Setup Considerations: Agent System of Record

Administrator Guide

You can use this topic to help make decisions when planning your configuration and use of Agent System of Record. It explains:

Why to set it up.

How it fits into the rest of Workday.

Downstream impacts and cross-product interactions.

Security requirements and business process configurations.

Questions and limitations to consider before implementation.

Refer to detailed task instructions for full configuration details.

What It Is

The Workday Agent System of Record (ASOR) is a centralized platform that enables you to manage your organization's digital agentic workforce by providing a single solution that you can use to:

Discover, add, register, configure, monitor and manage your diverse range of AI agents securely at scale across various systems.

Leverage a single control center for governance, security and auditing your agents, thereby reducing administrative overhead and mitigating security risks.

View data insights into agent observability, usage and performance, enabling you to analyze and demonstrate the value and impact of your agents.

You can manage these agent types through the

Agent Management Hub

, regardless of where they’re developed and run:

Workday-built agents (1P)

Partner-built agents

Self-built agents (3P)

In ASOR, the

Agent Management Hub

is the central location where you can register, configure, monitor, and manage your AI agents. An agent is a system that uses a Large Language Model (LLM) to reason and take action.

Every agent in ASOR has an agent definition. This definition provides a concrete manifestation of agent code, providing an administrator transparency of what an agent is and what it can do. Each agent definition in ASOR in Workday contains:

Agent Name:

The name of the agent.

Agent Description:

Brief description of the agent.

Skills:

The tasks or activities the agent can execute.

Execution Modes:

We support 2 execution modes for agent skills:

Ambient Mode: The agent performs tasks as itself and security is evaluated using its own permissions.

Delegate Mode: The agent performs tasks on behalf of a human and, in doing so, security is evaluated using the worker’s permissions.

Tools:

The resources the agent may use when executing the skill. For agents that interact with Workday, tools will be Workday APIs.

Resources:

Type of API that a tool uses. Examples: REST, SOAP, Graph, WQL.

Business Benefits

Enhanced Administrative Efficiency:

Provides a single control center for governance and auditing, ensuring agents are deployed securely at scale.

Operational Transparency:

Offers insights into agent activity, usage, and performance to help analyze the value and impact of your AI workforce.

System Stability:

Protects system performance - while ensuring that agents run securely with standardized authentication and auditing.

Agentic Interoperability:

Provides the ability for agents to interact across systems to get work done.

Use Cases Throughout the Agent Lifecycle

You can use ASOR and Agent Management Hub to manage your agents throughout the agent lifecycle, which consists of these 5 phases:

Define:

You can view the agent definition for an agent on its agent profile page in

Agent Management Hub

. For external agents, you provide a definition for your agent via API.

To start using an agent, you must first register it in

Agent Management Hub

to prepare it for configuration and activation. In order to register in Production Workday-built agents you must meet the following requirements:

uMSA

Workday product entitlements

Configure:

As you configure an agent, you can specify which skills it will use and who can access it.

Activate:

Once you register and configure your agent, you can then activate it - which enables users to interact with it based on the security permissions you configured.

Deactivate:

When you no longer need an agent, you can deactivate it. This hides the agent from users, but keeps it accessible in

Agent Management Hub

- where you can reactivate it in the future as needed.

Recommendations

Review the View Security for Agent Skill report when configuring agents:

Use this report to surface the domain and business process security policy configurations the Workday APIs and resources your agent may use. Upon runtime, Workday first evaluates the security groups that have access to the agent (Available To), then the security groups that have access to the tools (domain or BP policies for the APIs and resources).

Audit Trail Review:

Regularly examine the audit log for agent actions. The log clearly distinguishes between an ambient agent operating alone and a delegate agent acting on behalf of a specific user.

Requirements

To access ASOR, you don’t require additional SKUs.

To register and use agents in the Agent System of Record fully in your production tenants, your Workday account team needs to opt-in to the Universal Main Service Agreement (UMSA).

UMSA is required for non-production and production tenants for Workday-built agents. UMSA will not be required for non-production tenants for self-built agents.

Limitations

Agent Proxy and Delegation:

Workday agent interactions can’t be proxied or delegated.

Key Management:

While ASOR handles the creation and secure storage of the client credentials (private key) in the Credential Store, you must ensure your security environment supports the OAuth 2.0 token-exchange process that utilizes this key pair.

Usage Analytics in Sandbox Tenants:

For agent usage analytics on the overview page, the sandbox refresh will not wipe out usage activities from previous weeks.

ASOR Security

Domains	Considerations
Agent Compliance	Provides access to reports and data required for agent compliance.
Agent Management Hub	Provides access to the Agent Management Hub - the centralized hub for viewing and managing all agents.
Manage: Agents	Provides access to tasks for managing agents including configuration, activation and deactivation in ASOR.
Reports: Agent Reporting	Provides access to view agent analytics and report data sources and filters for agent level reporting.
Reports: AI Agent Security	Provides access to reports about agent security information. Example: the View AI Agent User Audit Trail report.
Setup: Agents	Provides access to setup tasks and activities for agents in ASOR.

Agent Security

As you configure your agents, you can specify who has access to interact with the agent by providing security groups on the

Available To

column for each delegate execution mode skill. Example: You can specify Employee as Self
to make a skill available to an employee who interacts with the agent. As long as a user is part of one of the security groups on one of the skills, they will have access to interact with the agent.

Agent interaction security permissions are separate from tool execution permissions. Consider evaluating alignment between the security groups in the

Available To

prompt and the security groups that have permissions for the

Tools

the agent may execute using the

View Security for Agent Skill

report. This report shows you detailed information about the Agent’s Tools APIs and the respective domain or business process security policies and permissions. The agent interaction security groups do not have to be a 1:1 match with the security groups on API policies. In order for the tool execution to succeed, a user must be a member of a security group on the agent skill and one that’s on the API security policies. Upon agent runtime, Workday evaluates both user access to the agent and the user access to the APIs the agent is executing as tools.

Example:

Agent	Available To Security Group	Domain Security Policy Security Group Configuration	Successful Tool Execution
Self-Service Agent	Employee as Self	Employee as Self	Yes
Self-Service Agent	Employee as Self	Managers	No

Ambient execution mode skills do not require interaction access because they take action without human user interaction, usually on a schedule or set trigger.

You can use the Set Up AI Administrator Security Permissions task to configure the security for ASOR, PCC and Sana all at once.

You can access the

All AI Agent Accounts

report secured to the Domains: Workday Accounts
domain in the System functional area to view all agent accounts in your tenant. On the report, you can click the

Workday Account

related actions to view audit trails.

Agent System User (ASU) accounts enable you to secure and manage your agents in Workday. Workday automatically generates an ASU for your agent after you complete the

Configure Agent

task. You can view your ASU accounts for your agents using the

All AI Agent Accounts

task (secured to the Workday Accounts domain in the System functional area). You can also view AI agent accounts using the

All Workday Accounts

task.

Agent developers can leverage ASUs to interact with Workday data through Agent Gateway.

Reporting

The

Overview

section of the hub provides access to these reports (secured to the Reports: Agent Reporting
domain) and tasks (secured to the Manage: Agents
domain):

Active Agents

report

Activate Agents

task

Configure Agent

task

Deactivate Agent

task

Total Agent Sessions Over Time

report (only available for Workday-built agents)

Total Agent Sessions

report (only available for Workday-built agents)

Total Unique Users

report (only available for Workday-built agents)

The

Agent Registry

section displays your registered agents, along with their status and when an administrator user in Workday last made updates to their configuration.

You can click the name of an agent to view the agent’s profile, the available skills for the agent, and these agent analytics reports (secured to the Reports: Agent Reporting
domain and only available for Workday-built agents):

Agent Sessions

Skill Interactions

Total Skill Interactions Over Time

Total Unique Users for the Agent

Total Unique Users for the Skills

The

Unregistered Workday Agents

section displays the

Find Unregistered Workday Agents

report (secured to the Agent Management Hub
domain), where you can discover the Workday-built agents you’re eligible to register.

You can access these reports for agent security, compliance, and auditing:

Agent Input and Output Request Details

(secured to the Agent Compliance
domain and only available for Workday-built agents)

All AI Agent Accounts

(secured to the Workday Accounts
domain)

Generate Agent Input and Output Report

(secured to the Agent Compliance
domain and only available for Workday-built agents)

Input and Output Report For Registered Agents

(secured to the Agent Compliance
domain and only available for Workday-built agents)

View AI Agent User Audit Trail

(secured to the Reports: AI Agent Security
domain)

View Security for Agent Skill

(secured to the Security Administration
domain and the Manage: Agents
domain)

For additional agent reporting, you can access these data sources:

All Agent Definitions

(secured to the Reports: Agent Reporting
domain) that provides fields for reporting on all registered and unregistered agent definitions.

All Agent Registrations

data source (secured to the Reports: Agent Reporting
domain) that you can use in custom reports to analyze agent registrations.

Agent Input and Output Details

data source (secured to the Agent Compliance
domain) that you can use to view your agent input and output interactions for the past 30 days. Workday can take up to 20 minutes to display the latest information.

You can also leverage these data sources (secured to the Reports: Agent Reporting
domain) that provide usage analytics data for the past 30 days:

Agent Sessions by Day

Agent Skill Interactions

Agent Skill Interactions by Day

Agent Unique Users by Skills

Connections and Touchpoints

Workday offers a Touchpoints Kit with resources to help you understand configuration relationships in your tenant. Learn more about the Workday Touchpoints Kit on Workday Community.