ServiceNow embeds Anthropic Claude as its default Build Agent model

ServiceNow Build Agent has a new default AI model: Anthropic’s Claude. ServiceNow will also use Claude to power other application development workflows and is building agentic workflows combined with Claude for industries such as healthcare and life sciences.

The collaboration deepens the integration between Claude and Build Agent so that professional developers can create more complex agentic workflows that “reason, act, and execute autonomously,” ServiceNow said, adding that it will enable customers to create their own agentic workflows or extend ServiceNow’s out-of-the-box automation with what it described as “enterprise-grade oversight.”

ServiceNow is also collaborating with OpenAI on automation and speech-to-speech technology, it announced last week.

“Each of these partners have unique use cases and experiences and expertise,” said Amit Zavery, ServiceNow’s president, COO, and CPO. “We’re looking at who are the right ones to work together, which areas we want to focus on, and with what.” Customers still get to choose which model they use, he said.

ServiceNow is also using Claude internally alongside purpose-built solutions to speed up software implementation, targeting a 50% reduction in implementation time from initial sales discussions through autonomous deployment. It will offer the same technology to partners and customers.

Working together on agentic workflows

Anthropic and ServiceNow will jointly market agentic workflows built on the ServiceNow AI platform for industries such as health care and life sciences, to support tasks such as claims approval and research analysis running within ServiceNow’s governed workflows. Health care claims authorization could be reduced from days to hours while also decreasing costs, ServiceNow said.

All that can be done under the watchful eye of ServiceNow’s AI Control Tower, which encompasses governance, discovery, operation, operations management, observability, and whole lifecycle management of AI systems, Zavery said: “We can provide you the ability to track usage, compliance, security policy, things like that, so that as a customer, you know you have control of your AI implementations and you’re not worried about some rogue things happening without any kind of visibility.”

Info-Tech Research Group advisory fellow Scott Bickley sees ServiceNow fast becoming the leading solution for enterprise AI orchestration, and is seeking to surround itself with other leaders.

“ServiceNow is clearly seeking the best tool for the use case at hand. The recent OpenAI partnership focuses on agentic AI and automation, speech-to-speech interactions, and general-purpose workflow management. OpenAI is embedded in the ServiceNow ecosystem as a preferred AI engine, however, the Anthropic announcement demonstrates ‘good enough’ is not good enough for [ServiceNow Chairman and CEO] Bill McDermott.  He wants the best tool for the job at hand.”

When it comes to automating applications and workflows, Bickley said, “Claude Code is arguably the hottest thing going in the vibe coding/code augmentation space, as it operates directly within the user’s terminal, versus traditional AI coding tools that may be embedded as extensions in IDEs such as VS Code or Cursor. Claude Code acts as an agentic partner, excelling in complex scenarios such as planning, executing, and verifying multi-step tasks automatically.”

A bet on orchestration

Moor Insights & Strategy Principal Analyst Melody Brue also sees ServiceNow focusing on AI orchestration. “ServiceNow is betting on its AI control and orchestration layer, with Claude as the latest proof point that customers can get more specialized, regulated‑domain value without ever leaving that governed platform. For customers, this should be a clearer, faster path to agents. Customers now get a default agent‑building stack instead of having to choose and wire up models themselves, which should reduce friction and standardize patterns for reasoning‑heavy agents,” she said.

“This is also a better fit for regulated and high‑stakes domains. By explicitly tying Claude to healthcare and life sciences use cases and positioning Opus 4.5’s safety and benchmark performance, ServiceNow can appeal to customers who want strong guardrails and domain‑specific AI without abandoning the broader multi‑model platform,” she said.

The Anthropic and OpenAI agreements are a strategic commitment to architectural control and model plurality, according to Sanchit Vir Gogia, chief analyst at Greyhound Research. “This is the company moving from an AI-enabled platform to an AI-governed operating system. What might look like overlapping partnerships on the surface is in fact a well sequenced shift in design philosophy: from treating AI as an embedded feature to architecting AI as the execution fabric of the enterprise.”

The order of the agreements was important, he said. “OpenAI’s models were brought in to enhance the user-facing interface: voice, summarization, and generative interactions. Claude, on the other hand, is being embedded as the default model behind Build Agent. That is the engine for writing, deploying, and automating workflows at scale. It’s not just supporting the user experience; it’s designing the work itself.”

This tells us that ServiceNow is “engineering a separation of cognitive duties across its platform,” he said. But, he noted, “the most significant product in ServiceNow’s AI portfolio isn’t Claude or GPT integration. It’s the AI Control Tower. That is the infrastructure that makes everything else workable. Without it, this entire multi-model strategy would collapse under its own complexity.”

Cost questions

There are still questions, a big one being pricing: Customers need to know how they will be charged when multiple models are in play. ServiceNow has not yet clarified this.

However, said Gogia, “The strength of ServiceNow’s approach lies in its understanding that trust, explainability, and choice are not optional add-ons. They are core to enterprise adoption. If the company continues to execute with the same clarity it has shown in this launch phase, the underlying architecture will gain traction quickly.”

He added, “Buyers are watching how fast ServiceNow can turn this vision into repeatable, low friction use cases. If Build Agent works as promised, if Control Tower scales with demand, if costs remain predictable, then this strategy has every chance of becoming the standard others are measured against.”