AI isn't Killing Search – it's Rewriting the Decision Layer

The dominant tech narrative of the past two years has cast AI and search as existential rivals, locked in a zero-sum fight for the future of how people find information online. That framing is wrong – or at least wildly incomplete.

What's actually emerging is a structural convergence: AI and search folding into a unified system that doesn't just retrieve information but resolves decisions. The implications for how platforms capture value and which companies come out on top are enormous.

Search becomes infrastructure

For decades, search engines served as the internet's primary interface, organising information, ranking relevance and directing traffic. AI changes that interface but not the underlying need.

Large language models, AI assistants and voice interfaces all depend on the same capabilities search has spent years refining: retrieving relevant data, ranking sources and understanding intent. Without that backbone, AI outputs degrade fast. Chatbots without retrieval layers hallucinate outdated or fabricated answers, a problem that has already become a reputational liability for early deployments.

That's why search-first companies, including Google, Microsoft, Baidu and Yandex, have moved faster on AI integration than almost anyone else. They already control the necessary infrastructure.

From navigation to resolution

The shift at the product level is from navigation to resolution. Instead of returning a ranked list of links, systems now synthesise answers, compare options and guide users directly to a decision, collapsing a multi-step process into a single interaction.

Jun Wang, former Baidu Brain Scientist and ex-AI researcher at Google, describes the change in architectural terms: "AI-driven search is transitioning from 'passive response to queries' toward 'proactive perception of intent.'

"Powered by multimodal understanding and user behaviour context aggregation, AI search has shattered the technical boundaries of traditional keyword-based queries. Systems can now accurately capture and transform natural language nuances and situational descriptions into precise results without requiring rigid user prompts.

"More importantly, AI search has established an end-to-end and closed loop spanning search, decision-making and action."

The decision layer: Where the money moves

The most consequential shift is the emergence of the ‘decision layer’ – the point where user intent is not just interpreted but resolved, end to end.

Historically, the user journey was fragmented. Search captured informational intent, marketplaces handled transactions and value leaked at every handoff. That separation is now collapsing as platforms combine real-time search infrastructure, recommendation systems and generative AI into a single loop.

Building that end-to-end system requires control across three layers: retrieval (search), ranking and personalisation (recommendation) and synthesis (generative AI). Only a handful of companies operate meaningfully across the full stack: Google, Amazon in e-commerce, Microsoft, Baidu and Yandex.

The latter has been particularly aggressive about integrating these layers in its home markets, making it a useful case study in what a vertically-integrated decision-layer platform looks like in practice. For everyone else, the practical path is external integration; all five expose parts of their infrastructure via APIs.

Commerce is where it gets concrete

The dynamic is most visible in e-commerce. Search has traditionally been the entry point for discovery; users query, compare and migrate to a separate platform to transact. AI collapses that boundary.

Karun Thankachan, Senior Data Scientist at Walmart eCommerce, frames it in terms of cognitive load: "The key shift I see is that the mental load of translating a need into the 'right' search query is disappearing. You don't have to guess something like 'lego puzzle' anymore; you can just say 'gift for my five-year-old' and AI uses its world knowledge to understand and bridge that intent.

"That's search and AI becoming a single decision layer that moves you from need to answer with far less effort, while creating new, more natural surfaces for brands to show up."

Publishers aren't dead – but their role is changing

As synthesised answers replace ranked link lists, publishers face a shift in how visibility works. Quality still drives ranking, but ranking now determines whether a source gets selected as a trusted input into a generated answer, not just whether it earns a click.

Mikhail Slivinskiy, Search Ambassador at Yandex, argues the change is less apocalyptic than it looks: "The ecosystem revolution makes us more adaptive but does not drastically shift our values-driven behaviour in the long run. Simpler interactions may conclude within the interface, but more complex queries still drive users to verify, compare and go deeper.

"For publishers, this extends their role: not just attracting clicks, but also acting as a bridge between the audience and the outcome, building niche expertise, guiding users toward decisions and partnering more closely with relevant businesses. And a choice, a purchase or a more long-standing affinity with their business partner, made by a user, could serve as a new indicator of publisher performance."

Convergence, not replacement

Vinija Jain, ML leader on Ask Gemini at Google, puts it plainly: "The goal isn't to constrain choice, but to improve decision efficiency. What matters is how effectively a system can move a user from intent to outcome by combining retrieval, ranking and generative synthesis into a cohesive experience.

"These capabilities aren't new; they are the same foundations search has spent years refining. Without that backbone, AI systems, assistants and agents quickly lose context, accuracy and trust.

"The competitive advantage will come from owning that decision layer, where the system doesn't just surface options but actively shapes the path to better, faster decisions."

For businesses, value is shifting from traffic acquisition to outcome ownership, across e-commerce, travel, financial services and beyond. The companies that will define this next phase aren't building standalone AI chatbots or incremental search features. They're the ones controlling the full loop between intent, answer and action.