Why Insurers Struggle to Scale AI: EXL’s Sumit Taneja on Breaking Legacy Mindsets and Data Silos

EXL Exec Sumit Taneja on AI Adoption in Insurance

Sumit Taneja, EXL’s SVP and global head of AI consulting and implementation, points out that a legacy mindset is a key barrier preventing many insurers from advancing quickly with AI. While some companies have applied AI to specific use cases, many struggle to grasp a broader, integrated vision for AI adoption.

During AI sessions held with American insurers last year, Taneja observed that many participants found it hard to relate to the long-term potential and practicality of AI. To address this, EXL suggested taking incremental steps with a clear roadmap for where they want to be in two to three years.

Insurers ‘Think Very Linearly’

Taneja believes that the insurance industry’s traditional workflows and processes encourage a linear approach to work. This mindset limits the ability to see how AI can be embedded seamlessly into daily operations rather than existing as isolated applications.

He explains, “They just can’t think multiple things will happen in parallel… It will probably take much longer to get people out of that thinking that multiple things you used to do can actually happen now in parallel.”

Legacy Systems and Data Limitations

Foundational issues like numerous legacy systems and siloed, unstructured data continue to slow AI adoption in the insurance sector. These challenges are particularly pronounced among U.S. insurers.

“It’s not that they are not keen to do it, but there are many foundational problems... One is the multiplicity of admin systems, and second is their data landscape is still very siloed. Unstructured data has not been managed properly, so using some of that data for AI is still a few months or maybe a year away,” Taneja shared.

Many insurers currently use AI primarily for simple automation or marketing tasks. This observation aligns with findings from EXL’s 2025 Enterprise AI Study, which highlights that while machine learning adoption is growing, generative AI use remains limited.

“We are still scratching the surface of what AI can deliver to insurance,” Taneja said. “At the same time, adoption regarding GenAI and Gen Tech is still early days.”

Study Finds Slow Adoption

The EXL Enterprise AI Study surveyed senior executives in insurance and other industries worldwide, including 50 U.S. insurance leaders. The results reveal a gap between insurers’ perceptions of their AI progress and their actual standing compared to other sectors.

• 24% of U.S. insurers believe they are “far ahead” of peers in AI adoption.
• Only 10% qualify as AI leaders when benchmarked against other industries.
• 58% have started using machine learning AI, mainly for claims automation and fraud detection.
• Only 38% have adopted generative AI, the lowest rate among all surveyed industries.

Nearly 90% of insurers acknowledge scaling AI is very important. About 60% plan to pilot GenAI in claims, and 46% are already running GenAI pilots in underwriting.

“I haven’t seen much end-to-end AI deployment yet. Larger insurers are moving in that direction, but full transformation of underwriting or claims is likely a few years away,” Taneja noted.

Addressing Foundational Challenges

Taneja stresses that insurers must tackle legacy systems and data silos to accelerate AI adoption.

He cited an example of an insurer with 28 policy administration systems, none equipped to support AI infrastructure. “They need to reduce that to two or three systems over the next three to four years to really unlock value,” he said.

Having many legacy systems complicates data access, since a single customer’s information can be scattered across multiple platforms. Insurers in this position will need to invest in data warehouses and platform modernization before AI solutions can be effectively scaled.

About EXL Service

EXL Service is a New York–based global data analytics and digital operations company founded in 1999. It employs around 1,500 data scientists and serves diverse industries.

For insurance professionals interested in AI skills development, exploring targeted courses can be a practical step. Resources such as Complete AI Training’s courses by job provide focused AI learning tailored for insurance roles.