Artificial intelligence is now a central theme in technology conversations across insurance. The harder question is no longer whether AI can create value, but how insurers can translate that potential into measurable business outcomes.

In a recent episode of TDI Talks, Hugh Terry, CEO and Founder of The Digital Insurer, spoke with Srinivasan Varadharajan, Senior Sales Director at HCLTech, about the ROI of AI in insurance and what it takes to move from isolated pilots to scaled adoption.

For Varadharajan, the starting point is the connection between industry knowledge and technology execution. Having begun his career in insurance operations before moving into technology, he said working at “the cusp of business and technology” helps him understand both what insurers need and how technology can be adopted effectively.

In the context of AI adoption in insurance, this understanding is important. Delivering AI ROI at scale means moving beyond technology deployment and addressing operating models, customer journeys, underwriting philosophy, data foundations and the way people work.

Moving from AI usage to business change

AI is becoming increasingly central to insurance operations, with orchestration across key processes beginning to be infused with AI. Varadharajan described AI as a “great equalizer,” particularly because it can help smaller organizations compete more effectively with larger insurers.

But he also warned that insurers should avoid measuring AI success simply by usage. The better measure is whether AI changes the way work gets done.

“Don’t measure your AI adoption by the usage, measure it by the change in the ways of working,” he said.

AI applied on top of an old process may create efficiency gains, but it can also become an expensive way of preserving the same operating model. Varadharajan said insurers need to address process engineering, data engineering and the underlying technology estate if they want AI to deliver meaningful returns.

“AI adoption should be considered as a top operating model transformation and not as a technology transformation project,” he said.

Where insurers are seeing traction

Across the insurance value chain, Varadharajan highlighted three priority areas:

1. Underwriting: AI can support better risk selection and improve the quality of the book. The challenge is explainability, particularly in an industry built on pricing models that must be trusted, governed and fair.

2. Claims: Claims can offer a more immediate ROI case. AI can help reduce cycle times, improve fraud control, limit leakage and reduce customer calls or complaints. But claims also require care because it is one of the areas where the “human element” and empathy remain especially important.

3. Distribution: Distribution offers a different value profile. AI can improve sales conversion, support better product positioning and reduce acquisition costs. But Varadharajan warned that insurers cannot simply bolt AI onto existing channels.

ROI needs a broader view

The ROI case for AI in insurance is not always straightforward. In life insurance, for example, the benefits of improved underwriting may take longer to appear because the impact on risk selection and profitability emerges over time.

Varadharajan also cautioned that revenue growth does not always translate into profit growth. An AI use case may appear successful in isolation while creating a negative effect elsewhere in the business.

“You have to zoom out and see the overall impact as well,” he said.

That means ROI needs to be assessed at both functional and enterprise level. Insurers need to understand immediate gains, second-order effects and the longer-term impact on the balance sheet.

Start small, but do not stay small

Many organizations begin with a focused use case because it helps build confidence, test user acceptance and prove value. Varadharajan said this approach can work, especially when a business is still developing its AI maturity.

But there is a risk in running too many disconnected pilots. The faster route to scale is to create a common layer that includes governance principles, approved frameworks, integration patterns, vendor approaches and data access policies. This allows insurers to start with targeted use cases while building the foundations for wider adoption.

“It’s not just a niche or a big bang approach. It’s about a niche and a big bang approach,” he said. “They both are different stages on the same path of adoption.”

As AI adoption matures, reusable capabilities such as document summarization, retrieval and workflow automation can be applied across multiple functions. That improves speed, reduces duplication and helps insurers move toward platform-led transformation.

Varadharajan said HCLTech is also supporting this journey through dedicated AI capabilities and its AI Labs, which help clients move from experimentation into implementation. These environments allow HCLTech to work with insurers and other organizations on AI use cases, test adoption approaches and develop scalable solutions across different markets and functions.

For insurers, the message is clear. AI can deliver meaningful ROI, but only when it is connected to business priorities, supported by the right foundations and embedded into the way the organization works.

“When you start, you start as a function-led initiative, but when you scale, you scale it as a platform-led initiative,” concluded Varadharajan.