The current AI search landscape (AI Overviews, citations, conversational discovery) is the intermediate state.
Agentic AI, where systems autonomously research, shortlist, and complete purchases on behalf of users, is the step-change ahead.
Where we are right now.
We are firmly in an advisory-before-autonomous phase. Consumers increasingly use AI to assist with research, but very few trust it to buy without oversight.
The gap between “help me decide” and “buy for me” will persist for most categories for several years. The immediate opportunity for marketers is ensuring that your brand is well-positioned at the point where AI-assisted research hands off to a human final decision.
Technical considerations.
Competing commerce protocols, i.e. Google’s UCP and OpenAI/Stripe’s ACP, mean structural access to AI distribution requires deliberate technical decisions.
Google’s agentic checkout, announced in January 2026, already lets US shoppers complete purchases within AI Mode without leaving the interface. Zero-click commerce has started.
The infrastructure for an agentic world (structured product data, accurate brand information across platforms, community reputation, machine-readable content, platform-specific technical access) is being built now. Brands that treat this as a future concern will find competitors have already laid the foundations.
The agentic future is coming, but the immediate opportunity is in the gap between AI-assisted research and the human making a final decision. Build the infrastructure now.
Recommended Reading
McKinsey: "The Agentic Commerce Opportunity"
McKinsey projects agentic commerce could account for up to $1 trillion in orchestrated US retail revenue by 2030. This report sets out why the shift from AI-assisted research to AI-completed transactions isn’t a distant prospect.
Google Developers Blog: "Under the Hood: Universal Commerce Protocol (UCP)"
Written by Google’s engineering team, this is the technical blueprint behind UCP – the open standard that lets AI agents complete purchases across Google’s surfaces without custom integrations.
OpenAI: Model Context Protocol documentation
MCP is the emerging standard that allows AI agents to connect to external tools and data sources — and it’s rapidly becoming the plumbing behind agentic experiences across platforms. If you want to understand how agentic AI actually works under the hood, this is the primary source.
Anthropic: "Code Execution with MCP"
Anthropic’s own engineering team explains how MCP enables Claude to take real-world actions — running code, accessing data, interacting with systems — rather than simply generating responses.