access, and revenue.
In other words, this is as much about brand and power as it is about traffic. Architecting the answer layer is how a publisher ensures that its editorial voice and authority survive translation into AI-generated experiences. It is not enough to create great journalism. The work must also be structured so that machines recognize and respect its value.
Many publishers are understandably starting from a defensive posture. The legal landscape is active. Scraping feels uncontrolled. Content appears in AI products without consent or compensation. In that context, blocking crawlers or tightening access can feel like the only rational option.
But defense alone is not a strategy for the next decade.
Blocking may be necessary in the short term to prevent uncontrolled exploitation, but it does not answer the question of how audiences will encounter the publisher's expertise in an AI-driven world. Nor does it address the risk that systems will simply lean more heavily on less trustworthy sources.
A more durable strategy begins with proactive design: deciding what kinds of AI use cases a publisher wants to support, under what conditions, and with what representations of their content. That could include structured feeds for retrieval-augmented generation, verified panels for high-stakes topics, or curated knowledge sets for specific verticals.
The key is that publishers are setting the terms of engagement. They are not passively responding to whatever a crawler decides to do. They are expressing, in machine-readable and enforceable ways, how their expertise should power AI answers.
Treating AI discovery as an architectural challenge has implications beyond technology. It is not a matter that can be delegated entirely to product or data teams.
Editorial leadership must define what "authoritative" means within their organization and which kinds of content are suitable to anchor answers in their domain. Legal and policy teams must define acceptable use, risk tolerance, and licensing boundaries. Commercial teams must determine how structured participation can drive revenue, partnerships, or strategic visibility. Technical teams must translate these decisions into concrete formats, protocols, and enforcement mechanisms.
If each of these functions approaches AI in isolation — editorial worrying about misrepresentation, legal about infringement, product about integration — the organization will remain reactive. A unified view starts from a simple premise: the publisher's content is not just a collection of pages, but the underlying knowledge layer for a set of topics that matter to its audience. AI is the new interface to that knowledge. The business must decide how that interface works.
That decision is a leadership responsibility.
It is tempting to wait for standards, regulation, or market norms to settle before making big moves. The ecosystem is fluid. Business models are still emerging. Regulatory conversations are active. None of that is trivial.
But waiting carries its own risks. Mental models about how AI "should" work are being formed now — in product teams, in regulatory bodies, and in public expectations. If publishers are not actively shaping those models, they are implicitly accepting whatever defaults emerge from elsewhere.
Once AI platforms normalize training on unstructured, unlicensed content, it becomes much harder to renegotiate terms later. Once certain domains of knowledge are associated with other voices, reclaiming primacy becomes more expensive. Inertia favours those who moved early to define the interface.
The real risk is not missing a specific deal or product integration. The real risk is allowing others to define how your expertise appears, or does not appear, in the answer layer that is increasingly mediating access to information.
AI is not simply another platform to syndicate into, nor a passing trend in user behavior. It is a structural shift in how people ask questions and receive information. That shift creates risk for publishers who do nothing — but it creates opportunity for those who recognize that they are not passive content providers but architects of the knowledge layer these systems require.
Architecting the answer layer means treating your content as more than pages and your AI posture as more than a legal position. It means deciding how your expertise is expressed to machines, how your authority is signaled and preserved, and how value flows back when your work is used to power AI experiences.
Infrastructure like FetchRight exists to make this practical, but the strategic decision sits with leadership. The publishers who embrace this responsibility will not simply endure the transition from pages to answers. They will define it, and in doing so, shape how their audiences continue to discover, trust, and depend on their journalism in an AI-driven world.
The audience has migrated to AI-driven discovery channels. Publishers who structure and govern what they contribute can become essential partners in the AI economy.
When discovery moves from pages to answers, the value of expertise increases. Publishers have an opportunity not just to participate in the AI ecosystem, but to shape its trajectory.
Publishers' biggest readers are no longer human — they're AI systems. Yet most publishers have no visibility into this interaction. Closing that gap is the first step toward deciding how AI should use your content.