B2B journey

Underwrite in 30 minutes.

An NBFC underwriting team · ₹20k–₹2L unsecured personal loans · 5,000 applications a day

A mid-size NBFC wants to compress a three-day underwriting cycle into thirty minutes — without sacrificing the audit trail their risk committee reads.

Marketing claimb2b-credit-risk-shap
See the lender story Talk to us
The story · 5 scenes

Beginning to end.

An NBFC underwriting team · ₹20k–₹2L unsecured personal loans · 5,000 applications a day
·A mid-size NBFC wants to compress a three-day underwriting cycle into thirty minutes — without sacrificing the audit trail their risk committee reads.
  1. Scene 01 · Receive the application
    An applicant submits through the NBFC's own funnel. Their consent bundle and identity documents are POSTed to the Mukodi B2B credit-risk endpoint within the same request.
  2. Scene 02 · Score with explainable reasons
    Mukodi returns a credit-risk score, a SHAP attribution per factor, and a Courtroom-signed bundle. The risk team's reviewer sees the same bundle the API caller saw.
  3. Scene 03 · Surface the document story
    Document verification runs alongside the score — PAN, Aadhaar, bank statements, salary slips. Every page is OCR'd, cross-checked, and tamper-flagged.
  4. Scene 04 · Listen for fraud signals
    If the same applicant identity has touched another flagged session in the last 24 hours, the fraud verdict comes back as a sub-200ms ping the underwriter sees on the same screen.
  5. Scene 05 · Audit the decision, years later
    Every step is hash-chained — score, attribution, verdict, override. The NBFC's compliance team replays the decision from the audit ledger when the regulator visits.
Helpers on call
The stack · helpers + models

Who's actually doing the work.

Features used
B2B API + Python / Java / Node SDKsCredit risk + SHAP + counterfactualsHash-chained audit trailDocuments & KYC36-format ingest (Tier 1-5)
Honest limits

What this flow doesn't do.

  • The credit-risk score is a decisioning input — the lending decision (approve, decline, price) stays with the partner under their licensed underwriting policy.
  • Sub-200ms fraud verdicts are bounded by the partner's network round-trip; the SLO refers to Mukodi-side processing time.
  • Replay-on-audit assumes the partner has stored the request bundle alongside the response; Mukodi's audit chain stores hashes, not raw PII payloads.
← Back to all use cases