ARCs turn to AI, analytics as retail recovery business gains scale

Asset reconstruction companies (ARCs) are increasingly deploying artificial intelligence (AI), machine learning, and data analytics to sharpen recovery decisions, automate processes and improve operational efficiency as they seek to extract better value from stressed assets.

In its red herring prospectus, Asset Reconstruction Company (India) (Arcil) said it is using AI and machine learning for compliance checks and is enhancing models to predict recovery potential and validate discounting frameworks.

“After each payment is made, our AI and machine learning systems check the payer’s name for anti-money laundering (AML), Financial Action Task Force (FATF) and politically exposed person (PEP) certifications, with relevant loans subsequently updated and tagged,” the document said.

The company is also taking AI into the recovery process. “We are focussed on enhancing our AI and machine learning models to predict recovery potential and validate discounting frameworks,” the company said in its statement.

It is incorporating real-time application programming interfaces (APIs), communication analytics covering bots, interactive voice response (IVR) and tele-calling, and natural language processing (NLP)-based speech analytics. These, it said, will “drive AI powered recommendations designed to identify effective collection channels and follow up tactics”.

The use of AI could extend across the asset recovery cycle, including acquisition, valuation, legal processes and borrower communication. According to experts, the growth also comes amid growing stress in retail segment.

The stress in retail and MSME loans has been rising. According to Arcil’s RHP, the retail stressed assets across banks and NBFCs grew at a compound annual growth rate (CAGR) of 12.3 per cent between FY20 and FY26, with bank retail stress rising 13.5 per cent and NBFC retail stress 9.6 per cent during the period.

MSME stress has also increased, with overall stress across banks and NBFCs rising at a CAGR of 7.6 per cent from about Rs 6.8 trillion in FY22 to Rs 9.2 trillion in FY26. MSME credit outstanding, meanwhile, grew at a much faster 18.8 per cent CAGR to Rs 82.2 trillion as of March 2026, while NPAs in the segment stood at about Rs 1.9 trillion.

“AI and technology adoption among ARCs is still at a relatively nascent stage, but it has been picking up, particularly with the growth in retail business,” said Hari Hara Mishra, chief executive officer (CEO) of the Association of Asset Reconstruction Companies. “The retail customer base expands and the industry has to deal with larger volumes of customer data, behavioural trends and other information.”

“ARCs are also beginning to explore the use of AI and technology in processing and analysing this data. Some larger ARCs have begun investing in dedicated technology centres and bringing multiple technology initiatives together, indicating that the use of AI and technology in the sector is gradually gaining traction,” Mishra added.

According to industry experts, AI could analyse historical recovery trends, collateral values and macroeconomic data to identify assets with higher recovery potential, while AI-led voice bots, chatbots and NLP could automate borrower communication and assess intent to pay. It also identified OCR/NLP-based document digitisation and legal analytics as potential applications.

Arcil already uses data analytics, account prioritisation, geo-tracking, UPI/QR payments and heat maps to identify stress clusters and improve collections. The company said its technology investments are aimed at scaling operations without a corresponding increase in costs.

“We believe that we will be able to increase the scale and effectiveness of our operations without a proportionate increase in our operational expenses,” the document said.