Skip to content
The Lendforge Platform

Lending infrastructure your engineering team can ship in two weeks.

Lendforge handles the underwriting model, decision API, adverse action generation, and audit log. Your platform handles the user experience.

Request Access Read Developer Docs

Architecture

How data flows through a Lendforge integration

Your Platform event data + applications POST /v1/decisions Lendforge API Layer Decision Engine model + policy rules response: decision, score, reason_codes, offer

Product modules

Three capability areas, one integration

POST /v1/decisions accepts an application payload containing the applicant identifier, platform ID, and lookback window. The response returns a decision (approved, countered, or declined), the approved amount and APR range when applicable, and an ordered list of reason codes. A credit policy rules layer runs after model scoring: you define cutoff thresholds, eligible product tiers, and hard exclusions, and Lendforge enforces them deterministically before any response leaves the engine.
Push events via a streaming webhook or ship a bulk CSV export of historical transaction records. Lendforge ingests raw events, applies normalization logic, and computes model features: 90-day GMV, rolling velocity ratio, category concentration index, tenure percentile, and 40 additional platform-native signals. You do not write feature engineering code. The pipeline handles it.
Every declined or countered decision generates an adverse action notice mapping model feature attributions to FFIEC reason codes, as required under ECOA and Regulation B. A disparate impact monitoring dashboard tracks approval rate disparity across demographic proxies and alerts you when a threshold is crossed, before the disparity becomes an exam finding. The decision audit log retains every decision at 25 months with the exact model version and policy snapshot active at the time, and a replay API lets you reconstruct any historical decision on demand.

Platform archetypes

Built for the platforms where commerce happens

E-commerce marketplaces

Offer installment credit at checkout. Underwrite sellers for working capital lines using their GMV history, not a bureau snapshot from six months before they started selling on your platform. Price both the buyer BNPL offer and the seller advance from signals your marketplace already holds.

B2B SaaS platforms

Embed net-terms or invoice financing into your billing workflow. Your customers' usage depth, payment tenure, and product engagement are better predictors of default than a business credit report on a two-year-old LLC. Lendforge models both inputs and returns a single underwriting decision.

Gig and creator platforms

Advance earnings to contractors before payout settlement. Traditional employment verification fails here because your workers are self-employed. Lendforge models income predictability from earnings volatility, payout cadence, and platform tenure, signals that a standard credit underwriting model cannot see.

Ready to build a credit program on your own data?

See Pricing Read Developer Docs