Tier 1 — 高信息差（值得立即关注）

1. OpenAI × Visa 合作：AI Agent 直接购物支付

• Visa 将支付网络接入 ChatGPT，AI Agent 可以代用户下单付款
• 综合分 9 / 差距分 4 | 来源：AP News, OpenTools
• 为什么重要：这意味着 AI Agent 从“帮你搜“进化到“帮你买“，Agentic Commerce

OpenAI and Visa Partner to Let AI Agents Make Purchases

Source: https://opentools.ai/news/openai-visa-agentic-commerce-ai-purchases-2026

Agentic Commerce

OpenAI and Visa announced a partnership enabling AI agents to make purchases on behalf of users through ChatGPT, marking the most significant infrastructure play yet for autonomous AI‑driven commerce.

AI Agents Get a Credit Card

OpenAIandVisaannounced a partnership on June 10 that will let AI agents make purchases on behalf of users through ChatGPT and other OpenAI platforms. The integration connects Visa’s global payment network directly into OpenAI’s ecosystem, allowing developers and merchants to accept agent‑initiated transactions with the same security infrastructure that powers Visa’s 300 billion annual transactions, according toBloomberg.

In practical terms, a user could tell ChatGPT to order paper towels or pay a recurring bill — and theAI agentwould handle the transaction within user‑defined guardrails. The move marks the most significant infrastructure commitment yet to agentic commerce, where AI agents move from discovery and recommendation to completing purchases autonomously.

How the Payment Infrastructure Works

The partnership is built onVisa Intelligent Commerce, a program launched in April 2025 that opens Visa’s network rails to developers building AI agents for commerce. Transactions use tokenized Visa credentials — replacing sensitive card data with secure network tokens bound to specific agents and use cases — with real‑time authorization and continuous fraud monitoring, according toVisa.

Each agent‑initiated transaction operates within clearly defined user permissions: spending limits, merchant category restrictions, and required approval thresholds. This means an agent can be told “only buy groceries under $50 at these three stores” — and the payment infrastructure enforces those rules at the network level, not just the application level.

• Tokenized credentialsSensitive card data replaced with network tokens bound to specific agents and use cases
• Spending limitsPer‑transaction and per‑category caps set by the user before the agent acts
• Merchant restrictionsAgents can be restricted to approved merchant categories or specific retailers
• Real‑time authEvery transaction authorized in real time with Visa’s existing fraud monitoring

Codex Integration and Enterprise Workflows

Beyond consumer purchases, the partnership also explores enterprise applications built onCodex,OpenAI’s coding agent, according toPYMNTS. The companies intend to embed payment primitives and trusted agent identity signals into developer‑focused experiences — meaning Codex could eventually handle procurement, invoicing, reconciliation, and payment execution within automated business workflows.

“By integrating with Visa Intelligent Commerce, we’re building the infrastructure for secure, transparent, and user‑controlled agentic transactions, helping people do more with AI agents while maintaining confidence that payments are being handled safely and securely,” Marco Mahrus, OpenAI’s Head of Partnerships for Commerce, said in a statement to.3

From Instant Checkout to Network‑Level Integration

The Visa partnership represents a more ambitious approach thanOpenAI’s earlier payment experiments. In March 2026, OpenAI discontinuedInstant Checkout, a simpler ChatGPT payment feature, after just six months, shifting instead to working with retailers on dedicated apps within ChatGPT, according toFinextra.

“AI will transform commerce more profoundly than the internet or mobile technology ever did. As AI agents become active participants in the economy, Visa’s focus is to ensure transactions are trusted, secure, and seamless,” Jack Forestell, Visa’s Chief Product and Strategy Officer, said in the2.

“AI will transform commerce more profoundly than the internet or mobile technology ever did. As AI agents become active participants in the economy, Visa’s focus is to ensure transactions are trusted, secure, and seamless.”

Jack Forestell- Chief Product and Strategy Officer, Visa

The Competitive Landscape: Everyone Wants Agent Payments

Visa andOpenAIare not alone in pursuing agentic commerce. Mastercard launched its own agentic AI payments tools in October 2025. Stripe partnered with OpenAI on ChatGPT payments in September 2025. Both Visa and Mastercard have been recruiting banks across the UK and Europe for AI payments trials, as the financial infrastructure layer races to be the rails on which autonomous AI agents run.

“The organizations that succeed in agentic commerce will not simply deliver smarter AI. They will deliver confidence,” Forestell added in the2. The emphasis on trust and security infrastructure — rather than just AI capability — signals that payments companies see their existing networks as a moat against pure‑play AI competitors who might otherwise commoditize the transaction layer.

What This Means for AI Builders

For developers building AI agents, the Visa-OpenAIpartnership provides a clear path to monetization: agents that can discover, recommend, and nowpayfor products create a complete commerce loop. The developer tooling angle — embedding payment primitives into Codex — suggests OpenAI sees agentic commerce as a platform feature, not just a consumer convenience.

The partnership also raises the bar for agent security. With Visa’s network‑level enforcement of spending limits and merchant restrictions, developers get a built‑in compliance layer — but also lose some flexibility compared to building custom payment flows. The tradeoff is clear: faster time‑to‑market for agentic commerce features, but within Visa’s controlled infrastructure. For startups building AI shopping agents, this partnership may define the default payment stack for the next generation of commerce applications.

Sources

1. 1.Bloomberg(bloomberg.com)
2. 2.Visa(corporate.visa.com)
3. 3.PYMNTS(pymnts.com)
4. 4.Finextra(finextra.com)

Tags