Takeaways:
- AI being used to orchestrate decisions and update operations in real time
- Application of AI should be done carefully and deliberately – and documented
- Insurers' added uses for AI require more governance work
This year, expect insurers to move past AI pilots and experiments and into orchestration of operational functions, say analysts at Datos Insights.
"2026 is going to be the year of scale. The hype cycle is over. We're no longer discussing what AI might do, but we're measuring what it actually is doing," said Tim Baum, senior principal at the firm, speaking in a
Insurers' use of AI is evolving, moving from just data management or extracting insights from data, to orchestrating decisions based on collected data and keeping information current for decisionmaking, according to Baum.
"The content infrastructure surfaces the right unstructured data at the right moment, turning real physical world observations into machine readable context," he said. "When an underwriter updates the guideline, the AI model can immediately take that. We've seen carriers actually doing this, where they will update a guideline. It's stored within SharePoint. SharePoint automatically sucks it into the AI model, and now when it's making decisions, it's making decisions real-time based upon those changes. You don't have to worry that the underwriter didn't realize that change was made, and now they're not in compliance."
With AI orchestration, AI agents can make decisions autonomously, according to Baum. To successfully manage this AI decision-making, insurers must document the architecture for the decisions, and deploy AI agents strategically in repeatable, data-rich workflows, he added. This capability should be rolled out progressively and piecemeal, acquiring buy-in from insurers and their customers.
AI-based underwriting solutions are making it possible to handle varied lines of business in a unified workspace, with a variety of features and functions, according to Meredith Barnes-Cook, senior principal at Datos Insights. "We are way beyond the original workload workflow solutions," she said. "Now we have those that have a lot of built in AI automation, analytical agentic capabilities, along with those that are partnering with standalone solutions in those arenas."
Advances in intelligent document processing (IDP), an AI technique, have made it indispensable for insurers, according to Carey Geaglone, senior principal at Datos Insights. "IDP is achieving accuracy rates of 90% or more, and that's significant from where we were just 18 or 24 months ago," she said. "IDP really is a key component in the success of the underwriting workbench implementations, because that submission process is so manually intensive and document heavy."
With these advances in insurance companies' use of AI, more governance efforts will be required, according to John Romano, principal at auditing firm Baker Tilly, who spoke in the firm's
"As our companies are implementing AI, do we understand how it behaves? Have we put in the proper AI governance over what we have, and any changes and updates expected," Romano said in the webcast. "If there's drift, do we know how to look for that? Do we know how to look for documentation for overrides? Are people still involved in the decisioning itself?"
