Paytm Launches AI-Powered Credit Scoring Engine for Thin-File Borrowers in Tier-2 Cities
Amit Yadav
Paytm's new AI credit engine uses 200+ alternative data signals to assess 400 million thin-file borrowers excluded from formal credit — showing 23% lower defaults in a 6-month pilot.
Paytm has unveiled an AI-based credit assessment engine designed specifically for thin-file borrowers — individuals with little or no formal credit history — in India's tier-2 and tier-3 cities. The system uses over 200 alternative data signals, including mobile recharge patterns, utility bill payment regularity, UPI transaction frequency, and app usage behaviour, to generate a creditworthiness score independent of CIBIL history.
The Thin-File Problem
An estimated 400 million adult Indians are either credit-invisible (no credit bureau record) or thin-file (fewer than three credit products in their history). Traditional lenders cannot assess this population using conventional scorecards, leaving them reliant on informal moneylenders who charge annualised interest rates of 40–120%.
Paytm's new engine, built on a gradient-boosted ensemble trained on four years of proprietary transaction data from 330 million registered users, aims to extend formal credit access to this population at rates competitive with personal loans from NBFCs.
Early Performance
In a six-month controlled pilot covering 85,000 borrowers in Uttar Pradesh, Madhya Pradesh, and Telangana, the AI engine demonstrated a 23% lower default rate compared to a control group approved under conventional thin-file models. Average loan size was ₹18,000 with a 14-month tenure.
The RBI's regulatory sandbox has granted Paytm conditional approval to scale the product nationally, subject to quarterly audits of model fairness across gender, geography, and linguistic community. Paytm has said it will integrate the engine with partner NBFCs and small finance banks starting in Q2 2026.