Prepare for the Future of AI Agents in HR

More Detail

AI Agents Shift to Multiagent Systems to Increase Reliability

• Phase 1 — Single-platform MAS (current phase): Most HR implementations of AI agents create MAS by leveraging multiple agents within a single HR platform such as an HCM suite. This approach enables streamlined integration of agents working on the same workflows and data, eliminating the need for protocols. AI agents can easily share information about candidates or employees and collaborate on tasks such as scheduling interviews, creating career paths or suggesting learning plans within the same platform. Operating on a single platform simplifies maintenance and updates. Administrators can implement changes across the HR system without disrupting individual agent functions.
  
  The challenge with this approach is that AI agents may not be able to execute workflows that extend beyond a single HR platform. Single-platform MAS require the workflow to remain within one system. Until agent communication protocols mature, HR organizations will remain in Phase 1 or will need to build out interoperability between systems with orchestration bots to realize cross-platform collaboration.

• Phase 2 — Cross-platform MAS: This approach uses agent communication protocols to facilitate interoperability and collaboration among AI agents, enabling workflows that span multiple HR systems. For example, a global retail chain is launching a new employee benefits program. In this scenario, the benefits administration agent in one system collaborates with the human resources agent in another system to efficiently enroll eligible employees, ensuring accurate benefits allocation and adherence to local labor laws across various countries.
  
  Despite the increased interest and adoption of emerging standards, cross-platform MAS still require market maturity and consensus on communication protocols. They are not yet a viable option for the vast majority of HR organizations.

• Phase 3 — Internet of Agents (agentic web): While currently a vision and not yet a reality, this approach envisions a global network of interconnected agents, including HR agents, functioning as a multiagent system. This future state would be similar to the concept of an app store on a mobile device, where users can download apps built for plug-and-play installation and supported by defined communication protocols.
  
  In such an expansive ecosystem, HR AI agents can automatically discover and interact with other enterprise AI agents, forming dynamic collaborations to address complex workforce challenges. This interconnected network would leverage advanced discovery mechanisms, enabling agents to identify compatible agent partners across the enterprise (e.g., finance, travel and expense, IT and central scheduling) and establish communication channels autonomously. Cross-functional AI agents will become increasingly sophisticated, capable of negotiating roles, sharing resources and coordinating efforts across the company.

Figure 1: The Evolution of Multiagent Systems

Prepare for the Future of AI Agents

1. HR competencies to support AI agents — Build AI literacy and technical expertise within the HR organization to achieve success with AI agents. HR professionals need the necessary skills to experiment with, implement and manage AI agents effectively. Upskilling the HR team is also crucial for building robust AI governance and teaching staff members how to leverage monitoring tools that identify anomalies and deviations from standard HR workflows. Maintaining a human-in-the-loop approach ensures that critical thinking, empathy and oversight remain at the heart of HR practices. This approach not only supports the ethical and equitable use of AI but also keeps employees informed and engaged as AI agents continue to evolve within the organization.

2. Workflow mapping and redesign — AI agents will integrate into broader HR workflows, requiring careful redesign to deliver tangible HR value. HR leaders should begin by mapping critical HR workflows and employee journeys — such as recruitment, onboarding, performance management and offboarding — to pinpoint opportunities for various AI use cases, including agentic solutions. This approach ensures that AI agents enhance processes in targeted, meaningful ways.

3. Experimentation — Experiment with agentic use cases in HR to build familiarity and proficiency with emerging design patterns in this domain. Pilot AI agents for tasks such as screening résumés, handling service center requests or managing routine employee actions like requesting paid time off or enrolling in learning courses. The skills and insights gained from these HR-focused experiments will prove valuable, regardless of whether the most ambitious future scenarios — such as an Internet of Agents — ultimately materialize.

4. Governance — HR organizations should proactively develop, communicate and implement comprehensive governance protocols before deploying AI agents to minimize legal and reputational risks. Organizations must carefully navigate a complex regulatory environment when using AI to inform decisions that affect employees or job applicants. Most jurisdictions require ongoing human oversight, and some even mandate direct human involvement in the decision-making process.

5. HR technology, integration and APIs — Assess the current HR tech stack and modernize where appropriate. Remediate gaps with legacy platforms and evaluate the consolidation of disparate platforms. Appraise the interfaces and integration capabilities of your core HR applications, focusing on how effectively HR data and functionality are exposed through APIs. This assessment enables you to establish robust agent interface management, supporting secure, policy-controlled and observable agent integration within your HR technology stack.

6. Retrieval-augmented generation (RAG) — RAG represents a critical capability to enhance, as HR AI agents will depend on context and memory to deliver accurate, personalized support. Strengthening RAG ensures that AI agents can access relevant information efficiently and respond to HR queries with greater precision and customization.