The Workday Agent Landscape
Going one level below the overview — what an agent is, what's GA, what's coming, and how to form a defensible point of view for every customer.
Welcome
Module 1 gave you the shape of the proposition. This module goes deeper: what a Workday agent actually is, how it's positioned inside a tenant, which agents are live right now, and what's on the near-term roadmap.
By the end, you should be able to walk a customer through the landscape confidently, point them at the GA agents that match their priorities, and flag the roadmap items worth watching.
Learning objectives
What is a Workday agent?
A Workday agent is a purpose-built AI capability that runs inside the customer's tenant and performs a specific job to be done — summarising, recommending, drafting, or taking small actions on behalf of a user. It is not a generic chatbot bolted on the side.
Capability
The job it does — e.g. recommend candidates, draft a business process, summarise a contract.
Data & context
The Workday objects and records it reads to do its job, respecting tenant security.
Guardrails
The governance layer — trust, security, human-in-the-loop, audit trail.
Every agent combines those three. If you can describe an agent's capability, the data it needs, and the guardrails around it, you can have a credible conversation with a customer about it.
Categories of agent
The portfolio isn't flat — it maps to the customer's operating model. Use these categories to orient conversations.
Most customers will start in one category — usually the one where they have the most pain, not the most data. The job of the Kainos AI Navigator is to widen the aperture from "one agent in one area" to a sequenced plan across categories.
GA agents today
These are live, generally available, and can be positioned as "something your customer can use now". Each card gives the plain-English job.
HCM
Finance
Planning
Platform (deployment & adoption)
This list is current to the roadmap we hold in the CoE AI. Workday updates the catalogue each release — always confirm GA status on the current Workday roadmap before committing to a customer.
Quick check — before you move on
Early access & near-term roadmap
Not every "coming soon" agent is equal. Use the status colour to set customer expectations honestly.
Early Access
Planned GA — H1 2026
Planned GA — H2 2026 and beyond
How to talk about the roadmap with customers
Lead with what's GA. Mention Early Access only if the customer has the appetite to be a design partner. For anything Planned GA, say "on the current Workday roadmap" — not "available next quarter". Workday dates move; trust is harder to rebuild than expectation is to reset.
Data access & guardrails
Every agent sits on top of the tenant's existing security and data model. Customers will ask four questions — know the answers.
What data does the agent see?
Only what the invoking user is already permitted to see. Agents inherit tenant security; they do not grant new visibility.
Where does the data go?
Processed inside Workday's trust model. Customers should check the Workday trust and privacy statements that apply to each agent.
Is there a human in the loop?
For agents that take action (draft, recommend, configure), the user approves before commit. For read-only insights, the user interprets.
Is it audited?
Agent activity is logged. Customers can review what an agent did and who approved it — same audit posture they expect from Workday generally.
When a customer's security or risk team pushes back on AI, walk through these four questions slowly. Most objections dissolve once it's clear agents inherit security rather than create new risk.
Sana Agent Builder
Sana Agent Builder is Workday's platform for building custom agents — the answer for needs that aren't covered by a GA or roadmap agent. It's central to the "build" leg of the Kainos AI Navigator.
When it's the right answer
- The need is unique to the customer's operating model.
- No GA or near-term agent covers it.
- The value is large enough to justify a custom build and its lifecycle.
When it isn't
- A GA agent already covers 70%+ of the need — use it and adapt around the gap.
- The customer wants AI because AI, not to solve a specific job to be done.
- The data or process underneath isn't clean enough for a custom agent to succeed.
Module 11 goes deeper on build patterns, reference architectures, and where Kainos delivers Sana-built agents as a managed capability.
Check your understanding
Three questions. Each explains why every answer is right or wrong — the reasoning matters more than the score.
Next steps
- All learners: complete Module 3 — Kainos AI Navigator: our approach in depth.
- Customer-facing: Module 4 — Positioning with customers.
- Architects: Module 5 — Flex credits & AI architecture, then Module 11 — When to build.
- Integrations consultants: Module 6 — Simplifying integrations with agents.
- Customer Success: Modules 9 and 10 — Change management for AI and Adoption, not activation.