YES Bank and Open Financial Technologies have launched an agentic payments platform designed to identify recurring payments at risk of failure and intervene before the transaction is presented, applying artificial intelligence to one of the most persistent operational problems in lending and collections.
The i-Mandate platform targets banks, non-bank financial companies, mutual fund providers and enterprises. Traditional recurring-payment systems generally attempt a debit against a predefined account on a predetermined date. When funds are unavailable, the payment fails, creating additional collection costs for institutions and potential charges and inconvenience for customers.
i-Mandate introduces an intelligent layer before that point. Its architecture is built around three functions - Predict, Decide and Act - allowing the platform to assess payment behaviour, determine an intervention permitted under an institution’s policies and engage with the customer before the scheduled debit fails.
Potential actions include changing the payment date, switching the account used for repayment, registering another mandate or completing the transaction using an alternative payment method. The AI payment agent can interact with customers in real time while institutions retain control over the policies governing the actions it can take.
FROM COLLECTION TO PREVENTION
The model represents a significant change in how lenders could approach recurring payments. Conventional collections processes typically begin after a payment has failed. Predictive and agentic systems create the possibility of moving intervention earlier in the cycle.
For example, if the system identifies that a customer is unlikely to have sufficient funds on a scheduled repayment date, it could engage before the debit occurs and help the customer choose an authorised alternative. That could reduce failed transactions, collection workloads and customer friction simultaneously.
The initial use cases include loan repayments and equated monthly instalment collections, although the infrastructure could extend to mutual fund investments, subscriptions and enterprise payments.
WHY IT MATTERS
The launch provides a practical example of agentic AI moving into core banking workflows. Much of the financial sector’s early adoption of generative AI focused on customer service, employee productivity and information retrieval. Agentic systems go further because they can determine and execute actions within predefined boundaries.
Payments and collections are particularly significant applications because actions directly affect customer funds and therefore require stronger controls around consent, authentication, governance and auditability. For lenders, even incremental reductions in payment failures can improve collection efficiency, reduce operational costs and produce a better customer experience across large loan portfolios.
The strategic significance of i-Mandate is not simply that it uses AI. It demonstrates how financial institutions are beginning to redesign established banking processes around systems capable of predicting an outcome and taking an authorised action before a problem occurs.