Oracle doubles the number of AI agents in Fusion HCM

Oracle yesterday announced more than a dozen AI agents that will be available as part of its Fusion HCM application, complementing the twelve already announced earlier this year, along with many others across the Fusion application family. The new agents provide insights and automations across internal mobility and recruiting, career development and skills, core HR, and employee lifecycle and payroll. The emphasis is on freeing up time so that employees, managers and HR teams can focus on more valuable activities, says Yvette Cameron, SVP, Global HCM Product Strategy. She explains:

It's really focused on transforming the way that work gets done. They don't just answer questions, like that first generation of Q&A that generative AI agents were focused on. They're actually automating workflows. They're surfacing information and insights in real time, and they're acting on behalf of users to remove friction from everyday tasks.

Oracle is continuing its policy of embedding its AI agents into the familiar Fusion application environment with no additional charge for customers. This, along with the ability to instruct agents using everyday language, removes barriers to adoption, says Cameron:

Launching into AI doesn't mean you have to spend weeks training and preparing for it. We make sure it's an entirely intuitive, natural extension of the experiences you've already been having.

As the catalog of agents expands, the vendor is looking to AI to perform more extensive actions as customer confidence grows in the technology. She adds:

Our vision as we move forward is to have fully autonomous workflows — not just tasks, but end-to-end workflows — whether it is managing the talent review process, whether it is sourcing and hiring individuals into the organization, or going from looking at current positions and managing the full end-to-end process of succession management. The end-to-end workflows [are] the next horizon of transforming work with AI.

Building confidence

Those actions will typically be configured to require approval from a human before the agent goes ahead and completes a task. But it's also open to organizations and users to take out that human-in-the-loop step if they're confident in the agent's capabilities. She explains:

As we look through each one of these [examples], we're bringing insight to the individual users. Here are suggested successors. Do you agree? Yes. Then they add the successor to the succession plan. Here is learning that should be taken. Do you agree? Yes. Then add that. And again, that's human in the loop.

If you take that checkpoint out, it can take it automatically, so it's not just recommending, it's taking that action. So that the individual isn't having to navigate to another part of the application and enroll somebody in learning, or add that to the succession plan, or start that approval process. We are taking action with these agents.

It's entirely up to organizations how much control their people retain over agent actions. She goes on:

In our AI Agent Studio, that element of human-in-the-loop is literally a toggle, which at every step you can say, 'Yes, I want a human in the loop here, yes or no.' That gives our customers complete control. They can turn off that human-in-the-loop checkpoint as they build confidence and scale these capabilities, but it's always in their control.

While many of the agents are focused on very specific sets of tasks, such as providing guidance on company policies, or setting and tracking team goals, there are two that go further. These concierge agents, one for managers and one for employees, are able to call on other agents to complete tasks and workflows, rather than leaving it to the user to specifically invoke each action. Cameron explains:

Those concierge agents answer the question of, 'Why should I need to know what each and every one of these 100-plus agents does?' You shouldn't. We have done the work to deconstruct and understand how to apply AI to automate intelligently these tasks and workflows, and then under the concierge agent, with one single entry point, I can ask a question, and the agent will determine, 'Oh, is this a benefits question, where I need to apply the benefits advisor, or is this around compensation? Or do I need to invoke the leave of absence advisor?' Individual users shouldn't have to know or care about what's happening under the covers...

Today, our concierge agents cover many different topic areas and invoke a series of different agents across compensation, leave, benefits. For managers, it includes employee-related inquiries, employment data, and talent-related inquiries. There are additional areas still to be incorporated in the concierge agents in upcoming releases, but it's a powerful advancement in the ease of use, really taking full advantage of the many agents available in Oracle.

Skills development

As agents take over more routine tasks and free up time for higher level activities, it’s inevitable that employee roles will change and staff will have to hone their skills in those more demanding areas. Cameron says:

We're not just automating processes and improving productivity, we're enhancing the employee's experience. We're unlocking the the potential of people by freeing them up from having to focus on administrative tasks and freeing them up to focus on other areas. We're bringing new insights that perhaps they hadn't anticipated before. We are evolving our capabilities and what's possible with this new way of working with each release.

Here too, agents have a role to play in assessing skill levels, advising on career paths, and guiding employees to relevant learning and development opportunities. She observes:

As we step back and think about the skills that we need in this new era of agentic AI, I think some of the biggest skills to focus on, for leaders and managers especially, are things like empathy, collaboration, communication, critical thinking.

These are what companies often call the softer skills, but they're the more human-centric skills that are going to make working alongside this new digital workforce successful in the future. As I talk with CHROs this topic of, how do we grow those skills that I just mentioned in our leaders, and how do we develop that, that's an important factor too.

My take

This is a big expansion of Oracle’s AI agent offering in the HCM space, and no doubt will be replicated in other application areas over the next few weeks running towards its annual AI World conference. (The new name for the event formerly known as CloudWorld gives a clue to the overriding theme this year!)

The new concierge agents are of particular interest, and will likely be the preferred starting point for many managers and employees, rather than having to delve into the individual characteristics of each separate agent. While it’s important that IT and HR managers have confidence in the detailed processes and actions that each individual agent is responsible for, everyday users don’t need to know about that level of detail. As with the discussion of human-in-the-loop approvals, this is all about giving customers the time and space to build confidence as they adopt AI agent technology.