How Agentic AI could help insurers reach more Indians by 2047

India’s insurance sector collected ₹11.93 lakh crore in premiums and issued 41.84 crore policies in FY2024–25. As per the Insurance Regulatory and Development Authority of India (IRDAI) Annual Report 2024-2025, insurance penetration was 3.7% of GDP, including 2.7% for life insurance and 1% for non-life insurance.

These figures highlight the limited scale of the industry and the opportunity to extend financial protection to more people. Doing so will require insurers to reach customers efficiently, keep premiums low, process applications faster and maintain service across the life of a policy. Agentic AI could help perform tasks across these workflows under defined instructions and controls.

On Mint’s series All About AI, moderator Abhishek Singh, Associate Editor at LiveMint spoke with Arun Kumar Parameswaran, EVP and MD of Salesforce South Asia; Amneet Singh, Partner - Advisory and Leader, Business Transformation at PwC; and Sankaranarayanan Raghavan, Chief Technology and Digital Officer at IndiaFirst Life. The discussion examined how insurers could use agentic AI and determine where autonomous decision-making makes business sense.

Giving Insurance Agents More Time With Customers

India's insurance industry relies heavily on distribution networks. Equipping large numbers of agents with technology while keeping costs manageable has been a longstanding challenge.

Parameswaran said that, in emerging economies such as India, human labour can still cost less than AI. Insurers therefore need to identify where the technology delivers enough value to justify its cost. He sees agentic AI as a way to extend technology to tens or hundreds of thousands of agents more economically.

Raghavan described one potential benefit as “green time”: the time agents regain when technology takes administrative work off their hands. Less time spent handling routine documentation could leave agents with more time for customer conversations, follow-ups and understanding individual insurance needs. Insurers do not have to replace the human distribution network to make it more productive. They can use technology to support the people who connect customers with policies and help them navigate their options.

From Routine Requests to Underwriting Decisions

Some insurance tasks are easier to automate than others. A customer changing an address or telephone number, for example, should be able to complete the request through a conversation rather than navigate a lengthy manual process, Parameswaran said. An AI agent could verify the information, apply the relevant rules and complete the transaction.

Amneet Singh said insurers could use risk thresholds to determine which claims an AI agent may process autonomously. Straightforward cases could move through the workflow more quickly, while those exceeding defined limits could be referred for human review. In underwriting, AI could help staff review documents, organise information and assess applications against established guidelines.

Parameswaran illustrated the potential speed gains with an underwriting example. A process taking four hours could, he said, potentially be reduced to less than 30 seconds when AI scans documents, checks them against rules and presents a recommendation with supporting reasons. This is a speaker-cited illustration, not a universal benchmark for underwriting.

The value lies in accelerating document-heavy reviews and giving underwriters more time to assess the cases that need their expertise. The appropriate level of autonomy depends on the task: a routine request may be completed end-to-end, while a complex claim or underwriting decision may require human judgement.

What IndiaFirst Life Is Connecting

IndiaFirst Life has built a connected workflow spanning sales leads, underwriting, servicing, claims and customer feedback. Raghavan said this foundation helps the company identify where AI agents can be introduced.

He outlined three areas for Agentforce: new-business and underwriting workflows, claims processing, and email automation. The company is prioritising tasks with sufficient volume to scale, decisions that can be reversed where necessary, and clear governance over agent actions.

Salesforce’s IndiaFirst Life case study reports that the insurer connected more than 50 internal and external systems through over 110 APIs. It also reports a 58% reduction in policy issuance time following streamlined sales processes and automated underwriting.

That improvement predates the announced Agentforce expansion and should not be attributed to Agentforce alone. In an announcement dated 27 August 2026, Salesforce said IndiaFirst Life had selected Agentforce for new-business underwriting, customer service and claims processing. The planned applications include reviewing customer documents, flagging discrepancies, generating compliant service responses and routing claims that need further review.

Connected systems make information available across the policy lifecycle; AI agents can then perform selected tasks within those workflows. The aim is to build on existing operational capabilities rather than automate every process indiscriminately.

Where Human Oversight Still Matters

Insurance decisions must be consistent, explainable, and able to withstand regulatory scrutiny. Parameswaran pointed out that AI systems are probabilistic. The same prompt may not always produce the same answer. Insurance processes often require consistent outcomes under the same rules.

Guardrails, transaction-level controls, and audit logs are essential. Insurers need to know what information an agent used, which rules it applied, what action it took, and when to escalate a case. The boundary depends on the consequences of a decision. An address change under established verification rules differs from a complex claim requiring experience and judgement. Singh’s risk-based approach provides a way to distinguish between these cases. Raghavan said IndiaFirst Life intends to keep human oversight to ensure decisions remain consistent with its principles.

Raghavan stressed that the company is introducing agents where transaction volumes justify scaling and where decisions are reversible. These criteria help determine where more autonomy is appropriate. Responsibility does not shift to technology. As Raghavan emphasised, even when an AI agent makes a decision, the organisation stays accountable.

Building More Continuous Customer Relationships

Life insurance needs change as people move through different life stages. Someone starting a career may need a different level of protection from someone raising a family or planning for retirement.

Raghavan argued that insurers need to stay engaged beyond the point of purchase or claim because customers’ financial goals and coverage needs change. AI could help insurers respond by organising customer information, identifying relevant follow-up moments, and helping distributors prepare informed conversations. The agent remains important because recommendations must connect with a person’s circumstances and decisions.

Raghavan also suggested that individuals could eventually use their own AI agents to assess changing needs and connect with insurance distributors. He presented this as a possibility for the coming years, not a capability IndiaFirst Life already deploys. For insurers, the immediate question is how to use existing customer and policy information to make interactions more relevant. Timely communication helps customers review their protection needs as their lives change.

Redesigning the Workflow Around AI

Raghavan cautioned against inserting agents into existing processes simply to demonstrate adoption. Insurers should examine how decisions are made, where information passes between teams and which steps can be handled more effectively by an AI agent.

That requires defining the business problem before selecting the technology and measuring the results afterwards. Faster turnaround, improved agent productivity, better customer service and more consistent decisions are meaningful outcomes; the number of agents deployed is not. Parameswaran made a related point about investment: technology should deliver business value rather than become an end in itself.

For insurers working towards “Insurance for All by 2047”, the opportunity is to combine connected systems, carefully chosen AI tasks and the reach of human distribution networks. Progress will depend on how well those elements work together to serve customers, while preserving the accountability and trust that insurance requires.