Meet your company’s new HR reps: AI agents

The tech industry has spent the past few years focused on AI as a productivity engine, rewriting code, optimizing search, and automating customer service at scale. Now a more delicate transformation is underway., with agentic AI is moving into human resources. A new wave of startups and enterprise platforms claims algorithms can screen candidates, predict attrition, and recommend career paths faster than managers. The pitch is simple. AI promises less administrative work and more consistent decision-making. As these systems take on more responsibility, they are beginning to redefine what the “human” in human resources means.

“Concerns are valid, because unlike other enterprise functions, HR directly affects people’s lives, careers, and identities, so the bar for trust and responsibility is much higher,” says Mahe Bayireddi, CEO of HR tech unicorn Phenom.

Several companies are building tools for AI-led workforce redesign, embedding intelligent agents into hiring, employee support, and internal mobility. And wrestling with how to do it without losing the “human” in human resources.

In this premium story, you’ll learn:

• How leaders at four major AI-powered HR platforms are enabling agents without forgetting the human in the loop
• Why the big opportunities in the tech are helping HR balance C-suite demands for speedy AI deployment
• The key risks still being hashed out with the move from automation to autonomy

“What we’re seeing right now is what I would describe as a phase shift,” Bayireddi says. “There’s a lot of fear around job loss, but that framing is incomplete. HR roles are not simply disappearing. They are being deconstructed and rebuilt.”

HR Is Moving From Process Automation to AI Execution

Phenom’s new platform offers a window into how this change might actually play out.

HR data is highly sensitive, raising concerns that biased or opaque algorithms could lead to discrimination claims or flawed hiring and termination decisions. Mohit Bhende, cofounder and CEO of the technical hiring platform Karat, says fragmented legacy systems make the challenge harder, since many organizations still rely on disconnected tools.

“We’re somewhere in between, as the gap between the vision and the reality is wider than most vendors will admit,” Bhende tells Fast Company. “AI is not good at appreciating context, organizational history, or the kind of implied knowledge that makes someone genuinely valuable to a team.”

Phenom’s new platform, WorkOps, reflects a broader evolution. The company has built an agentic architecture designed to orchestrate workflows, with a centralized engine governing agents in real time, enforcing policies, and escalating decisions when human oversight is required. In practice, HR begins to resemble an operating system.

Mahe says a structural tension is emerging inside enterprises. CEOs and CIOs are pushing to accelerate AI adoption in pursuit of efficiency and competitive advantage. Chief people officers and HR leaders are urging caution, aware they are accountable for the human impact.

“The reality is, people-driven systems often cannot move at the same speed as technology, so this tension is not only expected, it is inevitable,” he says. “In HR, especially in areas like talent acquisition and talent management, decisions were often based on experience, intuition, and limited data. With AI, those workflows are becoming more data-driven, and transparent, and can improve operational outcomes when implemented correctly.”

Adoption remains uneven. Some studies suggest global AI use in HR ranges from 21% to 45% of organizations, while deep integration sits between 12% and 31%. Roughly 62% of HR AI failures stem from poor data quality and lack of context.

Phenom says its approach targets that gap. The platform builds on enterprise-specific context and guardrails defined during deployment, drawing on models including Claude, OpenAI, and Gemini, alongside smaller fine-tuned systems. The goal is to better match AI to the complexity of enterprise and employee data.

“Agentic AI cannot handle everything end-to-end. It lacks true contextual understanding and common sense, so relying on it completely would create inconsistencies and risks in enterprise operations,” says Phenom COO Hari Bayireddy. “We try to understand the industry first, collect data from multiple sources, and structure it properly, creating a semantic layer that the AI system can understand. Without that foundation, generative or agentic AI cannot deliver meaningful results.”

The Rapidly Evolving Market for AI-Powered HR Platforms

Phenom is part of a broader move toward AI-native HR, but its platform offers a concrete example of how vendors think this transition will work in practice. Startups such as Eightfold AI, Beamery, and Gloat are focusing on skills intelligence and internal mobility. Enterprise platforms including Workday, SAP SuccessFactors, and Oracle Cloud HCM are embedding generative and agentic AI directly into HR workflows.

Salesforce recently launched Agentforce for HR Service, which integrates AI agents into a system that lets employees request time off or track HR cases through conversational interfaces. The platform draws on unified enterprise data, including policies and employee profiles, to deliver responses and execute actions in real time.

“With business and HR leaders reporting that 41% of their time is spent on ‘zero-value’ tasks (referring to a Deloitte study), the industry is no longer just primed for change – it’s hitting a breaking point,” Kishan Chetan, GM of Agentforce Service at Salesforce, writes in an email to Fast Company. “In fact, we often say that the ‘portal-to-ticket’ era is dead. Agents can help resolve routine queries autonomously, freeing people leaders to focus on what only humans can do: high-value culture building and strategic talent development.”

Chetan says the long-term vision centers on humans and agents working together. Not everyone is convinced the shift is heading in the right direction.

An Inevitable Yet Unsettled Future

Experts argue that the move from automation to autonomy introduces new risks, especially in a domain where decisions have direct human consequences. Regulation is one factor. New laws in the U.S. and Europe are beginning to govern how companies use AI in hiring, particularly around bias, transparency, and candidate rights. In 2025, both Workday and Amazon faced high-profile claims of AI-driven employment bias, intensifying scrutiny and political pressure for clearer rules.

But regulation is only part of the challenge.

“In HR, where decisions directly impact people, the biggest concerns are hidden bias and over-reliance on AI as decision-makers rather than signal generators,” says Dr. Helen Gu, founder and CEO of InsightFinder AI.

Gu notes that AI still struggles with less tangible factors like context, collaboration, and culture. “There is a real risk of overfitting models to what can be measured while ignoring what actually matters,” she says. “When systems influence hiring or workforce planning, you need continuous visibility into how those models behave and where they may be drifting.”

Others point to strategic risks.

“Organizations may end up using AI to execute broken strategies more efficiently,” says Hemant Kapadia, CFO at Anaplan. “Deploying agentic AI in that environment is not progress. It is just automating chaos at a speed no human can control.” He says companies focused only on automation and cost reduction risk creating systems that are difficult to understand and govern. “The real opportunity is to use AI as a decision intelligence layer that drives growth.”

Vendors building these systems push back on the idea that AI replaces human judgment. They say most AI outputs are probabilistic and still require interpretation, especially in high-stakes decisions. “Our systems are designed so that the final decision, whether it is hiring or internal promotion, remains with humans. That balance is critical,” says Mahe.

Chetan adds that agentic AI “is an amplifier of human judgment, not a substitute,” noting that modern HR demands already exceed what teams can handle manually. Expecting leaders to respond to every query and oversee every outcome risks burnout in a function that is already stretched thin.

If Phenom’s vision holds, enterprise leaders will face growing pressure to manage more complex workforces while adapting to rapid technological change. Integrating AI into HR will require reworking organizational systems and underlying data structures, alongside new roles focused on oversight.

“Entirely new job categories are already emerging, where someone is constantly observing how agents behave, identifying where systems break, and deciding when to intervene, whether that means inserting a human into the loop or recalibrating the orchestration itself. As organizations deploy agents more broadly, I expect the nature of HR roles to evolve, helping enterprises make better decisions with clear and traceable accountability,” says Mahe.