In AI, the winners will be those who build trust at scale: Mukul Jain, CTO, Axis Max Life Insurance
Every CIO has sat through the proof-of-concept parade: a dazzling demo, a promising pilot, and then a long silence while the project stalls against legacy integrations, risk committees and unclear ownership. In insurance, where the product is a promise kept years or decades later, that gap is wider than most.
Axis Max Life Insurance has been working to close it. The insurer has embedded AI across underwriting, claims and customer service, and its leadership argues the lessons reach well beyond insurance. “The biggest lesson has been that AI success has very little to do with the model alone,” says Mukul Jain, the company’s CTO.
Underwriting delivered first
Asked which use case has had the biggest business impact, Jain points to underwriting. Faster, more consistent decisions have improved customer experience, shortened turnaround times, and let the business scale without a proportional rise in operational effort. That decoupling of growth from headcount is the outcome most CIOs are chasing.
What made it work was everything around the algorithm. “The real challenge is bringing together quality data, process redesign, risk controls, integration with core systems, and business ownership,” Jain explains.
He also has a lesson for leaders who struggle with adoption. “Customers don’t care whether AI is involved. They care about speed, accuracy, and transparency. If AI improves those outcomes, adoption follows naturally.”
Claims: speed without compromising trust
Claims is where an insurer’s brand is tested. With a 99.8% claims settlement performance and a rapid-payout commitment, Axis Max Life has set a high bar, and Jain is clear about what it demands. “Claims is one of the most sensitive moments in a customer’s relationship with an insurer,” he says. “Delivering speed without compromising trust requires multiple capabilities working seamlessly together.”
The company digitised the entire claims value chain rather than automating isolated steps, combining workflow automation, intelligent document processing, API-led integrations, business rules, real-time data access and straight-through processing where it is appropriate. Around that pipeline sit the controls: risk segmentation, auditability, maker-checker checks, exception management and specialist reviews where judgement is required.
“The objective is not just faster claims,” Jain says. “It’s making the claims experience both empathetic and trustworthy while maintaining strong governance behind the scenes.”
Humans and AI, by design
Jain rejects the idea that this is a choice between machines and people. AI is extremely effective at processing volume, spotting patterns and improving consistency, he says, while human judgement becomes critical when decisions involve ambiguity, exceptional circumstances, ethics or customer sensitivity.
The corollary is proportionality. The greater the customer impact, the greater the need for transparency, explainability and human oversight. “Trust remains a human responsibility, irrespective of how much technology is involved in the process,” Jain says. For CIOs drafting AI governance frameworks, that is a practical rule: oversight should scale with the stakes of the decision, not apply uniformly to every workflow.
Agentic AI: the next wave, with guardrails
Jain expects the first wave of agentic AI to land where work spans multiple systems, stakeholders and decisions, including customer servicing, underwriting, claims, seller enablement, policy issuance and technology operations. The shift is from people navigating systems to agents orchestrating them. “AI agents will be able to orchestrate activities, gather information, complete routine actions, and proactively manage workflows end to end,” he says.
The payoff, he argues, goes beyond productivity. Response times shrink, operational complexity falls, and people spend more time applying judgement than managing processes. He is equally direct about the limits: “Autonomy without accountability is not a model that will work in a regulated industry.”
The unglamorous foundation
Behind the results is several years of modernisation: cloud adoption, API-first architecture, platform engineering, application modernisation and stronger engineering practices, alongside enterprise data foundations that give a unified, trusted view of information.
“Building an AI-ready enterprise starts with building a data-ready enterprise,” Jain says. The next step is a trust layer around data, so the organisation knows where data resides, who can access it, how it is used and whether it is properly governed. His conclusion pushes back on the more-data instinct: “AI readiness is less about having more data and more about having trusted, governed, and accessible data.”
What separates leaders from followers
Access to AI models will become increasingly democratised, Jain expects, so the model will not be the long-term differentiator. Leaders will combine trusted data, deep business context, modern platforms, disciplined governance and the ability to execute at scale. The real test is repeatability. “Many organisations can showcase impressive proofs of concept. Far fewer can industrialise AI across the enterprise and consistently translate it into better customer outcomes, stronger risk decisions, and measurable business value.”
He ties it back to the industry’s founding principle. “Insurance has always been a business built on trust. AI won’t change that. If anything, it will make trust an even bigger differentiator.” The winners, he says, will use AI “not just to automate work, but to strengthen trust with customers, distributors, and employees.”