Running AI Agents in Insurance | ITC Vegas 2026
From Deployment to Real Impact: Running AI Agents in Insurance | ITC Vegas 2026 Fireside Chat
Moving Insurance AI Beyond the Pilot
Insurers have spent the last two years experimenting with AI. The question now is how to run agents across claims, underwriting and servicing safely, at scale and with results the business can measure. At ITC Vegas 2026, Dan Dantus, General Manager of Agentic AI at Skan AI, sat down with Venki Arunachalam, former Senior Director of Business Innovation and Technology Strategy at Equitable, to talk through what it really takes.
Key insights
1. Agents need to understand how the business runs
An agent that only knows the documented process will struggle the moment a claim or application goes off script. Grounding agents in how insurance teams actually work, including the exceptions, gives them the context to execute complex workflows reliably.
2. Governance should be a prerequisite, not an afterthought
As agents move into regulated workflows, insurers need continuous visibility and controls, as well as a complete audit trail. Traceable context means every agent decision can be explained to risk committees and regulators.
3. Start with high-volume, rule-governed processes
The strongest early candidates for agents are high-volume, rule-governed tasks: claims intake, underwriting exceptions and prior authorization backlogs. Getting these right builds the trust needed to expand.
4. Own your AI end to end
Insurers want AI that runs on infrastructure they control and integrates with their existing architecture so the operational knowledge that makes them competitive stays in-house.
See Skan AI in action
Request a demo to see how insurers are running teams and AI agents together with full traceability.
Frequently asked questions
Leading carriers are deploying agents on high-volume, rule-governed work such as claims processing, underwriting exceptions and prior authorization, while keeping humans in control of complex decisions.
Insurance is heavily regulated. Traceable context means every agent action is grounded in observed process data and recorded in an audit trail, which supports compliance and builds trust with regulators.
Skan AI observes how insurance teams actually work across every application, then uses that context to identify automation opportunities, train agents and monitor their execution with governance built in.
Yes. The platform can run on-premises or in a private cloud, fully within the customer's environment and powered by NVIDIA AI Enterprise and NVIDIA NIM microservices. This deployment approach keeps an insurers’ data, models, and agents fully under their control.
