Onboarding SLA Agent AI

Invoked continuously — stage transition events from the LOS drive the TAT clock in real time

• –Maintains a TAT clock for every active application — measuring the elapsed time at the current stage and the cumulative elapsed time since application submission. The TAT clock stops when an application is on hold (waiting for borrower documents, waiting for co-applicant KYC, waiting for valuation) and resumes when the hold is resolved — distinguishing between time consumed by the lender's process and time consumed by the borrower's response. Both are tracked; only the lender-controlled time is counted against the SLA commitment.
• –Monitors each application against the stage-level TAT targets — the time budget allocated to each stage in the application process: KYC verification (24 hours), QC review (2 hours), credit assessment (48 hours for standard applications, 72 hours for complex cases), disbursement (24 hours after credit approval). An application that is within its stage-level budget is green on the dashboard; one approaching the budget is amber; one that has exceeded it is red.
• –Sends a 4-hour advance warning to the operations head for any application approaching its overall SLA boundary — providing enough time for the operations team to escalate the application within the current stage, adjust prioritisation, or communicate proactively with the borrower. A borrower who receives a proactive communication before the SLA is breached has a better experience than one who contacts LendingIQ to enquire about their application after the promised TAT has passed.

Invoked when: a stage's average TAT exceeds its design assumption by more than 50%, or a weekly analysis cycle runs

• –Detects bottlenecks at the stage level — identifying where applications are spending disproportionately more time than the stage's design assumption. A bottleneck is distinguished from a normal distribution of application complexity by its systemic nature: a single complex application taking longer than average at the credit assessment stage is not a bottleneck; a sustained pattern where the average credit assessment TAT across all applications is 50% above its target is a bottleneck. The agent identifies the latter through trend analysis over the prior 7 days, not from single-application observations.
• –Characterises the bottleneck by its likely type — volume bottleneck (more applications arriving at a stage than the stage can process in the available time), quality bottleneck (applications arriving at a stage with errors that require rework before the stage can be completed), or dependency bottleneck (a stage is waiting for an external input — valuation report, bureau refresh — that has a longer turnaround than assumed). The bottleneck type determines the appropriate remediation: a volume bottleneck requires capacity addition; a quality bottleneck requires upstream process improvement; a dependency bottleneck requires external SLA management.
• –Produces a weekly TAT analysis report — the average, median, and 90th percentile TAT for the full onboarding journey and for each individual stage, by product type and application channel. The P90 figure is the most operationally significant: a process where 90% of applications complete within the SLA target is meeting most borrowers' expectations even if the mean is slightly above target; a process where the P90 is significantly above the mean indicates a long tail of applications with unacceptably long TATs.

Invoked when: an application's cumulative lender-controlled TAT exceeds the SLA commitment

• –Escalates every SLA breach to the operations head immediately — with the application reference, the product type, the channel, the stage where the breach occurred, and the total elapsed time. The escalation includes the full TAT timeline for the breached application — where time was spent at each stage — so that the operations head can identify the primary delay driver without additional investigation.
• –Triggers the borrower communication workflow for SLA-breached applications — notifying the operations team to contact the borrower with an updated TAT and an apology for the delay. Borrowers who are not proactively communicated with after a SLA breach are more likely to escalate to the Banking Ombudsman; proactive communication converts a SLA failure into a managed relationship event.
• –Logs every SLA breach with its root cause stage, product type, and channel — building the dataset for the weekly TAT analysis and the operations head's process improvement agenda. A breach that occurs in isolation is an operational exception; a pattern of breaches in a specific stage, product, or channel is a systemic process problem that requires redesign rather than individual case management.