🧭 ServiceNow and Anthropic Announce Multi-Year Partnership — Claude Powers ServiceNow Build Agent

ServiceNow has announced a multi-year partnership with Anthropic that makes Claude the default AI model powering ServiceNow Build Agent — ServiceNow's AI-assisted workflow automation tool used by enterprise IT and HR departments to automate approval chains, incident routing, and service request fulfilment. TechCrunch reports the deal as one of Anthropic's largest enterprise integrations to date, with ServiceNow's platform deployed across more than 7,700 enterprise customers globally. The partnership deepens Claude's presence in the IT service management (ITSM) market alongside existing integrations in financial services and healthcare.

What changes for ServiceNow customers

• Claude as default in Build Agent — ServiceNow Build Agent, the natural language tool that enables non-technical employees to build and modify workflow automations, will use Claude as its default AI model; customers retain the option to switch to other models for specific workflows
• Multi-step workflow generation — the integration uses Claude's extended context window and multi-step reasoning to handle complex workflow specifications that span multiple ServiceNow modules (ITSM, HRSD, CSM) simultaneously
• Enterprise compliance posture — the partnership includes a data processing agreement aligned with ServiceNow's existing GDPR, SOC 2, and FedRAMP compliance posture; no customer data processed through the integration is used for model training
• Co-development roadmap — Anthropic and ServiceNow have announced a joint product roadmap, with planned expansions into ServiceNow's field service management and supply chain modules scheduled for later in 2026

Bill McDermott, ServiceNow CEO, is quoted in the announcement: "Claude's ability to understand and reason over complex, interconnected enterprise workflows is exactly what our customers need to move from manual processes to intelligent automation at scale."

🧭 Guide — Building Enterprise Workflow Automations with Claude Agents

Alongside the ServiceNow announcement, Anthropic has published a new developer guide on building enterprise workflow automation agents with Claude. The guide is targeted at platform engineers and enterprise architects who are deploying Claude as the reasoning layer for multi-step approval and routing workflows — the same use case at the heart of the ServiceNow integration.

Key guidance for enterprise workflow agents

• Human-in-the-loop checkpoints — the guide recommends inserting explicit human approval steps at decision points with financial, personnel, or compliance implications; Claude agents should route to human reviewers rather than auto-approve when uncertainty is above a threshold
• State persistence across steps — guidance on using Anthropic's context window efficiently for multi-step workflows: storing state summaries rather than full conversation history to maximise the number of workflow steps that fit in context
• Tool call validation — pattern for validating tool call results before proceeding, preventing a failed API call at step 3 of a 10-step workflow from causing the agent to proceed with stale or missing data
• Rollback design — architectural pattern for designing reversible workflow steps, so that an agent encountering an unexpected state mid-workflow can cleanly halt and report its stopping point rather than leaving the system in a partially-modified state