---
title: "Security workshops with Ask — Kiweely"
description: "Decide domain and business process policies in the workshop. Use Deployment Agent to inspect a non-Production tenant, then edit and activate the policy…"
url: https://kiweely.com/guides/security-design-workshops-deployment-agent
---

# Security workshops that follow Workday’s own access model

Workday security is functional areas, policies, and security groups. You cannot change a delivered domain’s contents. Ask can point at the policy that hides a field. The workshop ends when the right group has View or Modify and pending changes are activated, not when the chat sounds finished.

-   **Checked** against Workday on November 19
-   Applies to **Workday Extend**
-   6 min read
-   Published November 16
-   By Kiweely

A security workshop decides who can see and do what. Workday’s model for that decision is three parts: **functional areas**, **security policies**, and **security groups**. Deployment Agent with Ask can inspect how a non-Production tenant implements that model for a field or a task. As Workday documents it, Ask does not grant the access. You edit the policy and run **Activate Pending Security Policy Changes**.

Workday has not published a security-workshop product. The meeting is yours. The policy is Workday’s.

## Two policies, and the workshop has to name which

| Policy | What it decides | HCM course wording |
| --- | --- | --- |
| Domain security policy | Who can view or modify the tasks, reports, and data secured to that domain | Customers cannot change the delivered domain or the items secured to it. They put security groups on the policy. |
| Business process security policy | Who can initiate, approve, delegate, cancel, and otherwise act, for every definition of that business process type | One policy for the type. Definitions can route steps only to groups the policy allows. To do steps can be assigned to any active group, but a task on the to do still has to be allowed to that group. |

Putting a group on a business process step does not let them see a compensation field. Putting them on a domain does not let them approve the hire. The workshop writes two lines for every role.

Payroll’s setup topic is the same pattern, with payroll-specific group types: role-based (a payroll partner sees tasks and data for the pay groups they support), segment-based (decentralized administration of payroll business processes), and user-based (a payroll administrator unconstrained across pay groups). View is typical for reports. Modify includes the tasks and reports on the domain. Third parties can be brought in through service centers, with limited access to a subset of workers, and those representatives are not workers.

## Who may use the agent in the workshop

Only groups on the **Deployment Agent Users** domain, with View on report and task permissions, and on the Ask skill’s access list. Workday’s example group is Implementers. A group on the skill but not on the domain fails validation. Activating the agent is a separate switch in Agent Management Hub.

That access is not a license to show every workshop attendee the agent. The agent reads configuration in the active non-Production tenant, or a named connected tenant from Customer Central. It is not a Production assistant. Chat is deleted after 14 days. Do not put employee data questions into a thread you cannot retain; ask about the **policy**, not about a named worker’s pay.

There is no indicator that Ask is on. Before the workshop, ask for five calculation fields and open them. If you need a security answer, use the published shape:

> I am a Workday security administrator. In my tenant, users assigned to the \[role\] cannot view the \[field\] on worker profiles. Look in my tenant and tell me which domain security policy controls the field, which security groups currently have View, and how to update the policy.

Workday’s illustration uses Payroll Specialist and Bonus Plan Override, and describes a result on Worker Data: Compensation Explicit with the role missing from View. Treat that as the shape of an answer. Confirm the domain on your tenant before anyone edits. From Customer Central you can ask how activating the change would affect another implementation tenant. That question is an impact question, not a migration.

## A two-hour workshop that ends in an activated policy

**First half hour, vocabulary.** Use knowledge discovery if the room does not share terms. One new chat, lay language, bullets. Copy the terms into the decision record. Do not design in that thread.

**Next hour, one role, one field, one action.**

1.  Name the functional area. For HCM the course lists Staffing, Core Compensation, Organizations, and Personal Data. For payroll, open **Domain Security Policies for Functional Area** and pick the payroll area rather than guessing the domain name.
2.  Run the prompt above. Write down the domain it names.
3.  Open the domain. **View Domain**, then **Edit Security Policy Permissions**. See whether the group is on View or Modify.
4.  If the business process is also in scope, open **Business Process Policy > Edit** and see whether the group can perform the action. Remember you cannot usefully edit the definition’s secured step until the policy allows the group, and you must activate pending changes before the definition edit will see them.
5.  Decide in the room: View, Modify, or neither. Say the pay-group or organization constraint if the group is role-based or segmented. Unconstrained user-based groups see all instances of what the policy secures. That sentence should be said out loud before you grant one.
6.  A security administrator makes the edit in the Implementation or Sandbox tenant and activates it. A second person in that group opens the field or starts the action. That opening is the test. The chat is not.

**Last half hour, migration impact.** If this policy will move with Object Transporter, say how. Ordinary modified instances are overwritten on the target. Security policies **merge** unless you choose to overwrite. A security group’s business process security policy permissions migrate **additively** and do not remove groups already on the target. The workshop’s “revoke” decisions need a manual step on the target, written down now.

### Prompts for the three decisions people mix up

| Decision | Prompt | Confirm in |
| --- | --- | --- |
| Who can see a field | The published security prompt, with your role and field, plus “in my tenant.” | The domain policy’s View rows. |
| Who can act on a process | “In my tenant, which security groups can initiate and approve \[business process type\]? Which groups are on the step of the definition for \[organization\]?” | The business process security policy and the definition. |
| Who can use the agent | Do not ask the agent to list its own access as a substitute for setup. | Deployment Agent Users, the Ask access list, and **View Security for Agent Skill**. |

## Failure modes

-   Editing a copy of the policy in a chat and never activating. Nothing changes for the user until activation.
-   Granting Modify when the room asked for a report. Modify includes the tasks on the domain.
-   Using an unconstrained payroll administrator group to solve a “this pay group only” problem. Role-based or segmented groups exist for that constraint.
-   Believing a to do assignment grants the underlying task. It does not.
-   Migrating the policy and assuming Production groups disappeared. Additive migration will not do that.
-   Workshop notes that exist only in the agent. Fourteen days is the retention. Write the domain, the group, View or Modify, and the activation comment into the project record the same day.
-   Running the workshop in Production through Deployment Agent. You can configure the agent in Production so Sandbox refreshes keep it. You cannot use it there as a conversational assistant.

## Frequently asked questions

### Can Ask change the policy if we approve the suggestion?

Workday’s security example ends with you editing the policy and activating changes. Follow that. The concept topic calls the inspection read-only. Workday’s [Configuration Cloud](https://blog.workday.com/en-us/configuration-cloud.html), in opt-in early access for partners since 18 November 2026, lets Deployment Agent carry out changes once a plan is approved; that is not what the Ask example shows.

### We use implementers in Sandbox. Is that our security design?

No. Implementer groups are how you grant Deployment Agent and build access. The design is the groups your workers will have in Production. Test with a user in the worker’s group before you call the workshop done.

### How do service-center users fit?

Payroll’s topic says they are not workers, they have limited tenant access, and they support a subset of workers. Put that in the matrix as its own row. Do not model them as employees.

### What do we compare across tenants?

Tenant Compare can compare domain security policies, including whether a domain is suspended because its functional area is disabled, and subdomain differences on a matching parent. Use it after you migrate. Do not use a chat as the compare.

## Sources checked

Public Workday topics listed above, fetched 2026-11-16.

## Sources checked

-   [Overview of Workday HCM Core Concepts](https://doc.workday.com/workday-education/en-us/course-manuals/hcm-core-for-administrators/overview-of-workday-hcm-core-concepts.html)
-   [Example: Troubleshoot Security Policy and Access Configuration with the Ask Skill](https://doc.workday.com/admin-guide/en-us/workday-ai/agents/workday-built-agents/deployment-agent-in-implementation-environments-an/example--troubleshoot-security-policy-and-access-c.html)
-   [Setup Considerations: Payroll Security](https://doc.workday.com/admin-guide/en-us/payroll/payroll-considerations/yrp1604695036934.html)
-   [Business Process Framework](https://doc.workday.com/workday-education/en-us/course-manuals/hcm-core-for-administrators/business-process-framework.html)
-   [Set Up Deployment Agent in Implementation Environments and Customer Central](https://doc.workday.com/admin-guide/en-us/workday-ai/agents/workday-built-agents/deployment-agent-in-implementation-environments-an/set-up-deployment-agent-on-implementation-environm.html)

## Related guides

[All guides](https://kiweely.com/guides)

-   [### The 2026 Workday AI map: Sana, ASOR, Workday-built, custom, partner, and external agents Workday’s public AI story has five layers: Sana orchestrates, ASOR governs, Workday-built agents ship in the product, partners and customers bring their own, and build tools such as Developer Agent create new ones. This guide places each layer with Workday’s own words.](https://kiweely.com/guides/workday-ai-ecosystem-map-2026)
-   [### What Workday’s Deployment Agent Can and Cannot Do in Your Tenant Workday’s ASOR-enabled Deployment Agent reads tenant configuration in Implementation, Sandbox, Preview and Customer Central. It doesn’t run in Production, and edits configuration only in partner early access. Workday publishes no price; the Rate Card meters 5 credits per Ask interaction.](https://kiweely.com/guides/workday-deployment-agent-tenant-inspection)

### More guides on Custom agents and Workday AI

[All 91 guides](https://kiweely.com/guides#custom-agents)

-   [Pick a Workday integration tool before you prompt an agent](https://kiweely.com/guides/ai-workday-integration-design-workshop)
-   [Is there an AI agent that can do Workday configuration, business processes and security changes in a tenant for me?](https://kiweely.com/guides/ai-agent-for-workday-configuration-business-process-security)
-   [Our Workday consulting firm wants an AI tool that does tenant work for our consultants: what are the options and costs?](https://kiweely.com/guides/ai-tools-for-workday-consulting-firms-options-costs)
-   [Discovery questions Workday’s Deployment Agent can answer](https://kiweely.com/guides/ai-workday-discovery-fit-gap-workshops)
-   [Use Deployment Agent beside Object Transporter, not instead of it](https://kiweely.com/guides/ai-assistant-workday-object-transporter)
-   [Planning Agent and Decision Intelligence](https://kiweely.com/guides/workday-planning-agent-decision-intelligence)
-   [Partner-built agents and the Workday marketplace](https://kiweely.com/guides/workday-partner-built-agents-marketplace)
-   [Financial Audit Agent and Revenue Contract Agent](https://kiweely.com/guides/workday-financial-audit-revenue-contract-agents)

### From the blog on Custom agents and Workday AI

[All posts](https://kiweely.com/blog)

-   [Using an AI Admin Agent to Configure Workday API Clients and Test Role Boundaries](https://kiweely.com/blog/ai-admin-agent-for-workday-connector-setup-and-role-testing)
-   [Copilot, Claude, ChatGPT and Gemini in one company: how each one reaches Workday](https://kiweely.com/blog/microsoft-copilot-vs-claude-chatgpt-gemini-for-workday)
-   [What Actually Happens Behind the Scenes When an AI Agent Builds, Validates, and Deploys a Workday Extend App](https://kiweely.com/blog/from-business-brief-to-deployed-workday-extend-app-walkthrough)

## Let Workday check the work.

Kiweely writes WQL and builds orchestrations in your own tenant, and Workday, not Kiweely, says whether they passed.

[Start for free](https://kiweely.com/download) [See how it works](https://kiweely.com/vine/how-it-works)

## Get the next guide in your inbox.

One email a month with new guides and one Extend tip. Nothing else.
