Key takeaways:
ServiceNow has picked Claude as the default model for its Build Agent and a preferred model across the ServiceNow AI Platform.
The ServiceNow Anthropic Claude partnership pushes AI directly into the 80+ billion workflows enterprises already run on ServiceNow every year.
ServiceNow is rolling Claude and Claude Code out to its own 29,000-strong workforce, with early tests showing up to a 95% cut in seller prep time.
The two companies are targeting a 50% reduction in customer time-to-implement and building industry solutions for healthcare and life sciences.
Claude Opus 4.5 is being positioned as the reasoning engine for regulated tasks like claims authorization and research analysis.
Enterprise AI has moved past the pilot stage. Big companies are no longer asking whether to use large language models; they are asking how to wire them into the workflows that actually run the business, without losing control of security, cost, or compliance. That is the backdrop for the ServiceNow Anthropic Claude deal announced this week, which makes Claude the default model behind ServiceNow's Build Agent and a preferred model across the wider ServiceNow AI Platform. According to Anthropic, the partnership also puts Claude and Claude Code in the hands of ServiceNow's own 29,000 employees. Here's what is in the announcement, what it changes for enterprise buyers, and where the honest caveats sit.
What ServiceNow actually announced?
ServiceNow named Claude as the default model powering ServiceNow Build Agent, its enterprise coding tool for building apps and agentic automations, and as a preferred model across the broader ServiceNow AI Platform. Build Agent lets both professional developers and citizen developers describe an app or workflow in natural language and have the agent reason, plan, and execute the build.
ServiceNow expects Build Agent usage to quadruple over the next 12 months. The company also flagged a joint goal with Anthropic: a 50% reduction in customer time-to-implement, shrinking the gap between the first sales conversation and a live, autonomously deployed workflow.
Why the Build Agent choice matters?
ServiceNow's platform is the connective tissue for IT support, HR, customer service, and security operations at a large share of the Fortune 500. Enterprises push more than 80 billion workflows through it every year, according to the Anthropic announcement. Putting Claude at the default layer of Build Agent means those workflows can now be authored, extended, and automated with a frontier reasoning model, inside ServiceNow's existing access controls, audit trails, and usage monitoring.
That governance layer is the real story. Most enterprise buyers are not blocked on model quality anymore. They are blocked on identity, data boundaries, logging, and change control. Embedding Claude behind ServiceNow's platform is designed to fold those requirements into the same workflow the model runs in, rather than bolting on a separate AI stack.
Citizen developers get a real on-ramp
One of the more interesting claims is that Build Agent will let non-engineers create applications that used to need significant developer support. If that holds up in practice, it changes the queue at every internal IT shop that has a two-year backlog of small requests. Whether it holds up depends on how tightly the platform constrains what a citizen-built agent can actually touch, which is exactly where ServiceNow's governance controls have to earn their keep.
Industry solutions: healthcare and life sciences first
ServiceNow and Anthropic are also co-building agentic applications for regulated industries, starting with healthcare and life sciences. Claude will support tasks related to research, analysis, and claims authorization, running on ServiceNow's governed platform. The pitch: claims authorization that today can take days could be compressed to hours, at a lower cost.
Anthropic points to Claude Opus 4.5 leading major medical and life sciences benchmarks as the reason it is the right model for these workloads. That is a strong claim, and it comes from Anthropic's own evaluations, so treat it as a vendor-stated benchmark rather than an independent verdict. If you are a health or pharma buyer, ask for the underlying eval methodology before you plan around those numbers. Coverage of adjacent model releases gives useful context on how quickly the underlying capabilities are shifting.
ServiceNow deploys Claude across its own workforce
Internally, ServiceNow is rolling Claude and Claude Code across its full workforce. Two specific use cases were called out:
Internal use case | Tool | Reported early result |
Sales meeting preparation | Claude-powered coaching assistant | Up to 95% reduction in prep time |
Engineering + internal tooling | Claude Code | Faster idea-to-implementation across teams |
The seller-prep tool connects Claude to real-time enterprise data and web search, so reps can synthesize prospect context, account history, and market signals in one place instead of stitching together five tabs. The 95% figure is from ServiceNow's internal testing, so it is directional rather than an independently audited benchmark.
Claude Code, meanwhile, is now in the hands of ServiceNow's engineers for writing, reviewing, and debugging code, plus automating the repetitive parts of internal development. That mirrors what we are seeing at other large engineering orgs adopting AI coding assistants: the wins show up first in code review turnaround, and small refactors, not in headline productivity numbers.
What buyers should actually take from this
A few practical points for anyone evaluating the ServiceNow Anthropic Claude combination for their own stack:
If you are already on ServiceNow, this is the easiest way to add agentic automation for adding agentic automation without spinning up a parallel AI platform.
The governance story matters more than the model choice. Ask how Build Agent enforces data boundaries, who sees prompts and outputs, and how usage is logged.
The 50% time-to-implement and 95% prep-time figures are targets and internal results, respectively. Treat them as ambition, not SLA.
Industry solutions for healthcare and claims are early. If you are in a regulated sector, expect to co-design the first deployment rather than buy something off the shelf.
For teams still building out the base AI strategy before layering on a platform play.
Related service: AI Development
Frequently Asked Questions
What does it mean that Claude is ServiceNow Build Agent's default model?
Build Agent is ServiceNow's enterprise-grade tool for building apps and agentic automations. Making Claude the default means new Build Agent projects call Claude out of the box for reasoning, planning, and code generation, without customers having to select or configure a model themselves.
Can ServiceNow customers still use other AI models?
Yes. Claude is described as the default for Build Agent and a preferred model across the ServiceNow AI Platform, not the exclusive one. Customers keep the ability to use other models where they fit, and to layer Claude in for tasks that need more advanced reasoning.
Which industries are ServiceNow and Anthropic targeting first?
Healthcare and life sciences are the announced starting point, with Claude supporting research analysis and claims authorization inside ServiceNow's governed platform. The two companies say they will take these industry solutions to market together.
Where can I read the official announcement?
Anthropic published the full partnership details on its news page. According to Anthropic, Claude is now available as the default model for Build Agent and as a preferred model across the ServiceNow AI Platform for tens of thousands of enterprise customers.
How does this compare to other enterprise AI partnerships?
The pattern, embedding a frontier model inside an existing enterprise platform rather than selling it as a standalone service, is becoming the dominant shape of enterprise AI in 2026. What is distinct here is the depth: default-model status inside a workflow layer that already processes 80 billion enterprise transactions a year.
What to do next
If your team runs on ServiceNow, the practical next step is a short internal review of which existing workflows would benefit from a reasoning model, and which of those are ready for governance controls to be tested against. Start with one high-volume, low-risk workflow, and measure honestly.