Major corporations have quickly woven AI into every aspect of their business in the pursuit of cost savings and revenue growth. The digital commerce journey is no exception and effectively all large merchants in the US, UK and Europe have implemented generative AI in some part of their customer experience. LLMs and other AI tools are also growing as a traffic source for merchants: Adobe data shows that AI-driven traffic to retail and travel merchants has increased more than 10x over the last 15 months.
Despite these signs of progress, LLMs and other AI tools remained less than 1% of total traffic for retail and travel merchants during the 2025 holiday season. And Flagship market research indicates that true agentic payments (where AI agent can initiate and complete checkout on the customer’s behalf using permissioned payment tokens, virtual cards, digital wallets, stablecoins or other rails) are at the very early stages of deployment. We will continue to track progress on agentic commerce and payments as this new space develops in 2026 and beyond.
1. Agentic Solutions Observed at Top US/EU Merchants
(implemented solutions across US & EU top 10 companies ranked by revenue, n=80)
Across the top 10 US and top 10 EU companies per vertical, customer-facing Generative AI is now widely deployed, most often in Q&A chatbots, image searches, and product support. The more differentiated story is the step-up into agentic commerce (multi-step journey orchestration) and the relative scarcity of agentic payments (checkout execution by an agent).
Key observations:
- Most top US & EU merchants have deployed customer-facing GenAI for Q&A, guided discovery, and product support.
- Agentic commerce is still early-stage, deployments are concentrated in a smaller cohort where agents can plan and orchestrate multi-step journeys (discover → compare → assemble cart/itinerary) but still stop short of executing payment.
- Payments is the hard boundary to cross; the market is beginning to standardize on network tokenization and delegated authorization models (via emerging protocols such as UCP / AP2, ACP, and x402) to make agent-led checkout feasible.
2. AI-Driven Commerce: Traffic Signals
- AI-driven visits to merchants are up 10x over October 2024 but growth tapered or retreated during late 2025, which may be a sign that expansion beyond early adopters may require deeper integration with web browsers, search, social platforms, or other experiences beyond LLM chat.
- While data on agentic payments is not available, we believe they are still de minimis given Adobe’s data on AI-driven visits.
- New standards like Google’s Universal Commerce Protocol should help companies across the commerce value chain collaborate more effectively on monetizing agentic traffic.
3. OpenAI’s Largest Users by Vertical: Where Token Consumption Clusters
(selected companies' logos)
Flagship Commentary & Highlights
- Intelligent Copilots & Agents (29%) and Infrastructure & Enablement (26%) account for more than half of total token use, showing that most consumption supports the systems and tooling behind AI applications rather than end-user interfaces.
- Workflow & Orchestration (17%) forms the connective layer, driving automation and integration across agents, networks, and APIs.
- Digital Merchants (15%) are actively leveraging LLMs, introducing AI into the checkout page and assisting during the checkout flow.
- Research / Academia (8%) organizations are applying generative AI to technical exploration and innovation.
- Payments (3%) and Commerce Enablers (1%) are a small share of company count but play a key enabling role for others, especially platforms like Shopify and Mercado Libre which enable millions of SMBs.
Conclusion:
Near-term progress is likely to look like agent-assisted commerce at scale, with better discovery, planning, more automated cart/itinerary assembly, while agentic payments will rely on industry progress building solutions for trust, governance, and interoperability and finding the right use cases where buyers are eager to delegate shopping and payment to AI.
Please do not hesitate to contact Ben Brown at Ben@FlagshipAP.com with comments and questions.