Agentic Commerce
January 26, 2026
Today's Briefing: January 26th
NewsAgents Take
Today’s links all point at the same structural reality: agentic commerce will not scale on “protocol hype” alone. The near-term impact is still concentrated in discovery and decisioning, while fully on-platform checkout remains a small slice of ecommerce. At the same time, enterprise readiness is being overestimated, with high expectations but weak measurement and uneven operational maturity. Europe is trying to turn regulation into an adoption asset by pushing auditable control points and traceability. The common thread is that trust and governance, plus distribution, will matter more than agent UX.
What brands and retailers should take from Google’s UCP
EMARKETER’s framing is blunt: AI platforms may drive more shopping intent, but most transactions will still close off-platform for years. Their key stat is that on-platform sales will be just 1.7% of US retail ecommerce spending in 2029. That matters because UCP’s biggest near-term effect is likely to be standardised capability discovery and smoother handoffs, not a sudden migration of checkout into AI surfaces. Brands should optimise for how agents interpret product truth (catalogue structure, availability, policies) and for where they want conversion to happen, rather than assuming the “new checkout surface” will be the main battlefield.
Agentic AI expectations are high, readiness is patchy
The HBR Analytic Services report (sponsored by AWS) has useful numbers that map cleanly to the agentic commerce debate. 84% of respondents agree agentic AI will transform their organisation, and 90% expect most organisations in their industry will use agentic AI in the future. But only 5% say their organisation has well-defined success metrics for agentic AI, which is a quiet red flag for anyone betting on near-term ROI narratives.
The adoption distribution is also telling. In a July 2025 survey of 623 HBR audience members, 34% report they are already using agentic AI (11% widely, 23% in limited use cases), 40% are piloting or preparing, and 26% are effectively stuck at “considering or not moving forward.” Where organisations are already piloting or using agentic AI, reported benefits include greater productivity (36%) and better customer experience (26%). This looks like a story of selective deployment, not sweeping autonomy.
https://hbr.org/sponsored/2026/01/agentic-ai-expectations-readiness-results
Europe’s angle: make regulation an adoption asset
The European Commission summary of the StepUp StartUps Consortium work lands in a practical place. It highlights a core blocker for agentic systems: multi-step autonomous actions blur responsibility across tools and systems, which challenges compliance frameworks and increases the need for continuous traceability and meaningful human oversight. It also flags uneven digital readiness across Member States and risk-averse implementation due to limited reference cases.
The most relevant part for commerce is the direction of travel: build clear, auditable control points, strengthen sovereign data and infrastructure foundations, and scale via flagship use cases that prove safety and value. Even if you disagree with the policy posture, it is a reminder that agentic adoption will be pulled forward by governance primitives, not marketing.
Video context: the AI business model conversation is shifting
The linked video is a recent segment titled “AI honeymoon is over and this will be its hardest year yet,” featuring a discussion of OpenAI and its business model. This is relevant because agentic commerce sits downstream of the same constraint: distribution is abundant, but sustainable unit economics and trust frameworks are still being negotiated in public.
