Only 23% of Merchants Track AI Traffic as Agentic Commerce Grows

Payments orchestration platform Spreedly expects artificial intelligence to move beyond product discovery and recommendations into agentic commerce, where software can initiate purchases on a user’s behalf. The shift could turn automated traffic, long treated by merchants mainly as a fraud or security concern, into a source of customers with genuine buying intent and a distinct transaction channel for online businesses.

Spreedly said businesses will need to distinguish legitimate AI agents from malicious bots while reducing friction for approved agents across checkout, authentication and payment. That creates a new infrastructure challenge for merchants and payment providers seeking to accommodate machine-led purchases without weakening fraud controls. The report disclosed no merchant partnerships, transaction value, growth figures or specific announcement date, leaving the development at the market-strategy stage.

Ecommerce has traditionally relied on shopping carts, address forms and payment-verification screens, creating friction that can cause shoppers to abandon purchases. Stripe says advances in generative AI, digital payments and identity technology could shift those tasks to autonomous agents. Such systems may search for products, compare offers and complete transactions on a customer’s behalf, turning agentic commerce into a new gateway between merchants and buyers.

Stripe’s latest assessment suggests conventional checkout pages will increasingly move into the background. Shoppers could purchase directly from a product page or authorize an AI agent to place orders and make payments within predefined budgets and preferences. The change could sharply shorten the path from product discovery to purchase while requiring merchants to rethink payment authorization, fraud controls and accountability. The related report did not specify a launch date, transaction volume or investment amount.

Artificial intelligence is moving beyond product recommendations toward agentic commerce, where software can research options, negotiate terms and execute purchases for consumers or businesses. Item-level information—including SKUs, prices and availability—gives agents the structured context needed to compare offers against a buyer’s budget and preferences. Tokenized credentials and verifiable mandates can then connect those decisions to payment rails, although fraud, authorization and liability remain unresolved challenges for the emerging autonomous-enterprise model.

Global Payments said on Dec. 12, 2025 that merchant endpoints could expose item-level data, negotiate with shopping carts and capture buyer and cart mandates before submitting tokenized payment credentials. The company is participating in Google’s Agent Payments Protocol, or AP2, and plans to adapt its APIs, hosted checkouts and MCP servers for agent-led transactions. Its example involved an agent finding two Chicago concert tickets for less than $400 under preset conditions. Global Payments disclosed neither an investment amount nor a commercial rollout date.

Agentic AI is moving generative AI beyond recommendations to acting on consumers’ behalf, searching for products, comparing prices, placing orders and completing payments based on their preferences. PYMNTS Intelligence and Worldpay say the technology will reshape retail product discovery and payment authorization. Identity verification, transaction liability and consumer trust will also be critical to widespread adoption.

The latest report shows that 43% of retailers are piloting AI shopping agents, while 45% of consumers are willing to use such services, indicating that both supply and demand are in the early validation stage. Attention has recently shifted to payment mechanisms, including how to verify authorization, prevent fraud and resolve transaction disputes. The available event data does not specify the report’s publication date or any transaction value.

OpenAI and Visa are moving generative AI beyond providing advice and toward conducting transactions on users’ behalf. The partnership focuses on enabling AI agents to make automated payments in B2C shopping and enterprise workflows, using Visa’s global payment network, credential tokenization and identity verification to reduce the risk of fraud and unauthorized transactions.

As of July 20, 2026, Visa had announced plans to integrate real-time authorization and secure authentication technology into OpenAI experiences, allowing AI agents to complete payments within user-defined permissions. The companies have not disclosed a launch date, the value of the partnership or the initial markets. The industry is also watching whether banks could consequently be cut out of the agentic-payments gateway.

Visa Acceptance Solutions commissioned PYMNTS Intelligence to examine how the payments ecosystem is adapting to Agentic Commerce. Under this model, autonomous agents search, make decisions and pay on consumers’ behalf. Transactions may span websites, apps and physical channels, requiring identity verification, authorization and risk controls to be redesigned.

The latest report found that nearly 80% of acquirers are ready to support the omnichannel infrastructure required for Agentic Commerce, indicating that the industry is moving from infrastructure development to implementation. The report disclosed no transaction value, and the event information did not specify a publication date. The current priority is enabling autonomous agents to complete card transactions seamlessly across environments.

Generative AI is evolving from offering recommendations to selecting products and making payments on consumers' behalf, giving rise to agentic commerce. Andrew Torre, president of Visa Value-Added Services, said that as AI agents gain the ability to transact autonomously, the competitive focus will shift from model performance to trusted identity, payment authorization and accountability. These factors will directly shape the adoption of automated payments.

Visa recently said it is embedding identity verification, consumer authorization and transaction-dispute controls into payment flows. The aim is to ensure that an AI agent can confirm a user's intent and handle anomalous transactions when making purchases. The related report provided no specific launch date, investment amount or transaction volume, but indicated that Visa views security and trust mechanisms as central to deploying agentic commerce.