Anthropic’s Chief Commercial Officer Paul Smith told Semafor that enterprise customers are asking the company to build its own version of OpenClaw. When asked directly whether customers are making that request, Smith said: “They are. Without speculating too much on our product roadmap, [it] evolved pretty quickly.”
Smith declined to provide product details, but the admission is the first on-record confirmation from a named Anthropic executive that customer demand for an OpenClaw rival is shaping the company’s roadmap. Semafor’s Reed Albergotti characterized the situation as “more a matter of when, not if” Anthropic introduces a more direct OpenClaw competitor or embeds equivalent capabilities into its existing products.
What Anthropic Has Shipped So Far
Anthropic’s closest product to OpenClaw today is Dispatch, a feature inside Claude Cowork that lets users message Claude from a phone while it works on their desktop. Smith pointed to Dispatch specifically: “Think of the features that people found useful — the ability to have a whole bunch of agents working on my desktop but remote control from my mobile device.”
That’s a narrower offering than OpenClaw, which runs persistent multi-agent sessions, supports a marketplace of pre-built agents, and handles enterprise-grade orchestration. The gap between Dispatch and a full OpenClaw competitor is significant.
The Broader Arms Race
The Semafor report follows Axios reporting from March that OpenClaw’s popularity triggered a coordinated product sprint across NVIDIA, Anthropic, and Perplexity. NVIDIA shipped NemoClaw as a security-enhanced alternative. Perplexity launched its Personal Computer agentic system.
Semafor also references the 500,000-line Anthropic source code leak from late March, suggesting the leaked code revealed additional details about Anthropic’s agent platform ambitions, though the article doesn’t specify what those details showed.
The Adoption Problem
Smith offered a frank assessment of enterprise readiness. When asked whether Anthropic’s own customers can keep up with the pace of agent development, he used an analogy: “Rolling out broad-based AI to an entire employee base is like putting a Peloton in everyone’s sitting room. It doesn’t mean everyone’s going to start to get fit.”
For builders on Claude: if Anthropic ships a full OpenClaw-equivalent, the integration between model and runtime could be tighter than what OpenClaw offers for Claude users today. For builders on OpenClaw: the question is whether OpenAI’s ownership accelerates OpenClaw development fast enough that an Anthropic alternative arrives too late to matter.