Embed AI into workflows to unlock real capacity: mea Platform

Insurers often struggle to move AI projects past the trial stage. Julie McKenna explains how focusing on workflow integration and routine administration creates practical capacity.

Key points:
Embed AI into live workflows
Calibrate automation to risk
Avoid long-term tech lock-in

Commercial underwriters spend too much time on basic clerical tasks, pulling figures from complex submission documents and typing details into separate systems before pricing a risk. This administrative burden delays quotes, strains broker relationships and cuts into the time technical specialists need for risk assessment. And this manual state persists despite decades of “core system modernisations” and other innovations purported to alleviate this burden.  

But there is another way. Tackling this unnecessary workload should be the starting point for automation, Julie McKenna, vice-president of sales for North America at mea Platform, told APCIA Today.

“The biggest priorities are improving operational efficiency, accelerating decision-making and enhancing the customer and broker experience,” she said, adding: “Finally, with AI systems, we have technology that has caught up to the industry’s needs.” 

Escaping the pilot trap

Yet turning that ambition into working software is where the industry frequently struggles. Many insurers remain caught in a cycle of trials that never reach production, which McKenna said comes down to operational focus.

“There is rarely a lack of ideas. The real challenge is prioritisation of those use cases,” she said. 

She recommends organisations be selective and identify opportunities that are technically feasible, operationally practical and capable of delivering a clear ROI. “The most successful insurers, brokers and MGAs focus on a small number of high-value use cases where AI can create measurable improvements and build from there.”

Live deployment also does not require insurers to dismantle existing IT architecture. McKenna says software should complement legacy technology. “We don’t see it as a build-versus-buy decision,” she said. “The most successful AI strategies combine internal expertise with proven external capabilities. mea is designed to complement existing technologies … we work alongside internal teams as a technology partner.”

That distinction between a partner and a vendor matters more with AI than with any previous tech wave, McKenna argues. “A vendor delivers software and leaves the insurer to work out what it means for their operation.”

This is the fastest-moving technology the industry has ever seen, and the real question for insurers is who will keep them at the front of it and help them navigate what it means for their operations. 

She said: “They want a team with the insurance expertise to help find where their operation is constrained, solve the bottlenecks and commit to work with them on where to move next for the highest return. That is what differentiates a vendor from a strategic partner.”

Integrating into operations

Operating inside live workflows requires technology built specifically for the sector, with AI that understands the technical phrasing and complex layouts common to commercial cover.

“A domain-specific language model like ours understands the insurance context natively, which improves accuracy, reduces hallucinations and accelerates deployment.” But, insurance-specific does not mean insurance alone. 

To avoid technology lock-in as foundation models evolve, mea’s architecture integrates directly with large-scale hyperscaler systems, including Anthropic, OpenAI and Microsoft Copilot.

In one live deployment, she explained, an insurer embedded the technology directly into its underwriting workflow to extract and interpret complex documents. The integration resulted in faster processing, improved data consistency and reduced manual effort. This gave underwriting teams additional capacity to assess risk and focus on customer engagement during peak renewal periods.

“The key lesson is that the business value comes from embedding AI into operational workflows, as well as generating insights,” McKenna said. 

“The winners in AI won’t necessarily be those with access to the most technology. They’ll be organisations that combine the right data, domain expertise and operational discipline to turn AI into sustainable business value.” 

Oversight can also be tailored to the risk. Rather than apply blanket rules across an entire portfolio, insurers can decide where underwriters need to intervene.

“If they want human-in-the-loop, we’ll work with them to do that at the right times based on their guidelines,” McKenna said. “But on the other end of the spectrum, if they want to enable straight-through processing for cyber, for instance, we’ll enable that.” Underwriters step in only when the system flags exceptions based on triage guidelines.

Turning AI into a commercial asset depends on execution, McKenna said. “The winners in AI won’t necessarily be those with access to the most technology. They’ll be the organisations that combine the right data, the right domain expertise and operational discipline to turn AI into sustainable business value.”

Julie McKenna is vice-president of sales, North America, at mea Platform. She can be contacted at julie.mckenna@meaplatform.com.

For more news from the American Property Casualty Insurance Association (APCIA) click here.

Did you get value from this story?  Sign up to our free daily newsletters and get stories like this sent straight to your inbox.