Workday integrates Sana to turn its enterprise apps into agentic execution engines

Workday is rolling out agentic AI capabilities from its acquisition of Sana across its product range and beyond. With Sana for Workday, it aims to help its customers further automate its HR and finance workflows, while Sana Enterprise will extend those capabilities to non-Workday applications too.

Much of that automation was already possible, but not in such a conversational way, nor as visible to end users.

Workday acquired Sana for $1.1 billion in September 2025. It is now using the technology from that acquisition to create an agentic layer on top of its existing workflows, APIs, and data model, via a conversational AI interface that is backed by a series of agents, letting users trigger and complete tasks in natural language rather than navigating menus or building integrations.

What’s changing, said Dion Hinchcliffe, lead of the CIO practice at The Futurum Group, is that Workday is moving from merely embedding AI features into its enterprise applications to creating an agentic execution layer inside an enterprise’s system of record,

“Enterprises already had APIs, workflow automation, and integration tooling that could theoretically produce similar outcomes. What Sana appears to change is accessibility: It places a conversational orchestration layer directly on top of Workday’s HR and finance data, permissions model, and workflow engine so agents can safely (accurately) execute tasks without custom development,” Hinchcliffe said, alluding to existing Workday offerings such as Illuminate and Build, which the company has added over the last two years in its aggressive push to add agentic automation and related offerings to its portfolio.

HFS Research CEO Phil Fersht said the reduced need for custom development to deploy workflow automation will reduce the burden on CIOs and their development teams as they no longer have to wire capabilities together.

Battle for control of enterprise AI

However, analysts also see Sana’s integration pointing to a deeper challenge for CIOs, as other vendors aggressively push their own agentic offerings: Deciding whether control for AI agents will sit with the system of record or the system of engagement.

“Sana’s integration reinforces that split. Employees or users may ask Copilot or Gemini for help drafting something or planning a task, but when AI actually needs to update payroll data, approve an expense, or change a contract record, that action must happen inside the system of record,” said Hinchcliffe.

While Microsoft, Google, and AWS are offering broad agents and frameworks that can operate across productivity environments and multiple applications, vendors such as Workday, Salesforce, and ServiceNow are offering specialized operational agents embedded in systems that govern HR, finance, and enterprise workflows, Hinchcliffe said.

The key difference, he said, is that horizontal agents like the ones from Microsoft lack the deep domain expertise and context required to accurately run specialized enterprise processes.

“The result is a growing architectural split: Horizontal agents help employees navigate work, while vertical agents embedded in systems of record are the ones trusted to actually execute it,” he said.

For customers studying that split, availability may prove just as decisive as architecture.

Workday said that Sana for Workday and related components are available through its Flex Credits model, meaning the capabilities are bundled into existing subscriptions, allowing enterprises to begin deploying them without additional licensing negotiations.

Flex Credits, announced last year, is a subscription-based model for AI consumption within Workday’s offerings. Customers receive an annual allotment of credits with the option to purchase more if needed.