How an NBFC cut response times to <2 minutes and filtered 70% of non-viable files
Our customer is a mid-size non-banking financial company (NBFC) that distributes home loans across India through branch staff and a wide network of direct sales agents (DSAs). Through a period of steady growth, the lender built a strong frontline presence. But its sales conversations were consistently outpacing its credit operations. Leadership wanted the sales floor to give every borrower a fast, honest read at the very first meeting, without ever taking the lending decision out of human hands.
Before working with Gyde, every home-loan file hit the same friction: the borrower's question and the moment someone could answer it were days apart. The cost was lost bankable business and avoidable rework.
Agents couldn't confirm eligibility without routing files to the credit desk. This turned a two-minute conversation into a one- to two-day wait, which was long enough for borrowers to apply elsewhere.
Applications outside policy still traveled all the way to underwriting before being declined, burning agent and credit team time that should have gone to viable borrowers.
Outcomes and the reasons behind decisions varied from one officer to the next. This made it difficult to tell a borrower's story and left a weak position in a fair-lending review.
Risk and compliance would not delegate approval authority to opaque software, which is why earlier AI lending initiatives stalled before the pilot.
Gyde built an underwriting soft signal that sits with the agent as a simple chat during the borrower meeting. It answers the borrower's question in the moment and goes deeper on demand, and hands a fully documented decision packet to a human underwriter who always makes the final call.
The agent enters credit score, annual income, loan amount, and property value. Within seconds, a soft-signal card returns a clear status: it's likely approvable, needs a closer look, or isn't a fit yet. It also provides the maximum eligible loan amount, key affordability (FOIR) figures, loan-to-value ratios, and the exact documents to collect.
As the chat continues (employment type, existing EMIs and liabilities, preferred tenure, co-applicant income), the read gets sharper with every detail, all in plain language the borrower can follow.
Gyde tests how the file holds up under pressure (higher interest rates, a shorter tenure, a dip in income) so affordability risk is visible upfront and nobody is surprised later.
One tap turns the conversation into a full report for the underwriter: the eligibility maths in full, every policy rule passed or flagged, compensating strengths weighed in, stress-test results, and a clear recommendation — plus a plain-language version for the borrower.
Decisions run on the lender's own configurable policy, stay consistent file-to-file, and are saved with their reasons — so risk and compliance can reopen and defend any decision long after the fact. The system structures and explains; people decide.
In a focused pilot (one region, around 40 agents, roughly ten weeks), the underwriting soft signal changed the shape of the funnel and gave credit & compliance the confidence to widen it.
Time to a first answer, down from one to two days
Fewer non-viable files reaching underwriting
More applications progressed to the next stage
Saved per agent, every week
Soft signals that matched the eventual underwriting outcome
Muthoot Fincorp Slashes Customer Acquisition System Training and Onboarding Time by 63% and Data Errors by 78% with Gyde
Read Full StoryIf you're an NBFC or bank running home loans through DSAs, we can show you exactly how the soft signal works (on your policy, with your numbers).
Book a 30-minute walkthrough