Workday’s Bold Future With Responsible Agentic AI

In this article we discuss:

• Focusing on Both Tech and People
• Leaning Into the Future, Respecting Our Core

As AI enters its next evolution—agents—Workday’s responsible AI (RAI) governance framework continues to guide the company’s decisions about product development. AI agents, which are  systems designed to translate user intent into coordinated action, open up a world of potential in the workplace, and with these new capabilities come new considerations about responsible AI. 

We believe true innovation means bringing real, positive value to our customers while avoiding harm, which is why Workday's commitment to ethical and responsible AI is now more crucial than ever. It all comes back to our foundational values—Integrity and Innovation. These same AI Ethics Principles serve as our North Star, guiding every decision we make to help ensure the AI we build is trustworthy. Responsible AI isn’t an add-on for us; it’s at the core of our strategy.

Focus on Both Tech and People

At Workday, our approach to responsible AI is holistic—what we call sociotechnical. This means we look at the full picture: not just the tech under the hood, but how real people use it in real workplaces. We design our guardrails around how AI interacts with teams, processes, and the entire organization. Agentic AI introduces a new dynamic, moving from assistance to autonomous action, and that's why we’re adapting our existing, flexible framework to scale with the new risks and capabilities.

Technical Considerations

Agentic AI ushers in a major technical shift. To govern it, we focus on three core areas—risks that are also central to emerging regulations like the EU AI Act:

• Autonomy: We are deliberate about which actions agents can take on their own. We prioritize human rights and safety, helping ensure consequential decisions, like hiring, always remain with a human and are never automated.
• Transparency: Agent actions must be visible and traceable. An audit trail is essential for customers and developers to reconstruct behavior, spot misaligned actions, and help ensure accountability and compliance.
• Security: Agents with access to enterprise tools can pose security risks. To limit harm and protect trust, agents operate on a "least privilege" model, inheriting only the human user’s existing permissions.

Social Considerations

Technical guardrails are only half the story. The rest has to do with humans. When organizations adopt agentic AI solutions, we have to rethink how this may impact people, including what their jobs look like. New risks emerge when humans and agents start collaborating. That's why the social side of our governance focuses on three key areas: 

• Accountability: It is important to be clear about who is responsible for what. We make sure humans stay in control of consequential decisions, but users must also understand their role in understanding and overseeing the technology. 
• Evaluations: We evaluate beyond just output quality. We test agent behavior: how agents interpret instructions, make decisions, and respond when things go wrong. This requires ongoing, scenario-based evaluations based on an agent’s level of autonomy.
• Change management: We believe in guiding our employees through the AI journey to help ensure a smooth transition and build trust.  We lead with human-centered values and clear safety guardrails so employees understand both the "why" and the "how" for a successful AI partnership.

Leaning Into the Future, Respecting Our Core

Workday is committed to building responsible and trustworthy AI agents, grounded in our founding values and ethical principles. This is possible because our strong governance framework was intentionally built for agility, allowing our core principles to naturally and confidently extend to the agentic space. As technology, policy, and regulations continue to evolve, we are on solid ground and do not have to fundamentally rethink our approach. By staying true to integrity and innovation, we are confident our principled approach to agentic AI will unlock significant potential and help transform the world of work for the better.