BLOG

Back to Blog

Laura Drabik explains what smarter use of AI looks like in commercial insurance operations

February 2, 2026 | Insurance data decisioning, Insurance process automation, Insurance Underwriting

Commercial lines underwriting is performing well by historical standards. But strong results can hide structural issues. Submission volumes keep rising. Data arrives fragmented and unstructured. Underwriters are stretched thin. 

In this episode revisit of Unstructured Unlocked, Laura Drabik of Guidewire joined Tom Wilde to discuss why underwriting performance increasingly depends on how work enters the enterprise, and why modernization must start with intake, not downstream systems.

Listen to the full podcast here. >>>

The challenge is unstructured submissions, not missing data

Laura Drabik described what she calls the ā€œtower of bobbleā€ problem. In commercial and specialty lines, underwriting involves many parties, brokers, insureds, and carriers, all sending information in different formats. Submissions arrive as emails, PDFs, spreadsheets, and attachments, often with critical details scattered across documents.

As Tom Wilde explained, most underwriting systems assume clean, structured inputs. Pricing engines, rating tools, and policy systems are only as effective as the data they receive. When information arrives fragmented and inconsistent, teams spend hours reconstructing submissions before any meaningful work can begin.

The real task, then, is converting unstructured submission material into complete, accurate, structured data that downstream systems can actually consume.

Efficiency matters more as the market shifts

In a hard market, carriers could hit growth targets without reviewing every submission.  Drabik noted that many carriers have been able to tolerate this inefficiency during an extended hard market. When demand was high, insurers could meet premium targets without reviewing every submission. That dynamic is changing.

As competition increases, speed and precision become decisive. Underwriters need to evaluate more submissions, not fewer, to write the same premium. But adding headcount is not realistic. Underwriting is a specialized skill, and experienced talent is aging out.

The opportunity is to remove the work that should never require underwriter judgment. Data entry. Document sorting. Chasing missing information. When submissions are complete, normalized, and enriched before they reach the underwriter, experienced teams can focus on selecting and pricing risk. That is how carriers expand capacity without sacrificing discipline.

Related content: How AI enhances precision, speed, and efficiency in insurance underwriting

Trust is the foundation of AI in underwriting

Generative AI brings powerful capabilities, especially for summarizing complex submissions and adding context from guidelines and policies. But underwriting decisions are deterministic. A value is either correct or it is not. That is why no single type of AI is enough on its own.

The conversation emphasized combining approaches. Drabik highlighted AI’s ability to summarize complex submissions and compare incoming information against underwriting guidelines. Wilde explained that different types of AI serve different purposes. Deterministic AI is well suited for extracting specific values like dates, limits, or loss amounts. Generative AI can add context and summarize information. Combining these approaches helps teams move faster without sacrificing control.

Crucially, every step must be traceable and auditable. Underwriters need to understand where information came from, how it was processed, and what was added along the way. Human oversight remains essential, especially in early adoption.

Start with outcomes, not hype

As new AI terms gain attention, Drabik and Wilde offered a clear recommendation. Carriers should start by defining operational goals and work backward to identify where automation fits.

Modernizing underwriting is not about replacing people. It is about fixing the front end of the workflow so skilled teams can work faster, with cleaner inputs, and far less friction.

The takeaway is simple. Better underwriting decisions start upstream. Fix how work enters the enterprise, and everything downstream works better.

Subscribe to the Unstructured Unlocked podcast to get the latest episodes on your favorite platforms, including: 

Apple Podcasts

Spotify

Amazon Music

Ask Indico

Ask Indico

We help carriers make faster, smarter decisions across underwriting and claims — ask me how.