New You can now listen to Carrier Management articles!

Claims adjusting resonated with Gia Sawko.

Throughout her 30-year insurance career, she’s viewed adjusters as people who make others whole again. While working as the managing director of global safety and claims for Whole Foods Market, she saw her work as the process of putting folks back together.

“And that really was a big focus for me my entire career,” Sawko shared. “That it wasn’t a claim. It was an actual person.”

Sawko is a former adjuster and currently the director of claims for Gradient AI. She recently spoke about the roles artificial intelligence can play in the claims process during Carrier Management’s InsurTech Summit.

She doesn’t envision AI eliminating adjusters. Instead, she sees the technology making them more efficient.

“So that’s where I think it’s a benefit to adjusters or to carriers,” Sawko said, “and where people will start to embrace it more. Because they’ll see it as this virtual assistant.”

She began her insurance career as an adjuster at Kemper before joining the team at Whole Foods. She was named vice president of business impact at Gallagher Bassett in 2016 and joined Gradient AI six years later.

How AI Can Help Claims

Gradient AI is a predictive modeling company. On the claims side, the platform’s pattern predictions serve as another tool adjusters can use to make informed decisions. Sawko described Gradient AI’s capabilities as an expert “tap on the shoulder” that can sound early warnings for claims.

“And that does really push toward a very positive outcome or mitigating some of those losses,” she added.

Gia Sawko

In short, predictive modeling uses statistics to project outcomes. The models are “trained” on historical data that informs these predictions. Sawko believes capturing institutional knowledge – and pairing it with predictive modeling that can spot red flags – has the potential to be hugely helpful for supervisors.

The National Association of Insurance Commissioners reports that machine learning models “can help quickly assess the severity of damages and predict the repair costs from historical data, sensors and images.”

According to the Gradient AI website, the company’s product identifies claims risks earlier in life cycles, which enables proactive steps towards effective outcomes and cost containment. It also reduces the duration of outstanding claims to better contain the costs and identify high-risk claims that require the attention of a more seasoned adjuster who can help manage claims to more favorable outcomes.

When asked how the insurance industry talent gap is affecting claims management, Sawko said there’s not only a shortage of adjusters, but also that the industry is “begging for change.” She believes AI has the power to quicken and enhance claims processes and could prove especially important as institutional knowledge leaves the industry in the coming years.


When asked about the biggest challenges regarding claims handling and the use of AI in insurance, Sawko pointed to a need for good data quality and the ability to clearly explain to clients why the AI is telling adjusters to do something.

According to Rutgers University, bias “can cause artificial intelligence to make decisions that are systematically unfair to particular groups of people, and researchers have found this can cause real harm.”

The university reports that even if creators make conscious efforts to eschew bias in the data used in their algorithms, bias can still be baked into the data they select.

Another AI challenge is ensuring claims adjusters code unstructured data properly. When adjusters don’t understand the importance of properly entering incident tables and reports, entries can become directionless – and in turn, predictive models can suffer, Sawko said.

She added that being able to explain why a predictive model is designed a certain way is crucial. The inability to explain what the models are doing in a claim log means adoption suffers, she said.

“So, if someone says, ‘I’m going to raise this reserve,’ if the question was why, and somebody said, ‘Because the model told us to,’ as a risk manager, I wouldn’t allow that,” Sawko explained. “But I think understanding and having the ability to share that knowledge with clients, especially on the TPA (third party administrator) side, that would be really a very big deal.”

One Big Quote

Sawko said that there is so much work to do right now in the insurance industry that, without AI, claims can become very task oriented.

“I want adjusters to stay in the business, but I want them to be able to go to a soccer game and be able to not work the weekends,” she said. “So, this is really where balance between the two and that human touch can really come out, but it also can take away a lot of this duplication. And happy adjusters make happy claims and make happy clients, for sure.”

Watch the full interview with Gia Sawko on demand.