Insurers struggle to scale AI because fragmented data, legacy systems, and the need for human judgment in high-stakes customer interactions limit adoption.
Naresh Ramaswamy, SVP of Technology and Chief Architect at Virtusa, explains that scaling AI requires shifting from legacy data warehouses to targeted data products while treating the technology as a force multiplier rather than a human replacement.
Success comes from orchestrating AI across technology, governance, and reskilled teams, supported by structured change management and workflow-ready data products.
Insurance AI rises or falls on orchestration. Claims and service interactions demand empathy and judgment, so automation only works when it operates within clear governance guardrails and a unified data foundation. Leaders who align technology, data architecture, and workforce readiness turn AI into measurable operational gains, while fragmented pilots stall under regulatory pressure and siloed systems.
Naresh Ramaswamy is Senior Vice President and Chief Architect at Virtusa, leading digital transformation firm for insurance, consumer, and information services worldwide. With 25 years of experience, he has guided organizations such as American Express, United Health, Lufthansa, and Walt Disney Parks and Resorts through enterprise resource planning, robotic process automation, and agentic AI, modernizing operations, scaling platforms, and enhancing customer experiences.
He emphasizes that successful AI adoption in insurance requires a combination of technology, governance, and human-centered design. “AI in insurance isn’t about replacing the human in the loop. It’s about elevating the human focus on empathy, judgment, and the moments that matter most,” says Ramaswamy.
The productivity prize: Development teams report gains from AI-assisted coding. “At Virtusa, we see at least a 40 to 50 percent productivity improvement,” Ramaswamy explains, citing tools such as Copilot, Bedrock, and Gemini that accelerate output and reduce repetitive work. Scaling these gains enterprise-wide requires coordinated oversight of privacy, model risk, and evolving state-level AI regulations. Proactive governance creates stability, allowing teams to move faster with confidence.
Higher autonomy: Organizations are learning to define where AI operates independently, where it supports decision-making, and where human judgment remains essential. Ramaswamy compares enterprise AI to autonomous vehicle models such as Waymo. “Even the most advanced systems operate at Level 4 autonomy, where human oversight remains available by design. We are still years away from a completely autonomous CSR. There is no such thing as fully autonomous software at the moment."
Ramaswamy emphasizes that AI cannot scale without structured, high-quality data. Internal silos slow adoption, while proof-of-concepts tied to external public data move faster because privacy and governance risks are minimized. Insurers are shifting from data warehouses to productized pipelines, including Agent 360, Underwriting 360, and Claims 360, which enable curated, workflow-ready insights.
Structured self-service: Legacy insurers often struggle with fragmented or siloed data. “The common problem chief data officers report is business teams not getting the data they need. There’s no self-service,” Ramaswamy says. In contrast, he points to newer insurers such as Lemonade that design data architecture for straight-through processing from day one. Productized pipelines enable self-service analytics and faster operationalization.
Virtusa’s Helio platform exemplifies AI orchestration, connecting large language models to insurance systems while embedding FinOps discipline and compliance into design. “Embedding governance into the platform reduces risk and accelerates deployment, making compliance a structural enabler rather than a post-deployment hurdle,” explains Ramaswamy. Platforms such as Helio allow organizations to integrate multiple AI tools, maintain visibility over cost and performance, and operationalize models consistently across teams.
Human adoption: Technology delivers value when employees embrace it. Virtusa applies the ADKAR framework, which consists of Awareness, Desire, Knowledge, Ability, and Reinforcement, to guide workforce adoption. KPIs tied to objective key results quantify value realization, while reskilling ensures AI frees humans for high-value, nuanced decision-making rather than replacing them. “AI manages high-volume, repetitive tasks, allowing professionals to focus on exceptions and complex judgments,” he says.
The shift toward AI orchestration is moving teams from traditional development to designing automated workflows, managing agent interactions, and overseeing end-to-end processes. While AI can handle repetitive, high-volume tasks, human judgment remains essential for complex decisions, customer empathy, and exception handling. Organizations that align technology, governance, and workforce reskilling can unlock the full potential of AI, amplifying human performance rather than replacing it. “Human touch is critical in insurance. AI can amplify it, but it cannot replace it,” Ramaswamy concludes. “Is it going to displace jobs? No way. We need more people to do the things we want to do.”
