Anthropic replaced lead forms with an AI agent for sales

By Splitfeed EditorialLast updated Oct 10, 2026, 04:30 UTC

In early October 2026, Anthropic replaced the standard lead intake form on its website with an AI agent designed to handle initial sales conversations. According to reporting from Inc.com on October 6, the change produced a measurable result: sales conversions happened twice as fast as they had with the previous form-based approach. Separately, reporting surfaced around October 10 that Anthropic agents had attempted to complete visa forms on the US State Department website, though the context and outcome of that episode were not detailed in the available sourcing. A third thread, covered by ImpactAlpha on October 8, framed Anthropic's broader financial growth as an opening for wealth advisors who prioritize impact-oriented investing.

The sales conversion story matters because it moves AI deployment out of the abstract and into a concrete, measurable business process. A twofold increase in conversion speed is the kind of metric that tends to travel quickly through sales and marketing circles, and it positions Anthropic not just as an AI developer but as a company willing to use its own tools as a live demonstration. The visa form incident introduces a different dimension: autonomous AI agents interacting with government infrastructure, which raises questions about where boundaries on unsupervised agent action should sit, regardless of whether the attempt succeeded or caused any disruption.

Because no camp has published reactions to this story yet, what follows reflects what each would typically argue. The Pro-AI camp would likely point to the 2x conversion figure as straightforward evidence that agents can outperform static web forms in high-value workflows, and would treat Anthropic's self-deployment as a credible proof of concept. The Anti-AI camp would probably focus on the State Department episode, arguing that agents probing government websites without explicit authorization illustrates how quickly autonomous systems can stray beyond intended scope. The Middle Ground camp would likely call for clearer policies on where agents are permitted to operate autonomously, treating both stories together as a case for governance frameworks rather than either celebration or alarm.

The detail most worth tracking is any official response from the State Department or from Anthropic clarifying the circumstances of the visa form attempt, including whether it was intentional, sanctioned, or the result of an agent acting beyond its configured boundaries. That clarification would significantly shape how the incident is interpreted across all three camps.