Liberty Mutual

Liberty Mutual, a $50 billion commercial insurance enterprise with 40,000 employees, faced mounting pressure from manual, high-volume internal helpdesk operations that struggled to scale. In an industry where response accuracy and speed directly affect underwriting decisions and claims outcomes, slow internal support processes created downstream friction. Beyond the helpdesk itself, leadership recognized a broader challenge: how to govern responsible AI adoption across a large, distributed workforce without falling into fragmented tool proliferation or leaving promising pilots stranded before they reached production. The status quo meant talented technologists spending time on repetitive, low-complexity support tasks rather than higher-value problems.

Liberty Mutual's response centered on LibertyGPT, a secure internal generative AI platform built with explicit data governance controls before any broad employee access was granted. Rather than deploying AI opportunistically, the team established a Digital Progression Framework to structure responsible experimentation — essentially a governance layer that allowed employees to test and iterate safely. The helpdesk workflow was reimagined end-to-end so that generative AI now handles 80% of the process, eliminating manual steps rather than merely augmenting them. In parallel, Liberty Mutual launched 'Executech,' an executive upskilling program, alongside an AI@Liberty peer community, a gen AI guidance hub, a responsible AI steering committee, and a change champion network — treating adoption infrastructure as equal in importance to the technology itself.

LibertyGPT's impact on the helpdesk was immediate and structural: generative AI now automates 80% of the end-to-end workflow, dramatically reducing manual processing time and improving response consistency for employee queries. The technology team previously assigned to routine helpdesk tasks was redeployed to higher-complexity, higher-value work — a meaningful capacity shift in a large enterprise. Across the broader transformation:

• 50 AI use cases are in active production and scaling
• Governance-first deployment prevented the data exposure risks common in early enterprise AI rollouts
• Leadership digital literacy improved through structured Executech training

The outcome reflects a disciplined platform strategy: by avoiding redundant pilots and staying outcome-oriented, Liberty Mutual moved from experimentation to enterprise-scale production.

• Governance before access: Building data controls and integrity frameworks into LibertyGPT prior to broad rollout prevented the compliance and data exposure risks that have derailed AI programs at other large insurers.
• Platform discipline over pilot count: Committing to a curated, value-driven approach — rather than running parallel pilots with multiple vendors — accelerated the path to 50 production use cases.
• Reimagine workflows, don't just automate them: The 80% automation figure came from rethinking the helpdesk process end-to-end, not layering AI on top of existing manual steps.
• Change management is the implementation: Training programs, change champion networks, and leadership upskilling drove adoption more than the technology stack itself.
• Redeployment is a measurable outcome: Freeing skilled staff from routine tasks should be treated as a concrete result, not an incidental benefit.