The Platform

OpenCredit is architected as a trust-first credit aggregation platform, built around four coordinated agentic layers that compute and act on a borrower's Trust Confidence Profile.

Made in India

The Trust Confidence Profile

Every MSME, small business, vendor, or agri-business receives a score, component breakdown, and confidence band — computed from UPI, GST, and Account Aggregator-consented data, gated behind explicit, purpose-bound consent and never pulled before that consent exists. A borrower's confidence profile is never static: it is recomputed on a defined cadence and refreshed automatically as new consented data arrives.

Deterministic, Not a Black Box

How the Trust Score Is Computed

Every score is produced by a rules-based, deterministic scoring engine — not an opaque proprietary model. The same consented UPI transaction history is read across five dimensions, and every factor behind the resulting number is disclosed back to the borrower and the lender alike.

Velocity

Transaction frequency and volume over time.

Regularity

Consistency of payment timing, month over month.

Diversity

Breadth of merchant categories transacted with.

Seasonality

Stability of consumption across seasonal cycles.

Financial Resilience

Ability to hold positive balances and avoid overdrafts.

0.742
ROC-AUC — validated scoring accuracy
68%
First-time credit seekers scored with just 3 months of UPI history
Apache 2.0
Scoring engine open-sourced for community and regulatory audit

Four Agentic Layers

An agentic underwriting stack computes the trust signal and recommends each borrower to the lender types best suited to their confidence band.

Layer 01

Credibility Agent

Builds a borrower's trust and behavioural profile from consented digital data, producing the underlying confidence signal.

  • UPI transaction patterns
  • GST filing history
  • Consented Account Aggregator data
Layer 02

Underwriting Agent

Translates the Trust Confidence Profile into lending-ready, lender-type-specific risk parameters.

  • Lender-type risk parameters
  • Explainable factor breakdown
  • Confidence-band scoring
Layer 03

Matching Agent

Routes each borrower to the right lender type from a structured taxonomy, based on confidence band and profile shape.

  • Cooperative banks & credit societies
  • NBFCs
  • Merchant & scheduled banks
Layer 04

Offer Orchestration Agent

Manages offer generation, comparison, and transparent, borrower-facing delivery.

  • Multi-lender offer comparison
  • Transparent, explainable criteria
  • Borrower-facing delivery
Understood, Not Just Translated

A Trust Score India Can Actually Read

A confidence band means nothing to a borrower who can't understand why they received it. OpenCredit pairs the deterministic scoring engine with a fine-tuned language model whose only job is explanation — it has zero role in calculating the score itself, and can never move a number the rules engine already fixed.

That model turns each score's factor breakdown into a plain-language explanation in the borrower's own language — the same underlying number, delivered the way a person actually thinks about their finances.

Explanations available in

HindiMarathiTamilTeluguBengaliKannadaMalayalamGujaratiPunjabiOdiaEnglish

The scoring rules and their weights are identical in every language — only the explanation changes, so no borrower is scored differently because of the language they read in.

Bankability Score App

A standalone, zero-regulatory-approval Bankability Score mobile app is live for MSMEs today, with the farm and tractor-financing segment next.

It lets a borrower see their own Trust Confidence Profile directly — before any lender is ever involved.

Ready to Build on Verifiable Trust Infrastructure?

See how OpenCredit's agentic stack computes a transparent, explainable trust signal before a single lender ever sees a profile.