AI Brings Real Value in Insurance

Both of your companies have experience with AI. Describe your strategies with the technology.

TERNES: One of the most interesting things about AI, and I sincerely still believe it’s underrated, is the clean slate it provides our industry to redesign our operations and processes. It gives us the ability to reimagine, unencumbered by legacy, what best-in-class can be. For us, one of the biggest aha moments was when we realized that we would be able to scale certain workflows, like document retrieval, policy servicing, and revenue reconciliation, so much that we can start to fundamentally rethink work allocation across the middle and back offices, and reduce the back-and-forth that occurs when accounting teams find anomalies at month-end.

CHAWLA: At a high level, our approach is rapid experimentation paired with disciplined, responsible deployment. When we started building Eventual, it quickly became clear that modern AI represents a fundamental shift where it’s risky for any vendor to claim they know the future a priori. The only way to build real conviction in our product strategy was to challenge our assumptions and aggressively test new capabilities.

A concrete example is our R&D work on insurance-specific AI agents. Early reasoning-based language models, while powerful, experienced noticeable declines in accuracy as contextual complexity increased. Rather than pushing the technology beyond its limits, we rearchitected our evaluation, validation, safety, and governance frameworks from the ground up, with insurance-specific requirements at the center. While the underlying models have improved significantly since then, those foundational safeguards remain core to how we responsibly build and deploy AI for our customers.

Where has AI unexpectedly proven capable of adding value?

TERNES: 100% policy data integrity. Having clean data in your AMS and other systems is, of course, the ideal. Updated information about downloads, NAIC codes, coverage limits, and other details helps operators best serve their insured. But it’s something that requires constant maintenance and effort. So, when agencies have to balance carrier production requirements and revenue milestones, they often deprioritize data cleanliness. This is one place where AI has been impactful. It makes maintaining basic policy details almost automatic.

CHAWLA: What has surprised us most is how much AI’s value creation compounds when it drives impact across multiple teams rather than optimizing a single workflow in isolation. When we enable account managers to easily access and update policy information, policy servicing becomes faster and more consistent, which directly reduces downstream issues that often surface later in accounting and finance.

Conversely, when accounting teams flag anomalies for review, AI-enabled exception resolution empowers account managers to significantly shorten investigation and resolution cycles. The recurring lesson for us has been that brokerage operations are deeply interconnected. AI delivers the strongest ROI when it acts as connective tissue across those processes, compounding its impact as it integrates more broadly throughout the organization.

What vendor qualities should brokers prioritize when choosing an AI vendor?

TERNES: AI-“driven” companies are popping up all over insurance, and the ones that will prevail and have the greatest impact for our agencies are incredibly focused, forthright in their capabilities, and eager for feedback. Having a proven track record with complex systems is absolutely an advantage when we look at security, scale, and guardrails that make us feel more comfortable. On top of that, we look for teams that can move fast. Technology is changing so quickly these days that anyone who does not move fast enough gets left behind.

What do you hope to see from AI in insurance in 2026?

CHAWLA: I think we are going to have neverbefore-seen capabilities emerge. Agentic systems are getting more reliable over longer time horizons. Domain-specific model capabilities will improve. At the same time, as we’ve already seen in industries like law and healthcare, the players who have laid the right foundations around integrations, data access, and ontology (i.e., customer-specific concepts distilled from enterprise data) will reap significant benefits and pull ahead. I hope this is the year that “this is too good to be true” will become “I can’t believe we lived without it.”