How Liberty Mutual Uses Data and AI To Deliver Real Value For Customers And Employees
Liberty Mutual didn't treat data and AI as side projects. Under Global CIO Monica Caldas, the company built a unified tech engine that runs IT operations, engineering, AI, and the data stack as one system. The mandate: ship value faster, raise quality, and keep solutions reusable across markets.
That clarity matters for insurers. It turns AI from isolated tools into measurable outcomes in underwriting, claims, and service-where the customer feels it and employees get time back.
Make Technology An Enterprise Engine
Caldas co-leads the global enterprise data and data science office alongside the COO. The team focuses on shared data fundamentals (domains, quality, data products) and the repeatable ways AI creates value. The result is a single operating model with local delivery, built for speed and resiliency.
This structure lets teams plug into common platforms while solving market-specific needs. Less reinvention, more throughput.
From Predictive Models To Practical GenAI
Liberty Mutual's AI path started with predictive models for underwriting completeness and claims forecasting. Those wins created the discipline to adopt generative AI with intent, not hype.
Enter LibertyGPT, a secure internal platform. Today, 74% of eligible employees have access to enterprise GenAI tools, supported by required training and leadership sponsorship. The payoff: higher efficiency, lower costs, and better customer experiences.
Build Literacy And A Shared Language
Through the Executech program, leaders learn technology fundamentals, emerging trends, and how to link them to business outcomes. This common language helps teams prioritize the few capabilities that truly differentiate and fund them accordingly.
It also accelerates co-creation with business experts-vital for underwriting and claims where context drives decisions.
Culture, Talent, And New Ways Of Working
For a 40,000-person company, adoption is a people challenge first. Liberty Mutual's Digital Progression Framework gives employees a safe path to learn, experiment, and apply AI responsibly.
Everyday AI encourages practical use cases-drafting communications, summarizing documents, analyzing data. A GenAI Hub, AI@Liberty community, Responsible AI Steering Committee, and a Change Champion Network provide hands-on guidance and momentum.
What Value Looks Like
Liberty Mutual measures results across three lenses: operational efficiency, impact, and adoption. Here's what that means in practice:
- Efficiency: AI-assisted development has increased engineering delivery velocity by about 10% on average, while moving time to higher-value problem solving.
- Impact: More than half of eligible employees actively use large language model tools. The company is saving about 13,500 hours each week and reinvesting that capacity in customer outcomes.
- Adoption: Usage patterns, engagement, and employee feedback ensure tools are applied where they matter and used responsibly.
AI is embedded in core workflows. Underwriters get better decision support. Claims teams use copilots for document summarization and workflow automation. Customer service benefits from faster responses with less friction.
In the investments business unit, the company modernized core systems, moved data to the cloud, rolled GenAI into daily work, and is advancing agentic AI. That clears technical debt and improves speed and security for digital experiences.
Govern AI With Intent
A Responsible AI Steering Committee brings together Technology, Legal, Privacy, Security, and HR. The group balances risk and opportunity, sets guardrails, and ensures governance keeps pace with delivery. Data pipelines, model testing, and internal processes were in place before GenAI-so scaling has been controlled and accountable.
If you're formalizing your approach, frameworks like the NIST AI Risk Management Framework can help shape policy and controls without slowing teams down. See NIST's guidance.
What Insurance Leaders Can Do This Quarter
- Treat data and AI as an operating model, not a set of tools. Centralize platforms; federate use cases to business lines.
- Build a shared language. Run executive learning sessions and link them directly to funding decisions and roadmaps.
- Get GenAI into employee workflows with guardrails. Provide training before access, curated use cases, and prompt libraries. Track adoption weekly.
- Measure value in three dimensions: efficiency (time saved, throughput), impact (customer and employee experience), and adoption (usage and feedback).
- Stand up a cross-functional Responsible AI committee. Align to recognized frameworks and document decisions end to end.
- Start with 20-40 focused use cases across underwriting, claims, and service. Scale winners; sunset what stalls.
Where This Is Heading
Liberty Mutual's view: technology is a force multiplier for business goals. AI is shifting systems from fixed logic to context-aware methods that adapt in near real time.
The company is making data "AI-ready" for large language models and agentic systems, then embedding them directly into workflows. The outcome is faster decisions, less manual effort, and more time for deep work that improves customer outcomes.
Upskilling For Insurance Teams
If you're building similar capabilities, structured learning shortens the path. Explore role-based programs and certifications to equip teams for data analysis and GenAI in daily work: Courses by Job and AI Certification for Data Analysis.
Your membership also unlocks:
