IRMAn Talks by Puneet Gupta, PRM 20: AI & Data for Risk Management & Financial Inclusion
IRMAn Talks by Puneet Gupta, PRM 20
Sat, 28 Mar 2026 / 17:00 IST.
Puneet Gupta, founder of Kaleidofin, spent 25 years building AI-powered financial inclusion after noticing a stark problem: microfinance loan sizes had stagnated in real terms - a INR 15,000 loan that could buy two cows in 2001 barely buys one today. The industry was growing in volume but regressing in real impact.
His core insight: the sector had built process-based lending (group models, collateral) rather than genuine risk underwriting. Kaleidofin uses behavioral signals and alternate data instead - tracking how customers responded to demonetization, floods, or political loan waivers to assess willingness to pay. For customers with no credit history, lookalike models fill the gap.
A key design principle: one national model is exclusionary by design. Kaleidofin builds hyper-local cluster models, distinguishing, say, a rain-fed cereal farmer from an irrigated vegetable farmer near a city. This same methodology scaled seamlessly to Kenya and Bangladesh.
Results: sub-3% default rates while approving 60-80% of applicants, versus an industry approving just 10-25%.
Looking ahead, Puneet flagged that beyond credit, almost nothing has been built for informal-sector customers - no pensions, no insurance, no investment products - and called AI the tool to finally unlock all of these.