
State Bank of India has fully automated processing of cheques up to Rs 10,000 using artificial intelligence, covering about 25% of its cheque volumes. The bank also used AI to underwrite nearly…
State Bank of India has fully automated processing of cheques up to Rs 10,000 using artificial intelligence, covering about 25% of its cheque volumes. The bank also used AI to underwrite nearly Rs 1 trillion of MSME loans of up to Rs 5 crore each in the financial year 2025-26, managing director Rama Mohan Rao Amara said at FIBAC 2026.

SBI uses AI across customer lifecycle including credit underwriting, fraud detection, and customer service. The AI-led underwriting draws on GST data, bureau scores, and bank account information. Amara said the model has shown lower NPAs than traditionally underwritten portfolios. A control risk unit still reviews samples of AI-processed cheques to catch errors and retrain models.
The hype around AI replacing bankers is overblown. SBI's own numbers show a human control unit still samples every batch of AI-processed cheques. That is not automation replacing jobs; it is humans training machines. The real story is the asset quality benefit: lower NPAs from rule-engine underwriting. That is a measurable test to watch in quarterly results. If NPAs stay low, AI has earned its place. If not, the fanfare was premature. Does the annual report show NPA figures for this portfolio separately? That will settle it.
Source: bfsi.economictimes.indiatimes.com
This story was synthesised by AI from the source linked above.