How Sarvam helped transform frontline insurance sales for SBI Life's 3.5 lakh+ sales force
Built by Sarvam and McKinsey with SBI Life, SPARK is a bespoke Work Agent for SBI Life's 3.5 lakh+ frontline sellers.
In late March 2026, SBI Life launched SPARK to support its frontline sellers in their day-to-day work. Built directly into WhatsApp, SPARK brings together the information, tools, and collateral sellers need to prepare for customer conversations and act on opportunities.
Early results show a consistent pattern. Sellers who adopted SPARK grew faster than non-adopters, with the gap holding even among lower-performing sellers starting from similar performance baselines.
The Challenge
At SBI Life, frontline agency sales are driven primarily by Life Mitras and Unit Managers, distinct from Bancassurance and other distribution channels. Their productivity is tracked through a single metric, Individual Retail Premium, or IRP.
The challenge was twofold.
Give every seller faster access to what they need to sell
- Preparing for a customer conversation requires product information, benefit illustrations, financial calculations, customer-ready collateral, and a pitch relevant to the prospect. SBI Life needed to bring these capabilities together without asking its field force to adopt another application or change how they already work.
Improve productivity across the field force, including sellers who had become inactive
- The opportunity went beyond making high-performing sellers more efficient. SBI Life also wanted to understand whether AI could improve outcomes for lower-performing sellers and whether Life Mitras who had been inactive for months could return to production.
The Solution
SPARK brings these capabilities together in one place.
From a single conversation, a seller can:
- Generate customer-ready collateral such as Get Pay and Benefit Illustrations
- Run numbers using SBI Life's integrated financial calculator
- Get instant answers to product queries
- Put together a tailored sales pitch for a prospect
- Receive proactive nudges on the next best action
SPARK was built and rolled out in close collaboration with SBI Life's Agency, Bancassurance, IT, and Learning & Development teams, together with McKinsey.
Measuring Real Adoption
The analysis set out to answer three questions:
- Whether SPARK adopters outgrew non-adopters on IRP.
- Whether that difference held among lower-performing sellers starting from similar performance baselines.
- Whether deeper use of SPARK was associated with multi-policy selling and the reactivation of dormant Life Mitras.
To make sure the analysis captured genuine use rather than a one-time login, an "adopter" was conservatively defined as a seller who generated at least 10 collaterals or benefit illustrations, or sent at least 30 messages on SPARK.
With adoption defined, the analysis compared:
- Adopters and non-adopters on IRP growth, year-on-year over the same May-July period.
- Low-performing adopters and non-adopters starting from similar IRP baselines.
- Different levels of SPARK usage and multi-policy conversion, grouping sellers by the number of collaterals generated and comparing this with NOP ≥ 2.
- Activation among dormant Life Mitras, looking at sellers who had been inactive for the previous nine months and whether they returned to production.
The Impact
Across the analysis, four early signals stood out in how SPARK adoption and usage related to seller performance.
50-60%
YoY IRP growth among SPARK adopters
May to July 2026, compared with 0-15% growth among non-adopters over the same period
200-250%
YoY IRP growth among low-performing adopters
Versus 0-15% for non-adopters starting from a similar performance baseline
~65%
Multi-policy conversion at high usage
July 2026, among sellers generating 20+ collaterals, versus ~10% among those generating none
20%
YoY increase in activation of dormant sellers
June 2026, among Life Mitras inactive for the previous nine months
Adopters grew significantly faster
- From May to July 2026, SPARK adopters grew their IRP by 50-60% year-on-year, compared with 0-15% growth among non-adopters over the same period.
Lower-performing sellers saw even stronger growth
- Among low-performing sellers, SPARK adopters grew their IRP by 200-250% year-on-year, compared with 0-15% among non-adopters starting from a similar baseline.
- The difference remained even when sellers were compared within similar IRP performance tiers, making it harder to attribute the result simply to stronger sellers being more likely to adopt SPARK.
Deeper usage tracked with deeper selling
- In July 2026, approximately 65% of sellers who generated 20 or more collaterals on SPARK achieved a multi-policy conversion, defined as NOP ≥ 2.
- That compared with approximately 25% among sellers who generated up to five collaterals and around 10% among sellers who generated none.
- This is an association rather than proof of causality, but the relationship was consistent as platform usage increased.
Dormant sellers also returned to production
- Life Mitras who had been inactive for the previous nine months saw a 20% year-on-year increase in activation in June 2026.
- Among the sellers who reactivated that month, approximately 20% were SPARK adopters.
Building a More Productive Frontline
Three months in, the results point in a consistent direction. Across seller performance, depth of selling and reactivation, SPARK adoption was associated with stronger outcomes across the field force.
It is still an early vantage point, and no single measure establishes causality. But taken together, the results show what becomes possible when AI is built into the way a frontline already works.
For SBI Life, SPARK is putting that capability in the hands of more than 3.5 lakh frontline sellers, right where they already work, driving stronger productivity, deeper selling, and new momentum across the field force.
More Customer Stories
How Sarvam's voice agents work across Mahindra Finance's business
Mahindra Finance·Financial ServicesHow IIM Udaipur is scaling its bilingual curriculum with Sarvam
IIM Udaipur·EducationEqual scales its Call Screening Assistant to enterprise volume with Sarvam
Equal ·Technology