Supporting Mutual Insurers' Use of AI (Part 2): What to Fix First, What to Do Next
Vineet Bansal, chief information and technology officer at The Mutual Group, is steering practical AI adoption for mutual insurers. His message is direct: the models aren't the hard part. Data, governance, and legacy cores are.
Below are the key takeaways and an action plan for insurance leaders who want results without adding risk.
The real blockers aren't models - they're data, change, and cores
AI tech is proven. What slows mutuals is data quality, change management, and governance overhead tied to cybersecurity and privacy. Many mutuals don't have dedicated teams for policy, review, and ongoing controls.
Then there's the core. If your policy admin or claims systems aren't API-enabled, you can't embed AI into underwriting, claims, billing, and service workflows. AI must sit inside both technical and human workflows to matter.
Are mutuals embracing AI?
Not as much as they could - and that's OK. You're not late. Early entrants paid high costs and carried more risk. Mutuals also weigh member trust and capital constraints more carefully than bigger carriers.
Now is the time to move with intent. Run proofs of concept and let a few fail. You'll surface your gaps faster and make smarter bets over the next 12-24 months. Be strategic, thoughtful, and deliberate.
What to do next
Build your AI strategy on what makes mutuals unique: local presence, deep ties to policyholders, and a protection-first mindset. That advantage compounds when your core is modern, data is reliable, and governance is tight.
AI won't fix a 20-year-old policy admin system. You need APIs and event hooks to embed models where work happens. Consider partners for delivery and acceleration, but keep strategy, oversight, and final decision rights in-house. Have skin in the game.
Governance, transparency, and human judgment
Efficiency matters, but not at the expense of transparency and accountability. Insurers are trusted to protect people and restore lives. That trust is the bar.
Stand up clear AI and data governance. Keep human judgment in the loop for complex or sensitive decisions, and audit decisions that affect coverage, pricing, or claims outcomes.
A 90-day practical plan for mutual insurers
- Assess core and API readiness: Inventory current APIs, events, and data services. Map one or two priority workflows (e.g., FNOL intake, small commercial renewal) and identify where AI could be safely embedded.
- Data readiness review: Profile quality for key entities (policy, claim, customer). Establish lineage and access controls. Document PII handling and retention.
- Set governance fast: Create an AI oversight charter, owners, and a lightweight risk review. Align with industry guidance such as the NIST AI Risk Management Framework and the NAIC's AI principles and resources.
- Run 2-3 narrow PoCs: Pick low-risk, high-visibility use cases like email and document classification, claims triage suggestions, or underwriting prefill. Define metrics (accuracy, time saved, leakage avoided). Require human review.
- Partner with clarity: Use partners for build and integration. You own strategy, data boundaries, model acceptance criteria, and production sign-off. Lock SLAs, monitoring, and rollback plans.
- Change management: Train adjusters, underwriters, and service reps on how AI supports-not replaces-them. Publish quick-reference guides. Track adoption, error rates, and cycle time improvements.
Looking to 2030
Set a multi-year view for where AI will sit in underwriting, claims, distribution, service, and risk engineering. Keep iterating: modernize the core, expand APIs, refine governance, and grow the use-case portfolio. Small, steady releases beat big bets that never ship.
For insurance leaders who want to accelerate
If you need structured training on underwriting, claims, and policy AI use cases-with a focus on governance and workflow integration-see AI for Insurance.
Your membership also unlocks:
