The End of Search: Why AI Visibility is a Systems Problem

In this conversation, Liam Dunne and Ben Moore of Discovered Labs map the shift from traditional search engine optimization to the complex, non-deterministic world of AI search. The core thesis is that the paint by numbers era of SEO is over. Visibility is no longer about keywords; it is about influencing a multi-stage reasoning system. The hidden consequence of this shift is the transition of websites from human facing destinations to agent facing data sources, where the goal is to be ingested, processed, and cited by an LLM rather than clicked by a human. For business leaders and marketers, the advantage lies in moving away from surface level metrics toward modeling the gray box of AI reasoning. Those who treat AI visibility as a systems engineering challenge rather than a content marketing task will capture the long tail demand that traditional search has long ignored.

The Hidden Cost of Fast AI Solutions

Most marketers approach AI visibility by chasing immediate, visible results, such as getting a single prompt to cite their brand. Dunne and Moore argue that this is a tactical error. Because modern AI search engines use query fan out and multi step reasoning, a brand might be retrieved as a source but rejected at the final citation stage.

You can go into like essentially unless there was a Google Core update, essentially like the rankings were essentially very deterministic... and it was pretty finite, pretty manageable, all pretty controllable. And again, paint by numbers. And now we have kind of gone into a space where there are sources of non-determinism.

-- Ben Moore

The system is not a static index; it is a dynamic, gray box reasoning engine. When teams optimize only for the final output, they ignore the consensus and agreement algorithms that occur behind the scenes. Moore notes that agents often use sources to ground their reasoning without ever citing them. This creates a hidden cost where teams waste effort on content that grounds the model but provides zero brand attribution.

How the System Routes Around Your Strategy

A common mistake is assuming that high engagement, like upvotes on Reddit, correlates directly with AI citations. Dunne analysis suggests the opposite: the system prioritizes content relevance over social vanity metrics.

We have observed is engagement in terms of upvotes, we have not seen that correlate with increased citations. What we have seen highest correlation with is the actual content itself.

-- Liam Dunne

This reveals a systemic dynamic: the LLM is looking for specific entity relationships within the content to satisfy its internal reasoning. If your content contains the precise technical or contextual entities the model needs to build its answer, it will extract that passage regardless of whether the post has one upvote or one thousand. The implication is that gaming the system with bot driven engagement is becoming increasingly futile as models implement fingerprinting and hashing to identify and discard synthetic, low value noise.

The 18-Month Payoff: From Read-Only to Read-Write

The most profound shift identified is the transition from AI as a discovery tool to AI as an active participant. Currently, most businesses focus on discoverability, or getting their name into an LLM response. Moore and Dunne predict a rapid evolution toward read-write interaction, where agents do not just recommend a service; they execute tasks, book demos, and integrate tools on behalf of the user.

This creates a competitive moat for those who prepare their digital architecture for agents now. While most competitors are busy optimizing for human conversion rates, the early adopters are building agent friendly infrastructure. This requires patience, as the payoff is not in current traffic, but in the ability to capture the automated transactions of the near future.

Key Action Items

• Audit for Entity Density (Immediate): Stop focusing solely on keywords. Audit your website to ensure it clearly defines your brand, your ICP, and your specific value propositions in a way that allows the model to easily map your entity to the problems it solves.
• Shift to Consensus Marketing (Over the next quarter): Instead of chasing single links, invest in multi channel presence (Reddit, YouTube, G2) to build a web of votes of confidence. The model uses these to verify your brand authority during its agreement or consensus phase.
• Model the Gray Box (6-12 months): Stop treating AI visibility as a marketing task and start treating it as a data science problem. Build a pipeline to track your brand citation rate and sentiment across different queries to bound the uncertainty of your AI presence.
• Prepare for Agent-First UX (12-18 months): Evaluate your conversion funnels. If your demo forms or checkout processes are confusing for a human, they are likely impossible for an agent. Simplify your site architecture to be agent navigable to capture automated conversions.
• Prioritize Consistency Over Volume (Ongoing): Ensure your brand messaging is identical across all platforms. In an LLM driven world, the blurb surrounding your brand across the internet is more important than the link juice of a single high authority backlink.