AI is already beginning to separate workers’ compensation carriers into two groups—those struggling to turn investment into payoff and those that are already starting to show substantive results.

Building on findings from our new Guidewire-PwC report, Reimagining Workers’ Compensation in the Age of AI, this three-part series explores the impact of AI on this sector. In Part One, I explained why the case for change is accelerating in workers’ compensation. For Part Two, I’m turning to more consequential questions: Where is AI actually starting to deliver value, what do carriers hope it can unlock next, and what do their concerns reveal about the barriers to scale?

The First Wins Aren’t Flashy, They’re Consequential

In workers’ compensation, the earliest gains from generative, agentic, and other forms of advanced AI haven’t been glitzy. They’ve emerged as foundational in areas such as loss control, medical management, claims handling, and return-to-work (RTW) coordination. Indeed, some of the Guidewire customers in our survey of workers’ compensation carriers include:

• A southwestern US-based regional carrier using onsite video monitoring and AI vision to identify hazards or unsafe activities—resulting in lower claims frequency from participating insureds

• A workers’ comp carrier harnessing AI to detect patterns of abuse in medical practices and remove providers overprescribing opioids, reducing medical costs by up to 15% and improving claims outcomes along the way

• A Canadian workers’ compensation board that implemented an AI-driven predictive model to flag high-risk No-Lost Time claims for early intervention, preventing a significant percentage from becoming costly Lost Time claims

• A single-state workers’ compensation carrier that improved its return-to-work performance with a solution that provides both recommended activities as well as comparisons against similar RTW outcomes to identify and act on variances

Taken one by one, these may just seem like siloed use cases. Taken together, they’re proof that AI is beginning to compress the distance between signal and action. But even so, significant challenges remain.

The Market’s Moving, But the Payoff’s Uneven

In a Guidewire survey conducted for our study, 100% of workers’ compensation carriers said they expect operational efficiency gains from AI, and 60% expect better return-to-work outcomes. Yet while 53% of workers’ compensation carriers said they’ve realized at least some of the benefits they anticipated from their AI investments, 30% said they had yet to do so.

The broader industry picture mirrors this dynamic. A February 2026 regulatory survey of 347 insurers across 25 European countries found that nearly two-thirds were already actively leveraging AI across at least one business function, yet most were still at the proof-of-concept stage. Meanwhile, a recent global claims survey of 160 P&C insurers from Bain & Company found that 78% have adopted AI within their claims organizations. But only 4% have scaled it meaningfully beyond document summarization, customer communication, and fraud detection.

That’s the hard truth I’m seeing across virtually every line, but especially in workers’ compensation. As the market gravitates toward AI, the ability to scale is separating organizations that are building real operating muscle from those still learning in pilot mode. And whatever the stage of their adoption, carriers in this sector also have a robust list of AI-related aspirations to pursue.

Not Just Efficiency: What Carriers Hope AI Unlocks Next

The AI ambitions surfaced in our study aren’t abstract. They’re a definitive roadmap to the areas where workers’ compensation carriers still burn too much time, judgment, and expense.

• Underwriters want faster, more reliable ways to interpret and summarize rules and regulations pertaining to workers’ compensation

• Case management leaders want claims routed to the right adjuster faster, with the insights and context they need to impact return-to-work times

• Policy servicing teams want more intelligent segmentation and more tailored service aligned with the size and needs of the organizations they write

• Product leaders want fresher insights into compelling ways to modernize the workers’ compensation line

For these and other leaders, AI adoption isn’t just about those anticipated efficiencies. It’s about improving the quality, consistency, and timing of decisions where cost, complexities, and risks to customer outcomes compound fastest.

What the Study Tells Us About the Real Barriers to Scale

None of this points to a market that’s struggling to imagine where AI could help. Carriers are remarkably clear about where they think AI can deliver value. What’s far less settled is whether their organizations are built to capture that value at scale.

As demonstrated in our report, the same underwriters pining to cut through regulatory complexity worry about being left behind by savvier rivals. Case management leaders that desire faster and smarter claims assignment worry that their infrastructure isn’t modern enough to take full advantage of AI. Policy teams longing for more intelligent segmentation fear governance implications and operating-model disruptions. Product leaders seeking fresh approaches to modernization dread unwelcome impacts on downstream systems. In my view, none of this reflects hesitation. It’s an operating diagnosis.

How AI Leaders in Workers’ Compensation Are Pulling Ahead

As someone advising insurers in this line around the world, it’s clear to me that the carriers pulling ahead with AI aren’t necessarily the ones with the most clarity of vision. They’re the ones building the enterprise conditions to make AI useful, durable, and trusted. Those with a modern, AI-powered infrastructure, clear governance, strong data discipline, and workflows able to absorb better decision support have the obvious advantage. In workers’ compensation, scale won’t come from running more pilots. It’ll come from making AI operational in the places where speed and accuracy optimize outcomes for companies and injured workers, bend cost trajectories, and expedite return-to-work results.

This story doesn’t end with proof of value, however. In Part Three, I’ll reveal why early wins with AI in workers’ compensation foreshadow five broader shifts that our report argues will dramatically reshape this line through 2030.

Read the Report, Join the Conversation

Be sure to read the full report, which I co-authored with Imran Ilyas, Partner, PwC; Oliver Winkenbach, GM, Data Applications, Guidewire; and Matthew Wolff, Former Partner, PwC. You can also listen to a recent episode of my InsurTalk podcast, where my collaborators and I discuss key findings from the report, below: