Automating Loan Origination: How We Cut Approval Time From Days to Minutes

FundFlick (fundflick.in) is our own lending operations SaaS, and its Loan Origination System is the module where speed matters most — it's the first thing a borrower experiences, and the slowest legacy process it replaces. A simple application that used to take days now clears in 4-5 minutes. Here's what actually had to change to get there.

The old bottleneck was never the credit decision

Most lending teams assume the slow part of origination is underwriting judgment. In practice, the real time sink is almost always upstream of that: collecting documents, manually keying identity details, and waiting on a bureau report that someone has to request, download, and read. The credit decision itself, once the data is actually in front of an underwriter in a usable form, is often the fastest step in the whole flow.

Instant eKYC and bureau fetch, wired at the API layer

The fix is removing the manual data-collection step entirely, not making it faster for a human to do. eKYC and credit bureau fetch happen as direct API calls the moment an application is submitted — identity is verified and a bureau report is pulled automatically, before an underwriter ever opens the file. That's the single change that accounts for most of the time saved: a process that used to be "submit, then wait for someone to manually chase two separate external checks" becomes "submit, and both checks are already back by the time anyone looks at it."

SLA-driven task routing, not a shared inbox

Speed at the integration layer doesn't help if the application then sits in a generic queue. FundFlick's task management runs on Kanban boards with auto-allocation rules and SLA tracking — a submitted application is routed to the right underwriter automatically, and it carries a visible deadline instead of quietly aging in a shared inbox. The auto-allocation logic is doing real work here: without it, "instant eKYC and bureau fetch" just moves the bottleneck three steps downstream to whoever happens to check the queue next.

Multi-tenant, but the SLA clock is per-lender

Because FundFlick serves multiple lending businesses on one multi-tenant platform, SLA rules aren't a single global setting — each tenant configures its own approval thresholds and routing rules against its own risk appetite and staffing. A rule engine that only supported one global SLA configuration would force every lender on the platform into the same operational tempo, which doesn't match how different lending businesses actually run.

Where automation stops and a human has to decide

Not every application is eligible for the fast path. Simple, clean applications — clear eKYC match, clean bureau report, within standard product parameters — are the ones that clear in minutes. Anything with a flag (identity mismatch, adverse bureau signal, an amount outside standard parameters) routes to manual underwriting instead of getting forced through an automated decision it shouldn't make. The speed number only means something because it's honest about which applications qualify for it.

The 4-5 minute number isn't a demo trick. It's what's left over once you remove the two things that were never actually adding underwriting value — manual document handling and unrouted queue time — and leave the credit decision to happen exactly as carefully as before.

If you're evaluating a loan origination system, ask specifically what happens to an application that doesn't qualify for the fast path — a vendor's honest answer to that question tells you more about the system's real design than any average-approval-time number on its own.