Post by Gavin Lew, Conor Mackin, and Elena Beckman
If the Consumer Electronics Show (CES) 2026 confirmed one thing, it is that the traditional, linear retail funnel is obsolete. For decades, we operated under a model in which users searched, discovered, researched, compared, and purchased, usually within a predictable ecosystem. That era is over.
We are shifting from an economy of “finding” to an economy of “acting.” The emerging dominance of agentic AI and a potential future in which personal AI agents negotiate directly with retailer agents to execute tasks fundamentally rewrite the rules of engagement for product leaders, marketers, and researchers.
For executives, this shift presents a stark choice: cling to metrics that measure a dying linear journey, or pivot to understand the non-linear reality of the agent-to-agent economy.
From search to action: The rise of the agent
The immediate disruption is visible in how users interact with technology. We are moving beyond broad keyword queries. Large technology companies report that purchase completion rates are significantly higher when AI agents surface recommendations and transactional options directly, rather than through traditional search. Why? Because the friction of “browsing” is being removed.
Agent-driven commerce platforms show that storefronts are no longer static pages; they are dynamic interfaces where AI agents act on behalf of buyers to conduct research within the product catalog and provide comprehensive summaries to help aid customer decision-making.
This creates a new competitive reality: agent-to-agent commerce. CES suggests that a consumer’s personal agent will talk directly with a retailer’s agent to execute a purchase. This means that traditional marketing, designed to sway human emotion through display ads, becomes irrelevant in an interaction between two algorithms. To win here, your data must be structured to communicate value, availability, and specification directly to a machine, shifting the marketing focus to engaging agents that understand a user’s Jobs to Be Done (JTBD) and can act directly on their behalf.
Trust is the new currency
As consumers delegate purchasing power to AI, trust becomes the primary barrier to entry. We are currently in a “trust and verify” phase. To move to a “go buy this for me” phase, the underlying data interactions must be impeccable, and the agent must be consistent and reliable across time to solidify trust.
Data aggregation alone does not create personalization. Trust is built through tailored, guided interactions that clearly explain why a product recommendation fits a customer’s preferences, not by simply relying on purchase history. If an agent fails to demonstrate understanding of the user’s specific context, users will not authorize the transaction. To earn trust, the agent must act as a knowledgeable advocate on the consumer’s behalf, demonstrating expertise that allows customers to confidently relinquish certain controls.
Redefining value and identity
The definition of “value” has shifted. In an uncertain economy, customers no longer equate value with the lowest price alone. They now weigh time, reliability, quality, and durability alongside cost. As a result, customers go beyond simple price comparisons and rely on agents to help them determine which product delivers the strongest overall value across these dimensions.
Simultaneously, we see a divergence in how demographics define “relevance.” While some platforms rely on sales data (what is popular), others, like Pinterest, leverage visual data to algorithmically “tune to your taste.” This is critical for Gen Z, a demographic that rejects generic trends in favor of unique identity curation. For them, visual discovery is more helpful than text, and they use these platforms to discover and define “who they are,” not just what to buy. Brands that rely solely on “best seller” algorithms risk becoming invisible to a generation that values taste and individuality over algorithmic popularity.
Beyond the interface
This collapse of the funnel doesn’t make standard research obsolete, but it does expose the limits of focusing solely on isolated, linear tasks. While iterative usability testing remains vital for refining specific interactions, it provides the most value when paired with generative research to quickly understand how features and agent responses satisfy new JTBD use cases.
Real user journeys are now non-linear and cross-platform. A user might discover a product via a creator on YouTube, validate it through a query on Perplexity, and execute the purchase via an AI agent, never once visiting your homepage until the final click.
To reduce risk in this environment, we must change how we research:
- Investigate intent, not just interaction: We must revisit JTBD frameworks to provide the nuance AI agents currently lack. For example, a standard agent might find the cheapest product available, but a well-informed agent needs to know if the user historically prioritizes price or if they heavily research color, material, and finish because they value aesthetics and durability. AI agents can process transactions, but they struggle to intuit these deeper motivations without robust foundational research.
- Map the “spaces between”: Research must account for how users move between platforms. How does a user transition from a creator’s recommendation on TikTok to an AI-based validation step, or from a product listing’s review section to an independent source to verify consensus when they question the credibility of marketplace reviews?
- Test for trust: If trust is the currency of agentic commerce, we must measure it. We need to identify the specific interactions that build or erode confidence when a human hands off a task to an AI.
The path forward
The future of your product depends less on optimizing SEO for human readers and more on preparing for an ecosystem where agents make decisions. Your ability to clearly communicate value, especially how your product aligns with user identity and their job to be done, will determine whether the agent even surfaces your product for consideration.
The funnel hasn’t just changed shape; it has dissolved. Our research strategies must be bold enough to follow the user into the chaos that replaced it.