Workday announced a partnership with Anthropic on October 13, 2026, introducing a beta service called the Sana from Workday connector that plugs the HR and finance software company's agentic AI directly into Anthropic's Claude model. The announcement was made at the Workday Rising event in Las Vegas.

Agentic AI refers to systems that carry out tasks on a user's behalf rather than only generating text—here, applied to everyday HR and finance requests. The integration is designed to let employees complete routine tasks—such as checking their pay or requesting time off—by asking Claude rather than logging into Workday directly. Throughout these interactions, Workday's existing security framework, including roles, permissions, and audit controls, remains in place.

How the Claude Connector Works

Sana, which Workday recently rebranded as its agentic AI layer, serves as the central hub connecting Workday's functionality to external AI tools such as Claude. Every request still runs through Workday's permission structure: a manager sees only what a manager is permitted to see, and an individual employee can access their own payslip, but not that of the CFO.

That enforcement point carries weight because HR and finance systems hold some of the most sensitive data a company manages. Workday framed the connector as a safeguard against a broader risk: external AI agents that connect to business systems on their own terms could potentially bypass a platform's built-in infrastructure, and with it the safeguards companies rely on. Routing Claude requests through Sana is the company's mechanism for preventing that from happening.

The connector is available as a beta service and requires an active Workday subscription. At launch, it focuses on self-service HR and finance functions, and Workday said it plans to expand the tool over time to support complex, multi-step agent workflows.

The Numbers Behind Workday's AI Push

Workday acquired Sana, originally known as Sana Labs, for $1.1 billion in November 2025. The Sana from Workday product became available in March 2026, and seven months later that same technology is now being extended outward into a third-party model.

Company figures underscore the momentum. As of Q2 2026, more than 5,500 Workday customers were using AI agents, a 35% increase from the previous quarter. Workday's AI-driven SKUs accounted for nearly $600 million in annual recurring revenue, up more than 200% year over year.

Alongside the Claude connector, Workday is also launching Agent Harness, a tool designed to let companies customize AI agents using their own HR and finance data.

Why Workday Is Opening the Door to Claude

Rather than directly competing with general-purpose AI models, Workday is pursuing a strategy of embedding its functionality into those environments—including Google's Gemini and Microsoft's Copilot—so that it remains the secure execution layer for enterprise applications. The company aims to manage the complexities of security and compliance while maximizing usability for users. That strategy reflects a broader shift in enterprise software, where the question is less which assistant employees use and more which system enforces permissions when they do.

By keeping roles, permissions, and audit controls intact, Workday is betting that IT and compliance teams will be more comfortable letting employees use Claude for HR and finance tasks. Positioning Sana as the hub between Workday and third-party AI tools also makes the company the gatekeeper for its own data, even when the conversation is happening somewhere else.

For Anthropic, the partnership extends Claude's reach into a large base of enterprise workflows. Every connector that lets Claude complete real business tasks makes the model more useful inside companies, not just as a writing or coding assistant. The milestones to follow are the ones Workday has already set out: expansion from self-service functions to complex, multi-step agent workflows, and the rollout of Agent Harness for tailoring agents to company data.