Agents start taking your money: in-conversation payments hit 71.3% as shoppers hand AI the list

The checkout moves inside the conversation

The moment of purchase has always been where AI agent demos break: the customer gets transferred to another system or sent a payment link, splitting one conversation into two. Sierra and Stripe are trying to remove that seam. Their partnership puts payments inside the AI agent conversation — when it's time to pay, the agent moves into a secure payment mode, the customer enters card details on the keypad, and the details are routed to Stripe to process. Sierra is now listed in the Stripe App Marketplace, so businesses can connect an agent to their Stripe account without a one-off integration. m:…2512

The proof point is Sierra's own case study and should be read as such: Cedar, which helps health systems and physician groups with billing and payments and was among the first companies to take payments over the phone through the integration, reports that 71.3% of calls involving Kora end in a successful payment. Patients calling about a bill can now get their questions answered and pay in the same conversation. A single company's first-party number is not an industry benchmark — but as a first data point on whether people will actually hand money to an agent mid-call, it clears the bar. m:…2512

Demand is already here

Demand looks further along than the infrastructure. Forter's survey — conducted online by The Harris Poll for Forter on September 15–16, 2026, among 2,017 US adults, accurate to within ±2.7 percentage points — found 67% of Americans plan to engage in AI holiday shopping in 2026. Among shoppers, 47% are comfortable letting AI find deals, 44% are comfortable with AI product search and comparison, and 20% are comfortable letting an AI assistant select the product itself. The trust boundary is visible in the same data: 26% would be more comfortable if they could require final approval, and 24% if they could set spending limits. m:…0654

The younger cohort is already there: 68% of Americans ages 18–34 have discovered or purchased from a new retailer because an AI tool recommended it, and more than three-quarters of every age group under 55 plans AI holiday shopping this year. Forter also says it has seen a sharp uptick in agentic orders across its network since Meta launched its personal AI assistant, Muse. The pattern is consistent — shoppers hand the AI the list, and they want a veto at the end. m:…0654

Agents need to talk to each other

The other half of the commerce story is agents talking to each other. Genesys and Salesforce say they are introducing Agent2Agent orchestration across the Genesys Cloud platform and Salesforce, letting AI agents resolve issues directly with each other instead of routing everything through human handoffs and custom middleware. The pitch is that A2A connects platforms at the intelligence level rather than just the data level — a Genesys Cloud agent can create a case or update an opportunity in Salesforce, while a Salesforce agent surfaces CRM data or triggers actions back in Genesys Cloud, in real time. m:…4179

The important caveat is tense: this is announced, not shipped — the companies are preparing to break through the ceiling of desktop integrations, and A2A will handle coordination at the agent layer. What they promise is that instead of maintaining custom middleware to pass data between platforms, coordination happens agent-to-agent, with agents nudging stalled approvals or triggering process steps in the background. It follows a September announcement of A2A interoperability with Salesforce Agentforce. Take it as infrastructure for a world where the agent that takes your order talks to the agent that fulfills it — not yet as that world. m:…4179

What it means

Reasoning, not reporting: the agentic-commerce stack is arriving in the right order — the payment rail (Sierra-Stripe), the demand (holiday shoppers), and the coordination layer (A2A). There's also a protocol bet: Sierra says it's working with Stripe and others on the Personal Agents Protocol, an open standard with Meta for how personal agents interact with businesses, payments included. The open questions are all about trust and scale: 71.3% is one company's number, A2A is a promise, and the 20% of shoppers who'd let AI pick the product is still the minority. m:…2512