Using AI to unlock human potential at Liberty Mutual

Liberty Mutual’s approach to gen AI reflects a deeper understanding of enterprise transformation in that the hardest challenges aren’t technical but human. So global CIO Monica Caldas focuses on the central priorities of enterprise leaders: establishing effective AI governance, moving promising pilots into scalable solutions, and empowering employees to confidently adopt new ways of working.

Such a strategy requires balancing seemingly contradictory goals: maintaining rigorous integrity while encouraging experimentation, building platforms without proliferating tools, and driving adoption without overwhelming the organization — and all of this without a proven playbook as gen AI reaches an inflection point. The result so far is a transformation that treats AI as a fundamental opportunity to rewire how 40,000 employees work and better serve customers.

How is Liberty Mutual using AI to improve your employee experience?

We’ve been using AI to understand and predict for years, so when gen AI emerged in late 2022, we were already thinking about how it could change the way we work. Our early AI focus was on predictive tools that improved underwriting and claims forecasting, strengthening our core operations. Those investments built the discipline and foundation we needed to adopt gen AI thoughtfully when it emerged. Our first lens of it was protecting integrity of the data and decisions, and how to govern in an environment when the technology’s promise is so expansive yet unproven. Then we adopted the philosophy of “You can’t learn to swim by reading about water,” and provided employees with training and safe spaces to test, learn, and develop intuition.

Our plan was to first understand this new capability. We launched LibertyGPT, our secure, internal gen AI platform to ensure we were protecting data with the right controls. In parallel, we developed an employee training program, which functioned like a driver’s license. Once you have it, you can test and learn.

In parallel, we established an experimentation framework we call our Digital Progression Framework that enables employees to learn, experiment, and innovate responsibly while understanding how AI can support their everyday work and decision-making in a meaningful way.

We also focused on leadership training. When I joined Liberty Mutual, we started Executech, an executive training program focused on improving digital and data literacy. The program covered technology fundamentals, emerging trends, and how technology can drive business value. So shifting our focus from digital to AI was a natural progression, and these efforts are supported by a growing learning ecosystem, including a gen AI hub for hands-on guidance, an AI@Liberty community for peer collaboration, a responsible AI steering committee, and a change champion network that helps teams adopt and integrate new tools.

Lastly, our broader ambition is to embed AI into the way we work every day. We’re not just improving speed and quality, we’re giving people the confidence and capacity to focus on what matters most: solving more audacious challenges and delivering meaningful outcomes for customers. Ultimately, we’re empowering employees to see AI as a core enabler, to imagine what’s possible, and to apply it in thoughtful, innovative ways.

How are you approaching the technical aspect of the transformation?

We began by defining a destination that’s aligned to our business aspiration of rewiring the enterprise for AI. From a technology perspective, that means enabling data to move securely and seamlessly across the entire stack in near real time while creating a platform that allows our teams to innovate and deploy new capabilities at speed.

Achieving this requires modernization on three interconnected fronts. First, we’re transforming our core systems so they’re modular, interoperable, and built for continuous evolution. Second, we’re redesigning our data pipelines and ingestion layers to ensure that high-quality, well-governed data flows reliably across the enterprise. Third, we’re strengthening identity, access, and governance frameworks to enable broader use of advanced AI capabilities, while maintaining rigorous controls, security, and regulatory compliance.

Rewiring for AI isn’t simply a technology upgrade. It’s a multidimensional transformation that aligns architecture, data, security, and governance to responsibly realize the full power of AI across the organization. Ultimately, this isn’t just a technical shift, rather it’s technology acting as a force multiplier for our business. So it’s a technological transformation, but it’s also an operational and cultural transformation — one interconnected body of work.

How do you avoid pilot purgatory?

My peers and I are aligned around a simple rule: we must be outcome oriented and aligned to our business aspirations as the first guiding principle. We also agree that greater value comes from disciplined platform-orientation, not tool-proliferation. So we won’t run multiple pilots with different AI companies just to get a headline or check a box. Rather, we take a thoughtful, value-driven approach that’s yielded 50 use cases in production and scaling. The discipline of use case selection, input and output criteria, sponsorship, and governance is how we avoid too many pilots.

Also on tool proliferation, we’ve listed over 30 AI capabilities that we’ll build or buy. If we buy, we’re deliberate that the tool solves a real need. Because there’s no AI playbook yet, we exchange ideas with a network of multi-industry CIOs and their technical teams. We also benchmark with consulting companies to make sure we don’t miss anything. As new tools and capabilities come to market, we know how to respond and capture value.

How do you tell the story of AI to your stakeholders?

I talk about AI unlocking human potential, which is a very personal transformation. When we all initially learned our jobs, we developed a set of skills, applied them, and developed behaviors and expectations. And here gen AI is introducing a completely different dynamic.

This is a vulnerable moment for all of us, where we reflect, learn, and evolve. So I don’t talk about the technology inflection point as one dimension. Rather, the conversation is about all the dimensions of the shift. For example, operationally, we need to change the way we work and think about tasks we do today that can be offloaded into a well architected and governed gen AI solution. For example, we had a manual help desk process that was time consuming and slow to respond to internal employee questions. So our helpdesk team reimagined the workflow to leverage gen AI to take on 80% of the process. This eliminated manual steps and improved our ability to respond to employee questions faster and with greater accuracy. We were then able to redeploy the technology team that was doing the manual work toward the backlog of higher complexity problems.

When you’re a CIO talking about AI, you need to show possibilities relevant to your different audiences. For me, the best strategy is to talk about realizing human potential with a few examples. It removes the abstract nature of the possibilities and helps provide real examples. We can also talk about what works and what’s still evolving. Despite the progress, there’s much that’s still left to be fully figured out.

With such change upon us, how do you advise up and coming tech leaders on how to manage their careers?

The technology leader of tomorrow will have domain expertise in their industry and will always have to be imagining the art of the possible.

Ten years from now, CIOs will operate in an environment defined by agentic and collaborative intelligence at scale. The pace of execution will be extraordinary, and the operational backbone of technology will be deeply instrumented, observable, and increasingly autonomous. Many of today’s tactical and repetitive responsibilities will be orchestrated by intelligent agents, enabling faster and more precise decision-making across the enterprise.

But this doesn’t diminish the CIO role, it elevates it. As automation absorbs operational friction, the CIO gains capacity to focus on higher-order impact: deepening industry expertise, shaping business strategy, and partnering with domain leaders to reimagine products, services, and experiences. The technology executive becomes not just a steward of systems, but a co-architect of competitive advantage.

The technical mandate remains critical in terms of modernizing platforms, enabling trusted data flows, and safeguarding security and governance at scale. Yet the differentiator will increasingly be leadership. The CIO of the future must maximize human potential, guide organizational change, and cultivate a culture that blends human judgment with machine intelligence.

In an AI-powered enterprise, the CIO’s influence broadens from running technology to shaping the future of the business itself.