One platform for origination, servicing, collections and the ledger — with AI doing the reading a credit officer would otherwise do by hand. Built to be run by the team you already have, not a platform team you would have to hire.
A CRM nobody likes, an LOS from one vendor, an LMS from another, a collections tool bought separately, and a finance team reconciling all of it in spreadsheets at month end. Six licence fees, six support queues, and no single view of a customer.
They are not integrations of separate products. A lead, an application, a loan and a collection case are the same record moving through its life.
Leads, campaigns, activities and a customer 360 that survives the loan being closed.
Application, KYC, bureau, credit, underwriting, offers, sanction and disbursal — with every integration call visible on the page.
Schedules on reducing balance, repayments, charges, foreclosure, restructuring and closure.
DPD computed nightly, bucket movement, cases allocated to agents, PTP tracking and field visits.
An append-only double-entry ledger. Every disbursal, repayment, charge and write-off posts a balanced journal entry as it happens.
Credit memos written from bureau, bank statements and KYC. Underwriting recommendations that show their reasoning.
One lender wants DigiLocker; another insists on CKYC; a third accepts a document upload. One wants video KYC before sanction, another after. In most platforms each of those is a release.
In MODNAR a journey is a versioned document attached to a lender's product. Reorder the steps, swap the provider, add a condition, save. The next application runs the new flow — and applications already in progress keep running the version they started on.
Both the borrower journey and the LOS stage tracker render from the same definition. There is nowhere left to hardcode the order.
Not a chatbot in the corner. The credit memo is written from the bureau report, the bank statement analysis and the KYC record — the same evidence a credit officer would read, assembled into the document they would have spent forty minutes writing.
The recommendation comes with the factors it weighed and the ones that gave it pause, so an underwriter can disagree with the reasoning rather than argue with a score.
Every AI decision is stored with its inputs and its output, so a decision from six months ago can still be explained.
Not a single-lender product that will be made multi-tenant later. Every record carries its tenant, and isolation is enforced by row-level security in the database rather than by developer discipline in the application.
Each lender gets its own branded journey on its own subdomain, its own products, its own credit policy and its own users — operated from one platform.
The isolation is a database policy you can read on screen and a query you can run to prove.
Bureau, PAN, DigiLocker, account aggregator, penny-drop, eSign, mandate, payout. Each call is logged against the application that triggered it, with the request, the response, the HTTP status and the latency in milliseconds.
Open any application and the whole vendor conversation is there in order — including the retries, and including the calls that failed.
The same log answers the operations question and the finance question: what happened, and what it cost.
A schedule on reducing balance, and a balanced journal entry behind every event. Here is a ₹5,00,000 personal loan at 16% over 36 months — the EMI never changes, but what it pays for does.
Every external service is a sandbox stub for now — the gateway is built so a live provider is a configuration change, not a rewrite. Everything else is real.
The fastest way to judge this is to put one of your own products into it and push a real application through, end to end. That takes an afternoon — and you keep whatever we configure.
Start the conversation →