A TAT commitment to a borrower — "your loan will be disbursed in 7 working days" — is only as good as the institution's ability to know, at any moment, whether every step of the 7-day process is on schedule. If the credit assessment sits idle in a queue for 14 hours while the analyst handles other files, the 7-day commitment is broken before the borrower even notices — and the institution finds out at Day 9, not Day 3. The Onboarding SLA Agent AI monitors every step of every active application in real time, computes the SLA position of each application at each stage, and alerts the responsible party before the delay becomes a breach — not after.
The problem with end-to-end TAT measurement — and what step-level monitoring changes
Most institutions measure TAT as a single number: the time from application submission to disbursement, averaged across all applications. This number tells management how long the process took, not where in the process the time was lost. An application that disburses in 11 days instead of 7 may have spent 6 of those days in credit assessment, or 6 of those days waiting for a QC review, or 6 of those days because the borrower was slow to respond to a document request. The intervention required is different for each cause: a credit assessment queue problem is a staffing allocation problem; a QC review delay is a prioritisation problem; a borrower document delay is a communication problem. End-to-end TAT measurement obscures the distinction between these causes. Step-level monitoring reveals it.
The Onboarding SLA Agent AI assigns an SLA clock to every step in the onboarding pipeline — not just the overall application. The credit assessment step has a 24-hour SLA. The QC review step has a 4-hour SLA. The property valuation step has a 3-working-day SLA. If the credit assessment clock has been running for 20 hours with no action taken, the SLA Agent AI knows — and alerts — 4 hours before the breach, while there is still time to intervene. A notification that arrives 4 hours after the breach is an incident report. A notification that arrives 4 hours before the breach is an opportunity to prevent it.
The SLA framework: step-by-step TAT commitments by product
| Onboarding step | SLA (HL) | SLA (MSME) | SLA (LAP) | Alert at | Owner |
|---|---|---|---|---|---|
| Application receipt to acknowledgement | 2 hours | 2 hours | 2 hours | 75% | Ops / Digital |
| Document collection complete | 3 working days | 2 working days | 4 working days | 70% | RM / Onboarding AI |
| QC review (Onboarding QC AI) | 4 hours | 4 hours | 6 hours | 75% | QC team |
| Bureau pull and CIBIL processing | 1 hour | 1 hour | 1 hour | 90% | System / automated |
| Credit assessment (analyst) | 1 working day | 4 hours | 2 working days | 75% | Credit officer |
| Credit committee / senior approval | 1 working day | 4 hours | 1 working day | 75% | Credit committee |
| Property valuation (HL/LAP) | 3 working days | N/A | 4 working days | 70% | Empanelled valuer |
| Legal review (HL/LAP) | 2 working days | N/A | 3 working days | 70% | Legal team |
| Sanction letter issuance | 4 hours | 2 hours | 4 hours | 75% | Credit / Ops |
| KFS issuance and borrower acknowledgement | 24 hours | 12 hours | 24 hours | 75% | Ops / Onboarding AI |
| NACH mandate submission | 1 working day | 4 hours | 1 working day | 80% | Ops / borrower |
| Disbursement processing | 4 hours | 2 hours | 4 hours | 80% | Finance / Ops |
| Overall end-to-end commitment | 7 working days | 3 working days | 10 working days | 80% | MD / Ops head |
The real-time pipeline: all active applications · November 14, 2025 · 14:30
An alert that fires 4 hours before the credit assessment SLA breaches is an opportunity. An alert that fires 4 hours after the breach is a notification of what already happened.
LA-2025-18851's credit step has been in breach for 3 hours 20 minutes. The SLA Agent AI sent an alert to the Credit Manager at 11:00 AM — when the step was at 75% of its 4-hour SLA. The analyst had until 12:00 PM to complete the assessment. The alert fired at 11:00 AM. The Credit Manager has not confirmed action. The SLA Agent AI will escalate to the Credit Head at 14:00 PM (2 hours past breach) if no action is confirmed. The borrower has not been notified yet — if resolution happens by 16:00 PM, the borrower's experience is unaffected. If it reaches tomorrow without resolution, the SLA Agent AI triggers a proactive update to the borrower and re-assesses the overall disbursement date. The 75% threshold alert is the mechanism that keeps the breach from becoming a borrower experience problem — but only if someone acts on it.