How Workday’s Sana Self‑Service Agent Works Inside Microsoft 365 Copilot

This architecture preserves a clear division of roles:

• Microsoft 365 Copilot: the user interface where employees interact with the agent.
• Workday Sana agent: the execution layer that accesses Workday data and runs workflows.
• Workday platform: the system of record that enforces permissions, approvals, and policies.

Because actions ultimately run inside Workday, organizations keep their existing governance and auditability while still giving employees a conversational experience.

HR Tasks Employees Can Complete

For employees, the agent functions as a self‑service HR assistant embedded in their daily tools. Common tasks include:

• Checking vacation or time‑off balances
• Viewing payslips or payroll details
• Asking general HR policy questions

These tasks can be completed without leaving Microsoft 365, allowing workers to get answers or take action directly from the Copilot chat interface.

Manager Approvals and Actions

Managers can also perform routine supervisory tasks through Copilot. One example highlighted in launch coverage is the ability to review and approve employee timesheets.

Even though the request originates in Copilot, the approval workflow itself still runs in Workday. That means approval routing, audit trails, and policy enforcement remain unchanged.

Finance Tasks the Agent Can Handle

The agent also extends to finance‑related self‑service workflows. Examples include:

• Managing or submitting expense requests
• Checking finance‑related information tied to employee accounts

Because the system connects to Workday’s financial workflows, these interactions remain tied to the same business rules used in traditional Workday interfaces.

How Workday Keeps Governance and Security Intact

A key design goal of the integration is to avoid bypassing enterprise controls.

Instead of exposing raw Workday data directly to Copilot, the system keeps Workday as the authoritative execution environment. Copilot simply triggers agent actions and receives results.

This means:

• Role‑based permissions from Workday still determine what users can see or do.
• Business processes and approvals continue to run in Workday workflows.
• Policies and compliance controls remain enforced by the Workday platform.

Microsoft’s integration guidance also requires configuration of Workday security components such as authentication policies, integration users, and security groups to scope access correctly.

Why Deployment Is Considered Low‑Friction

Organizations do not have to build a custom AI assistant to connect Copilot and Workday.

Instead, Microsoft provides an Employee Self‑Service agent framework with a Workday extension pack. IT teams primarily configure the connection rather than developing the integration themselves.

Typical deployment steps include:

• Setting up Workday single sign‑on and integration users
• Configuring security groups and authentication policies
• Installing the Workday extension package in Copilot Studio

These steps still require Workday administrators and security reviews, but the predefined architecture reduces the amount of custom development required.

How This Fits the Broader Workday–Microsoft AI Strategy

The Sana Self‑Service Agent is part of a larger partnership between Workday and Microsoft to support interoperable enterprise AI agents.

The companies have been building infrastructure so agents created in Microsoft tools such as Azure AI Foundry or Copilot Studio can be registered and governed through Workday’s Agent System of Record (ASOR).

In that model:

• Microsoft provides the AI interfaces and productivity surfaces.
• Workday provides the governed business context and operational systems for HR and finance.

The Sana‑in‑Copilot release demonstrates how those pieces fit together in practice: employees interact with AI through familiar tools, while enterprise systems of record still control data, permissions, and business processes.

The Bigger Trend: AI Agents in the “Flow of Work”

The integration reflects a broader shift in enterprise software design. Instead of forcing employees to switch between multiple systems, AI agents increasingly act as a unified conversational layer across enterprise applications.

In this model, productivity tools like Microsoft 365 become the interaction layer, while systems like Workday remain the authoritative platforms that execute business logic and enforce governance. The Sana Self‑Service Agent illustrates how companies are trying to combine both worlds: conversational AI convenience without sacrificing enterprise control.