What: Four distinct AI-driven retail futures are now plausible — and retailers must build for all of them, not just the most likely one.

Why it is important: Because in agentic commerce, being absent from AI systems is not a visibility problem. It is a revenue problem.

AI agents are emerging as the new intermediaries in retail — systems that discover, compare, and purchase on a consumer's behalf, potentially without the consumer ever visiting a product page. They insert themselves between brand and buyer at the moment a purchase decision forms, making it possible to complete a transaction without the shopper ever encountering a brand's message, page, or advertisement. Retailers now face a specific decision: adapt infrastructure and engagement models for a world where an algorithm makes the first product judgement. This demands data governance, agent-ready APIs, and a consistent focus on two imperatives BCG identifies as constant across all agentic scenarios — discoverability (being found by the systems that mediate discovery) and desirability (being chosen by the consumers those systems serve). The strategic choice is equally concrete. Retailers must decide whether to become a destination platform — owning the consumer relationship through loyalty programmes, membership, and private labels, as Target has done — or to excel as an evaluation platform, competing on data transparency, fulfilment reliability, and machine-readable product information, as Wayfair has. Each path requires full commitment; half-measures in either direction compound the risk. Scenario planning, not static forecasting, is the appropriate tool here. BCG maps four plausible agentic futures, each demanding different capabilities, and argues that building across all four — rather than betting on one — is the only defensible strategy. Retailers who read the signals early and build accordingly will not just adapt to the agentic future. They will write its rules

IADS Notes: McKinsey (November 2025) projects up to $5 trillion in global retail revenue from agentic commerce by 2030. That figure makes data governance and agent-ready APIs a commercial priority, not a technical one — and sets the context for the rest of this reading list. Liontree (April 2026) documents the present-tense version of that shift: AI-driven shopping is already mainstream among younger and higher-income consumers, with retail power moving from brand websites to AI platforms. Journal du Net (September 2025) confirms the urgency for retailers to structure data and APIs for machine readability, as agentic commerce moves from early adoption to the expected standard. Inside Retail (November 2025) focuses on what that restructuring means in practice — AI-powered agents are already automating shopping decisions, and the retailers currently invisible to those agents are not failing strategically; they are simply absent. Ian Jindal (February 2026) closes the picture: AI is dissolving sector boundaries, and the retailers who invest in agent-ready systems and first-party data now are positioning themselves to maintain relevance as AI-driven shopping accelerates. These five sources make the same case as the BCG article, from the ground up: the agentic infrastructure question is not theoretical. It is already determining who is found, compared, and purchased.

Agentic scenarios every marketer must prepare for