3 weeks ago
SBI Uses AI to Revolutionise MSME Lending and Cheque Processing
The State Bank of India is the biggest bank in India, and it has started using smart computer programs called artificial intelligence, or AI, to do some of its work.
In one year, the bank used AI to approve loans for small businesses worth almost Rs 1 lakh crore — that is a huge amount of money.
Each small business loan can be up to Rs 5 crore.
Before saying yes to a loan, the AI looks at many kinds of information about a business, like its tax records, bank accounts and credit history.
AI also reads small cheques all by itself and checks that they are filled in correctly.
About one out of every four cheques the bank receives is small enough for AI to process this way.
Human workers still check some of the machine's work to make sure it does not make mistakes.
AI also helps the bank spot money problems early, so it can help businesses before big troubles start.
Because the machines handle boring tasks, bank workers have more time to spend helping customers.
The bank says AI is making its work faster, safer and better for everyone.
SBI used artificial intelligence to underwrite nearly Rs 1 lakh crore (around ₹1 trillion) in MSME loans of up to Rs 5 crore each during FY26, covering both new-to-bank and existing customers.
Underwriting draws on digital footprints including GST Network filing data, bank-account information, credit-bureau scores and other structured and unstructured data, combined through a business rule engine (BRE).
Loans underwritten through the BRE have shown lower delinquency, and AI has freed up relationship managers from data collection and preliminary analysis tasks.
The bank uses AI, including Large Language Models, to process cheques up to Rs 10,000 — about 25% of its cheque volumes — through straight-through processing with minimal human intervention.
A dedicated control risk unit samples AI-processed cheques to identify errors, while AI is also used for early-warning signals, fraud risk management, customer-care bots and the YonoG corporate banking assistant.
- Who
- State Bank of India (SBI), India's largest lender, with Managing Director Rama Mohan Rao Amara revealing the details at the annual FIBAC 2026 event.
- What
- SBI used AI to underwrite nearly Rs 1 lakh crore in MSME loans of up to Rs 5 crore each during FY26 and to automate processing of cheques up to Rs 10,000.
- Where
- Mumbai, India, at the annual FIBAC 2026 event.
- When
- During the 2025-26 financial year (FY26); announced on a Wednesday at the FIBAC 2026 event.
- Why
- To improve efficiency, customer satisfaction and risk management, and to free up relationship managers for higher-value work.
Key facts
- Bank
- State Bank of India (SBI)
- AI-underwritten MSME loans
- Nearly Rs 1 lakh crore in FY26
- Loan limit
- Up to Rs 5 crore per loan
- Borrowers covered
- New-to-bank and existing customers
- Automated cheque processing
- Cheques up to Rs 10,000 (about 25% of cheque volumes)
- Processing method
- Straight-through processing (STP) with minimal human intervention
- Underwriting data sources
- GST Network filing data, bank-account information and credit-bureau scores
- Oversight
- Dedicated control risk unit samples AI-processed cheques to identify errors
Quotes
SBI Managing Director Rama Mohan Rao Amara
Managing Director of State Bank of India
“"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."”
livemint.com
“"Now the role of the relationship manager is more in terms of validating that, looking at even the stress testing, whatever is coming out of that model."”
livemint.com
Sources
Related news
SBI shifts entire cheque processing to AI, freeing workforce
RBI Governor says AI alone can help limit AI frauds
SBI chairperson warns AI fraud can outrun conventional banking systems
RBI Deputy Governor Says AI Needs Human Judgment In Banking
SBI Chairman Pushes AI Banking Transformation Beyond Urban Customers
