HR technology is expanding – but HR work isn't getting easier

The analyst reports are unanimous. AI will transform HR. Talent acquisition, performance management, workforce planning, employee engagement – every corner of the HR function is being reimagined through the lens of artificial intelligence.

And yet, most HR functions that have attempted to implement AI-driven capabilities report the same experience: the pilots look promising. The enterprise rollout stalls. The ROI does not materialize.

The technology is not the problem. The foundations are.

The AI promise and the HR reality

Executive expectations for AI in HR are high and rising. Boards and CEOs are expecting AI-driven productivity improvements in the range of 17–23% across business functions. HR is expected to be part of that story.

Learning and development is the number one investment priority for HR leaders in 2025, with 39% planning to increase budgets – in significant part because of the need to reskill workforces to work alongside AI. HR technology (including AI-enabled tools) is the second priority at 34%.

Ambition is clear. But ambition without infrastructure does not deliver results. It delivers failed pilots and wasted investment.

Why AI in HR underdelivers

• Executives expect AI to deliver 17–23% productivity gains across functions. In HR, this depends on data quality, data completeness, and system connectivity that most HR functions have not yet achieved.
  Source: Forrester Research, 2024
• HR teams spend up to 36% of working time searching for documents – meaning the majority of HR operational data exists in unstructured, disconnected document archives that AI systems cannot effectively process.
  Source: IDC, 2024
  

AI systems – regardless of sophistication – require structured, connected, accessible data to operate. In HR, the most operationally significant information about employees does not live in structured database fields. It lives in documents:

• Employment contracts define the legal terms of the employment relationship
• Onboarding records capture the employee's initial agreements and acknowledgements
• Performance documentation holds the narrative of an employee's development trajectory
• Compliance certificates prove that mandatory training and policy acknowledgements have occurred
• Termination records define the legal and financial terms of employment endings

If these documents are stored in shared drives, email inboxes, or physical archives – disconnected from the HRIS – AI systems have no way to access, process, or learn from them. The AI operates on partial information. The outputs are partial at best, misleading at worst.

The structural problem beneath the AI problem

The document infrastructure problem in HR is not new. What is new is its strategic consequence. For years, manual document management was an operational inconvenience – slow, costly and occasionally risky from a compliance perspective. But it was manageable.

AI changes that entirely. When AI is positioned as the mechanism through which HR will deliver measurable business value, the quality and connectivity of the underlying data and document infrastructure becomes a strategic constraint – not an operational inconvenience.

HR functions that attempt AI transformation without first establishing:

• Structured document management integrated with HRIS
• Automated document workflows that create consistent, machine-readable records
• Compliant document retention and deletion processes
• Connected employee file structures accessible to analytics and AI layers

...will find that AI pilots succeed in controlled environments and fail in production. Not because the AI is wrong. Because the infrastructure it depends on does not exist at scale.

The business consequence

The consequences for HR credibility are significant. When AI initiatives that HR championed – and that the business invested in – fail to deliver the promised productivity gains, the finger is usually pointed at HR leadership or the technology vendor. The root cause (the document and data infrastructure gap) remains unaddressed and unacknowledged.

Meanwhile, HR continues to receive the lowest proportion of revenue investment of any major support function (0.80%, compared to 3.14% for IT and 7.5% for marketing), making it harder to fund the remediation work that would actually enable AI to succeed.

What the most AI-ready HR functions are doing

The HR organizations that are successfully operationalizing AI share a common foundation: they have addressed the document infrastructure problem before – or alongside – the AI initiative, not after.

Practically, this means:

• Conducting a structured audit of where employment documents live and how they are managed across the employee lifecycle
• Prioritizing integration between document management and HRIS, so that documents are connected to the system records they relate to
• Implementing automated document workflows that create consistent, structured records from high-volume HR processes like onboarding, performance review, and offboarding
• Establishing compliance-grade retention and deletion policies that are enforced automatically – not manually

These aren't AI projects. They are infrastructure projects. But they are the projects that make AI possible. Without them, AI in HR remains a perpetual pilot.

The Gartner 2026 document management research note provides a strategic framing that directly mirrors the HR context: AI document management platforms are expected to lead to unplanned budget overruns in 80% of enterprise deployments by 2030 – driven precisely by the gap between AI ambition and document infrastructure readiness. In HR, where the document estate is particularly complex and compliance-sensitive, closing the document gap before the AI investment is the risk-management imperative. (Gartner, Evaluating AI for Document Management, January 2026)

Why AI in HR will keep stalling without better foundations

AI will transform HR. That much is likely true. But the transformation depends on something far less glamorous than the AI itself: the quality, structure and connectivity of the documents and data that HR has accumulated over decades of operation.

HR leaders who want to be serious about AI need to be serious about infrastructure first. Not because it's the exciting work – but because it's the necessary work. The organizations that get to AI-powered HR are the ones that built the foundations the AI requires.

This is one of six articles in the "The HR Gap" series, examining why enterprise content management is becoming a prerequisite for HR transformation. The full series is available on the Doxis Blog.