In what marks one of the most consequential transformations in the history of Indian public sector banking, State Bank of India (SBI) has deployed artificial intelligence and advanced machine learning to underwrite nearly ₹1 trillion (₹1 lakh crore) in loans to Micro, Small, and Medium Enterprises (MSMEs) in fiscal year 2026 (FY26).
Speaking at the annual #FIBAC 2026 industry summit in Mumbai on Wednesday, August 12, 2026, SBI Managing Director Rama Mohan Rao Amara revealed that the country’s largest lender used automated Business Rule Engines (BREs) and proprietary predictive models to evaluate and disburse thousands of credit applications up to ₹5 crore each.
This algorithmic expansion isn’t just a backend software upgrade; it represents a fundamental shift in how capital flows to India’s commercial heartland. By automating the arduous paper-chase that previously defined SME credit evaluations, India’s banking giant is rewiring relationship management, slashing delinquency rates, and proving that legacy state-owned institutions can out-innovate nimble fintech competitors at massive scale.
1. The Death of the Paper Trail: Inside the Machine Driving ₹1 Trillion in Loans
Historically, securing an MSME loan from a public sector bank meant weeks—if not months—of back-and-forth document submission. Loan Officers and Relationship Managers (RMs) spent the majority of their working hours collecting tax records, verifying physical bank statements, cross-referencing credit bureau logs, and drafting manual risk assessments.
SBI’s AI-led underwriting model turns this paper-centric approach on its head by analyzing the dynamic digital footprint left behind by modern Indian businesses.
The Multi-Data Ingestion Framework
Rather than relying solely on audited balance sheets—which are often outdated by the time they reach a credit officer—SBI’s Business Rule Engine (BRE) ingests real-time, structured, and unstructured datasets directly from primary source networks:
- GST Network (GSTN) Filings: Real-time visibility into operational revenues, month-on-month sales velocity, and client concentration risk.
- Banking & Cash-Flow Data: Automated analysis of daily bank account transactions, cash inflows, and electronic clearing metrics.
- Unified Payments Interface (UPI) Flows: Granular insights into micro-merchant transaction volume and daily liquidity.
- Credit Bureau Scores: Integrated credit behavior algorithms evaluating past debt servicing performance.
- Alternative Sectoral Data: Industry-wide trade indicators, geographic supply-chain health, and macroeconomic signals.
By fusing these disparate datasets, SBI’s AI algorithms synthesize a comprehensive credit profile within minutes, allowing both existing clients and New-to-Bank (NTB) borrowers to secure approvals with unprecedented speed.
2. Unlocking “Thin-File” Businesses and Lowering Default Rates
One of the greatest bottlenecks in expanding priority-sector lending across India has been the “thin-file” problem. Millions of small-scale entrepreneurs, micro-enterprises, and sole proprietorships operate primarily through digital payments like UPI, keeping minimal traditional paperwork or official credit histories.
Through AI-powered alternate data scoring, SBI can now evaluate thin-file applicants with high precision, advancing financial inclusion without taking on unquantified risk.
[Digital Footprint: GSTN + UPI + Bank Logs]
│
▼
[SBI Business Rule Engine (BRE)]
│
┌────────────┴────────────┐
▼ ▼
[Instant Approval / Risk] [Early Warning Signals]
│ │
▼ ▼
[Lower Delinquencies] [Pre-DPD Intervention]
The Paradox of Higher Speed and Lower Risk
Conventional wisdom in financial services dictates that as lending volume speeds up, default risk inevitably spikes. SBI’s FY26 performance data directly challenges that assumption.
According to Amara, the portfolio underwritten through the BRE demonstrated lower delinquency rates than manually processed loans. Because machine learning models evaluate thousands of historical risk parameters without human bias, fatigue, or subjective leniency, the quality of credit decisions improves even as operational speed accelerates.
3. From Data-Gatherers to Risk Managers: The Evolution of the Relationship Manager
The deployment of automated underwriting has fundamentally redefined human roles across SBI’s branch network.
For decades, Relationship Managers functioned primarily as high-paid data gatherers and administrative coordinators. The integration of AI has eliminated this administrative burden, allowing RMs to pivot toward strategic, higher-value operations.
“It has actually freed the bandwidth of the people who were otherwise the relationship managers, particularly, who were otherwise spending a lot of time in terms of gathering the data, doing some analysis, et cetera.”
— Rama Mohan Rao Amara, Managing Director, SBI
Reallocated Bandwidth: What RMs Do Now
- Model Validation: Verifying algorithm outputs against ground-reality business contexts during site visits.
- Stress Testing: Running localized impact scenarios against borrowers exposed to broader market volatility.
- Deepening Client Relationships: Consulting with business owners on working capital optimization, cash management, and trade financing.
- Targeted Business Development: Spending time in the field acquiring high-potential clients rather than processing back-office paperwork.
4. Deep Tech in Banking: Early Warnings, LLMs, and Agentic AI
SBI’s technology adoption goes far beyond basic credit-scoring engines. The bank is integrating advanced AI technologies—including Large Language Models (LLMs) and Agentic systems—across every layer of its operations.
| Operational Area | AI Application & Impact |
| Pre-Delinquency Tracking | Models monitor market news, sectoral shifts, and transactional flags to generate Early Warning Signals (EWS) before standard Days Past Due (DPD) triggers occur. |
| Cheque Automation (LLMs) | Large Language Models parse handwritten and unstructured fields on cheques up to ₹10,000 (25% of SBI’s total cheque volume), achieving Straight-Through Processing (STP) with minimal human friction. |
| Customer Experience | End-to-end voice and text conversational AI bots handle routine customer care queries, escalating only complex edge cases to human support reps. |
| Corporate Trade Finance | YonoG / YONO Ji, an agentic AI assistant embedded in Yono Business, autonomously reads unstructured financial documents, populates loan systems, and conducts automated risk evaluations. |
| Tech Operations & Fraud | AI systems process billions of daily log entries in SBI’s Security Operations Centre and predict infrastructure outages via its Resiliency Operations Centre. |
Governance and The “Human-in-the-Loop” Safeguard
To keep machine bias and algorithmic hallucination in check, SBI maintains strict control protocols. A dedicated Control Risk Unit continuously samples and audits AI-processed cheques and credit approvals. If an anomaly is flagged, the error is fed back into the development pipeline to re-train and fine-tune the operational models.
5. Macro Impact: How AI Is Reshaping India’s Lending Ecosystem
SBI’s massive AI roll-out comes at a time of explosive growth across its balance sheet. For the quarter ended June 2026 (Q1FY27), SBI’s gross advances expanded by nearly 19% year-on-year to ₹50 trillion, while its dedicated SME portfolio soared over 22% year-on-year to ₹6.46 trillion.
┌──────────────────────────────────────────────────────────┐
│ SBI Q1FY27 Financials │
├─────────────────────────────┬────────────────────────────┤
│ Gross Advances │ ₹50 Trillion (+19% YoY) │
│ Domestic Corporate Advances │ ₹14 Trillion (+18% YoY) │
│ SME Loan Book │ ₹6.46 Trillion (+22% YoY) │
│ AI-Underwritten MSME (FY26) │ ~₹1 Trillion │
└─────────────────────────────┴────────────────────────────┤
The New Benchmark for Public & Private Sector Lenders
- Closing the Credit Gap: India’s 63 million MSMEs have long suffered from an estimated $300 billion credit gap. SBI’s ability to process ₹1 trillion in small business credit proves that alternate-data AI can economically bridge this deficit at national scale.
- Compressing Cost-to-Income Ratios: While SBI has not yet cited a final numerical drop in its cost-to-income metrics, removing manual overhead from 25% of cheque processing and standardizing MSME evaluations delivers massive, structural productivity gains.
- Pushing Fintechs Out of the Niche: Traditional public sector banks were once thought to be too slow to leverage real-time digital infrastructure. By coupling its vast 23,000+ branch footprint with cutting-edge AI pipelines, SBI is successfully protecting its market share against emerging neobanks and non-banking financial companies (NBFCs).
A New Blueprint for Global Banking
State Bank of India’s journey through FY26 offers a global case study in how heritage financial institutions can harness artificial intelligence responsibly.
By deploying AI to handle complex data aggregation while repositioning human personnel as strategic decision-makers, SBI hasn’t just underwritten ₹1 trillion in MSME loans—it has built the operational engine that will power India’s digital economy for decades to come.
