A dispute that reaches the Banking Ombudsman is not a dispute that was too complex for the institution to resolve — it is a dispute that was not resolved within 30 days, or was resolved in a way the borrower found inadequate, or was never acknowledged. The Ombudsman is not the destination for complex cases; it is the destination for cases the institution managed poorly. The Dispute Manager Agent AI's triage function exists to identify, at the moment a dispute is raised, which category it falls into, what the resolution pathway is, and whether the institution's existing records support or contradict the borrower's claim — so that 92% of disputes can be resolved at the institution level, by a structured, documented, time-bound process, without the borrower ever needing to invoke the Ombudsman Scheme.
The RBI's dispute resolution framework — what the institution is obligated to do
Under the Integrated Ombudsman Scheme 2021 and the Digital Lending Guidelines, lending institutions are required to acknowledge every complaint within 5 working days and resolve it within 30 calendar days. A complaint that is not resolved within 30 days can be escalated to the Ombudsman without any further action by the borrower. The institution that resolves 92% of disputes within 21 days — before the borrower has to think about the Ombudsman — does not have an escalation problem. The institution that resolves 70% of disputes within 30 days and handles the Ombudsman on the remaining 30% has a permanent operational overhead, a reputational risk, and the near-certainty of supervisory observations in the next RBI examination.
The cost of an Ombudsman escalation is not the fine — which is typically modest. It is the management attention required to respond, the examination record it creates, and the borrower who told 8 people about the experience before the Ombudsman found in the institution's favour. Dispute resolution at the institution level, done correctly and within SLA, is cheaper in every dimension than Ombudsman resolution — and the Dispute Manager Agent AI makes the structured, evidence-based, within-SLA resolution of most disputes operationally achievable without a large grievance team.
The 6 dispute categories — and the AI's handling approach for each
Borrower disputes a penal charge for late payment — claims they paid on time or the NACH failed due to a bank error
The most common dispute category. The borrower maintains they paid on time; the institution records show a bounce or late credit. The Dispute AI retrieves the NACH return code from the payment gateway (specific bank return codes distinguish borrower-side insufficiency from bank-side technical failures), the borrower's bank statement credits for the relevant date, and the CBS DPD entry timestamp. Where the return code indicates a bank-side technical failure (codes R29, R01-bank) rather than a borrower-side failure (R01-customer), the penal charge reversal is system-recommended without requiring an officer decision. AI resolution rate: 89% without human escalation.
Borrower complains that a collections agent called outside permitted hours, contacted family members, or used inappropriate language
All collections calls are logged in the call management system with a timestamp. The Dispute AI retrieves the call log for the date range specified by the borrower, checks each call against the FPC permitted window (8 AM to 7 PM, no national holidays), identifies whether any contact was made with family members (logged as "third party contact"), and pulls the call recording where available. If an FPC violation is confirmed, the AI generates the remediation record — a formal acknowledgement to the borrower, an internal complaint against the collections agent, and a refund of any penal charges that were being collected by the violating agent. AI resolution rate: 78% (remaining 22% require investigation where recording is unavailable).
Borrower disputes the rate being applied — believes their rate is different from what was committed at sanction
The Dispute AI retrieves the KFS issued at sanction (timestamped, borrower-acknowledged), the sanction letter, and the rate currently applied in CBS. Where the CBS rate matches the KFS rate, the AI generates a clear explanation letter showing the borrower exactly where their agreed rate is documented, with a side-by-side comparison of the KFS rate and the CBS rate. Where there is a rate discrepancy — which occurs occasionally when a rate change in CBS was applied to an existing account incorrectly — the AI flags the error to the Credit operations team for immediate correction and confirms the correction timeline to the borrower. AI resolution rate: 91%.
Borrower claims they paid but the payment is not reflected — or disputes a double debit from the NACH mandate
For payment attribution disputes, the Dispute AI requests the borrower's bank statement for the relevant month via WhatsApp (or Account Aggregator if consent is active), retrieves the NACH debit records from the payment gateway, and cross-references the two. A double debit — which can occur in rare NACH processing errors — is immediately confirmed and a refund process is initiated without requiring officer intervention. A payment claimed but not in the institution's records is cross-checked against the IMPS/NEFT settlement records before the dispute can be formally closed either way. AI resolution rate: 84%.
Borrower claims they were not given the KFS before accepting the loan, or that the KFS did not reflect the actual terms
The Dispute AI retrieves the KFS from the document repository with its creation and borrower-acknowledgement timestamps, the eSign record, and the sanction letter date. If the KFS was acknowledged before the sanction date — as is required — the AI generates a response with the complete documentation chain. If the KFS creation timestamp is the same day as the sanction date (an onboarding QC error flagged elsewhere in this article series), the AI acknowledges the procedural gap, confirms the terms were in fact as stated, and issues a formal apology and remediation for the process failure. AI resolution rate: 88%.
Borrower disputes foreclosure charges applied on prepayment, or claims full payment was made but the account remains open
For prepayment charge disputes, the Dispute AI retrieves the product's prepayment terms from the KFS (which states the prepayment charge schedule), confirms the regulatory requirement (for floating rate loans, zero prepayment penalty per RBI directive; fixed rate loans may have a charge up to 2% on the outstanding principal), and generates a clear computation showing the charge applied and the regulatory basis. For accounts claimed closed but still open, the Dispute AI retrieves the closure payment record, any NOC issued, and the current CBS account status — and initiates a reconciliation if there is a discrepancy. AI resolution rate: 82%.
The live triage dashboard: November 2025 dispute queue
The 0 Ombudsman cases this month is not a lucky run — it is what happens when every dispute is triaged within 48 hours, evidence is retrieved automatically, and resolution timelines are tracked to the day
The prior quarter's 4 Ombudsman escalations were all cases where the institution's response to the initial complaint was a generic acknowledgement letter with a "we are investigating" message that was not followed by a substantive response within 30 days. None of the 4 were complex disputes. All 4 were resolvable with the same evidence the Dispute Manager AI would have retrieved within 24 hours of receipt. The borrowers who escalated to the Ombudsman did so not because the institution's position was wrong — in 3 of 4 cases, the Ombudsman found in the institution's favour — but because 30 days passed without resolution. The Ombudsman is the regulator's provision for borrowers whose institutions did not respond adequately within the prescribed window. An institution with the Dispute Manager AI does not give its borrowers a reason to use it.