State Bank of India has transitioned AI from pilot projects to core operations, underwriting nearly ₹1 trillion in small-business loans and automating 25% of its cheque processing, signalling a major shift in digital banking practices to boost MSME credit access.
State Bank of India is accelerating its push into artificial intelligence, saying it has used the technology to underwrite nearly ₹1 trillion of small-business loans in the current financial year while also automating a quarter of its cheque processing. Managing Director Rama Mohan Rao Amara said at FIBAC 2026 that the bank has moved AI from pilot projects into core operations across lending, servicing and risk controls.
Amara said SBI has been using AI throughout the customer lifecycle, including credit underwriting, portfolio monitoring, fraud detection, customer service and day-to-day operations. The system draws on GST records, bureau scores, account data and other structured and unstructured inputs to assess borrowers, allowing the bank to extend loans of up to ₹5 crore to both new and existing customers.
The bank’s use of technology comes as public sector lenders are under pressure to expand MSME credit. The finance ministry has set a target of ₹17.31 trillion in outstanding MSME lending across public sector banks in FY26, while reporting elsewhere has suggested the government wants digital-footprint lending to scale further, potentially lifting the loan threshold to ₹1 crore to ₹2 crore. SBI’s own digital business loan programme has already processed ₹74,434 crore across 225,000 accounts through August 2025, according to Business Standard.
Amara said AI is also improving loan quality by helping SBI spot risk earlier and cut delinquency in portfolios screened through its business rule engine. He said the bank is using the same tools to support unsecured lending, particularly for thin-file borrowers such as small businesses and proprietorships that may not have much formal credit history but still need access to finance.
Beyond lending, SBI is deploying AI to detect vulnerable exposures before stress becomes visible through missed payments or days-past-due metrics. It is also using the technology in its security operations centre to process large volumes of system logs and in its resilience unit to predict possible breakdowns.
One of the clearest examples of the shift is cheque processing. Amara said cheques of up to ₹10,000 account for about 25% of SBI’s cheque volumes and are now handled through a straight-through processing system that uses vision-based large language models to read the cheque and check mandatory fields. He said human involvement is now minimal, although a control risk unit still samples processed cheques to catch errors and decide whether the model needs further training.
Disclaimer: This article is intended to inform and educate, not to recommend or endorse any financial product, investment or strategy. Please consider your own financial circumstances and seek professional advice where appropriate before making financial decisions.
