Challenges HR Leaders Face in Adopting Enterprise AI

Talent data fragmentation has emerged as a foundational constraint on enterprise HR’s ability to make strategic workforce decisions, persisting because organizations have systematically underinvested in unified data architecture.

Despite organizations operating an average of 897 applications, only 34% provide an integrated user experience across their channels, per the 2025 MuleSoft Connectivity Benchmark Report.

This distortion cascades as 92% of corporate learning programs cannot connect their costs to measurable results, yet organizations spend an average of $1,283 per employee annually on training, according to a 2024 study by the Association for Talent Development.

In a recent conversation on the ‘AI in Business’ podcast, Emerj Editorial Director Matthew DeMello spoke with Raúl Monroig, People Organization Vice President for the Intercon Region at Bristol Myers Squibb.

Monroig emphasized AI’s role in unifying fragmented HR data into actionable talent intelligence — and the need for HR leaders to narrow their skill‑building focus to the capabilities that truly drive business value.

In the following analysis of their conversation, we examine two key insights for HR leaders:

• Unifying fragmented talent data ecosystems: Consolidating dispersed employee information across HR technology stacks into trustworthy, verified data models that eliminate reliance on self-assessed competencies and enable precise talent mobility and development decisions.
• Disciplined skill focus and scientific outcome measurement: Replacing programs that develop 15-20 skills simultaneously with focused investment in just two to five critical capabilities, measured by whether they actually improve business performance and ROI.

Listen to the full episode below:

Guest: Raúl Monroig, People Organization Vice President for the Intercon Region at Bristol Myers Squibb

Expertise: Human Resources Leadership, Talent Acquisition, Change Management, and Global Workforce Development.

Brief Recognition: Raúl Monroig Ruiz is a distinguished HR leader at Bristol Myers Squibb, driving talent strategy and organizational transformation across diverse regions, including China, Southeast Asia, the Middle East, Africa, and Latin America. He champions the evolution of HR through change management and positive reinforcement learning, significantly enhancing employee engagement and operational excellence within global pharmaceutical manufacturing and distributor networks.

Unifying Fragmented Talent Data Ecosystems

Enterprise HR teams face a foundational problem that most AI tools don’t address: the data used to make talent decisions isn’t reliable enough to support them. The issue isn’t data scarcity; it’s that the information HR teams rely on is fragmented and rife with partial fiction.

Monroig distills this predicament with precision: “We work with only part of the data that we should have available, but we don’t.”  In other words, HR teams operate across sprawling technology stacks — Workday, dashboards, Eightfold, and dozens of specialized tools — but these systems exist in isolation.

“The data is scattered all around. Most of us are not working with a system that allows us to put it all together and work with it with a common view.And I think that’s fundamental for how we start working on developing individuals and developing teams.”

– Raúl Monroig, People Organization Vice President for the Intercon Region at Bristol Myers Squibb

Raul notes that the quality problem runs deeper than disconnected systems. When HR teams attempt to map skills, build pipelines, or design development programs, they begin from a compromised foundation: self‑assessment. Employees who rate their own skills face the same incentives that shape public professional profiles: a tendency to overstate strengths and minimize gaps.

As Monroig notes, people generally rate themselves as more capable than their peers, making self‑reported competency data structurally unreliable.

Any skill‑building program built on self‑reported data is effectively built on sand. The flaw is not only methodological; it reflects a deeper structural issue in how enterprises approach talent intelligence.

The trust deficit becomes acute when organizations attempt to implement AI‑driven mobility or skill‑matching workflows. These systems inherit the weaknesses of their inputs. Fragmented data scattered across platforms is not a technical inconvenience; rather, it is a strategic failure to invest in consolidated insight.

What makes the problem solvable is that alternative data sources already exist in:

• Performance metrics
• Peer feedback
• Project outcomes, and
• Behavioral signals

These sources contain ground truth about what people can actually do. HR platforms continue to treat self‑assessed skills and a siloed system‑of‑record data as sufficient.

Monroig argues that vendors should be selected and configured for the specific task of consolidating fragmented data, rather than relying on generic, off‑the‑shelf AI solutions. Machine learning systems can ingest disparate data sources and create unified employee profiles that surface patterns humans cannot see across siloed systems. Instead of waiting for perfect integration, organizations can use AI as a consolidation layer, translating scattered signals into coherent talent intelligence.

For HR, Monroig notes, the ability to consolidate data in this way becomes a template for the broader enterprise: AI as a foundational engine that makes fragmented systems coherent and actionable.

​

Disciplined Skill Focus and Scientific Outcome Measurement

Monroig argues that HR leaders often fail not because of intent, but because of overload. Organizations frequently launch broad portfolios of skill initiatives, including, among others:

• AI literacy
• Leadership development
• People management
• Digital fluency, and
• Legacy compliance programs

Without a clear hierarchy of which capabilities actually drive business value. The result, Raul notes, is predictable: “We end up trying to develop skill‑building systems that, frankly, cost a fortune, and we don’t know if they even deliver.”

The deeper issue is that HR cannot measure impact when the underlying data is fragmented. Without a unified view of employee performance, progression, and outcomes, attribution becomes impossible.

If HR cannot correlate skill investments with revenue growth, retention, customer satisfaction, or team performance, it has no basis for prioritizing which capabilities matter.

Monroig’s view is that this measurement gap pushes HR toward comprehensiveness — building everything and optimizing nothing — because the data foundation cannot support disciplined choice.

He argues that when HR lacks a unified, trustworthy data architecture, three consequences follow:

• No attribution: HR cannot see which skills correlate with business outcomes.
• No prioritization: All skills appear equally important, so everything gets funded.
• No strategic focus: Skill portfolios expand, but impact remains unverified.

In Monroig’s framing, the root cause is not ambition but architecture: fragmented data makes focused skill strategy impossible.

The consequence is not just wasted spending. HR organizations often build large‑scale development systems without ever validating whether the skills they target influence business outcomes. Basic questions go untested:

• Does this skill change bottom‑line performance?
• Does it drive revenue or retention?
• Does it improve customer outcomes?

In most enterprises, these questions remain unanswered, and Monroig notes that the pharmaceutical context makes this blind spot even more acute.

“We should be very focused on what the two, three, maybe five skills that are going to build our business in the coming years. [We should] be relentless and very disciplined in building those skills at the right level with the right people… We want to build leadership capabilities, and we want to build people management capabilities,” he argues.

– Raúl Monroig, People Organization Vice President for the Intercon Region at Bristol Myers Squibb

Monroig also points to a deeper uncertainty: HR has not historically operated from a validated scientific understanding of skill acquisition or behavioral change. Reward systems and development programs have been deployed for decades without rigorous testing.

Looking ahead, Monroig identifies three capabilities as essential for organizations navigating AI adoption:

• Curiosity
• Agility, and
• A customer‑service mentality.

Curiosity, in his framing, is the willingness to experiment even without guaranteed output. He describes an employee in Latin America who, without formal AI training, built a ChatGPT agent that reduced brand‑launch preparation from three weeks to 30 minutes.

That willingness to test, iterate, and adapt reflects agility — the ability to adopt new tools as they emerge. And the customer‑service mentality ensures that any AI application remains anchored in business needs, not novelty.