AI-Powered Credit Infrastructure for African Lenders
Score before
you lend.
Wealthstein turns real credit, financial, and identity behavior into one persistent credit score per credit account — delivered through an API, not a bureau report. You keep the borrower relationship. We do the underwriting math.
500 neutral score
5 weighted factors
$0.28/mo + $1/check
How It Flows
Data in, a decision-ready score out.
One Identity, Not One Profile Per Bank
An account number is a credit account. Not a login — an identity.
Wealthstein doesn't score a bank relationship. It scores a person, persistently, across every lender that ever touches that account number — updated as new data arrives, never static.
01
Access Bank
Kemi applies for a loan
Access Bank sends her permitted data to Wealthstein. A Credit Identity is created for her account number, and a score is returned — just to Access Bank.
02
GTBank
Months later, a second relationship
Kemi opens an account at GTBank. That data doesn't start a new profile — it updates the same Kemi identity Access Bank already contributed to.
03
Wema Bank
A new lender, an existing identity
Kemi applies to Wema. Wema doesn't see Access Bank's score or GTBank's score — it requests Kemi's current Wealthstein Score, built from everything permitted so far.
Kemi's account number never has three separate scores — it has one Wealthstein Score, three lenders permitted to see it (each under its own consent), and a score that kept improving as more of her real financial behavior came in. That's the difference between a bureau report and a credit network.
API Reference
Call it, get a result back.
Four endpoints cover the whole flow — onboard, connect, score, decide. Every response is what you see below, no separate docs portal required to understand the shape.
— Onboard a customer by phone number
Call
curl -X POST https://api.wealthstein.com/v1/customers/quick-onboard \
-H "Authorization: Bearer sk_live_..." \
-H "Content-Type: application/json" \
-d '{ "phone": "+2348012345678" }'Result
{
"customerId": "cus_8f2a91c3",
"existing": false,
"status": "onboarded"
}What you're actually up against
What gets in the way, and what Wealthstein does about it
The problem
Most applicants have no formal credit file to underwrite against.
How Wealthstein solves it
Score off real credit history and financial behavior instead — data your applicants already generate every day.
The problem
Building a credit model in-house means hiring a data science team you don't have yet.
How Wealthstein solves it
Plug into a scoring engine that's already built, weighted, and explainable — live in an API call, not a hiring cycle.
The problem
Bureau reports are static snapshots that go stale the moment they're pulled.
How Wealthstein solves it
Scores refresh as new data arrives, so a borrower's risk picture stays current across the life of the loan.
The problem
A single risk model rarely fits every product on your book.
How Wealthstein solves it
Set your own minimum score, amount bands, and refer thresholds per product — a payday line and an SME facility can read the same score differently.
The problem
Sharing customer data across systems without clear consent is a real liability.
How Wealthstein solves it
Every score request is gated by a named, revocable consent record — logged and auditable from day one.
Why Wealthstein
Built to get you underwriting faster, not slower.
No bureau, no gatekeeping
Score real payment behavior directly — no waiting on a third-party bureau's coverage of your applicant.
Explainable AI, not a black box
Every score ships with the AI-weighted factors behind it, ready for your own adverse-action notices.
Your rules stay yours
Thresholds and amount bands are set per product, per lender — never shared, never overridden.
Pay only for what scores
No platform fee to start. You're billed per connection, per score, per decision — nothing upfront.
Product
Built for a risk team, not a data science team.
Account-based onboarding
Your customer's account number is the only field you need to start. An existing identity gets a one-tap consent reuse instead of a fresh data pull.
Explainable by default
Every score ships with a per-factor breakdown — the same reasons your team needs for adverse-action notices, not a black-box number.
Your rules, not ours
Set minimum score, amount bands, and refer thresholds per product. A payday product and an SME facility can read the same score differently.
Consent-gated, always
No score reaches a lender without an active, named, revocable consent record. Every access is logged and auditable.
Usage-based billing
Pay per connection, per score, per decision — plus an optional monitoring fee for portfolio-wide score refreshes on your existing book.
Sandbox-to-live parity
Build and test against synthetic data today. Swapping in a live provider integration changes the data source, not your integration.
How it works
Four steps, one API.
01
Onboard by account number
Submit a customer's account number. A new identity starts a data connection; an existing one gets a one-tap consent reuse instead.
02
Connect consented data sources
Financial, repayment, and identity signals stream in under an explicit, named, revocable consent record.
03
Score is computed
Weighted factors — credit history, financial behaviour, income stability, identity & account stability, risk & fraud behaviour — produce a score from a neutral 500 start, with a full explanation attached.
04
Your product rules decide
The score runs against your own product's thresholds and amount bands. Approve, decline, or refer — your call, every time.
A Day in the Life
How Metro Trust Bank underwrites, before and after.
A fictional bank, a realistic workflow: approving loan applications and tracking repayments, with and without Wealthstein underneath.
Without Wealthstein
01
Manual document collection
Metro Trust Bank collects payslips and bank statements by hand, then waits on a bureau check that may return nothing for a thin-file applicant.
02
Judgment-call approvals
Loan officers approve or decline based on incomplete data and gut feel — inconsistent across officers and branches.
03
Spreadsheet repayment tracking
Repayment status lives in spreadsheets, updated manually. Missed payments surface late, if at all.
With Wealthstein
01
Instant score at application
One API call returns a score, band, and confidence level — built from real payment behavior, in seconds.
02
Automatic, consistent decisions
Approve, decline, or refer — decided against Metro Trust Bank's own product rules, the same way every time.
03
Real-time repayment monitoring
Scores refresh as new data arrives. A missed payment triggers a webhook, not a quarterly spreadsheet review.
Wealthstein vs. building it yourself
Skip the build, keep the control.
Where Wealthstein sits
You hold the lending license and the borrower relationship. Wealthstein is the AI-driven scoring engine underneath it.
That's deliberate: your team already carries the compliance and consumer-protection obligations that come with lending. We don't duplicate that — we give you a faster, better-evidenced number to lend against.
Pricing
One onboarding fee, then two predictable numbers.
Borrowers never pay. A one-time onboarding fee gets you live; after that you're billed only for what you actually create and check.
Onboarding
$4,500 one-time
Covers integration support, sandbox setup, and your pilot walkthrough. Billed once, when you go live — not per pilot.
Subscription
$0.28 / credit account / mo
A credit account is one lender's link to one borrower — not the borrower themselves. If two lenders connect to the same person, that's two accounts and two $0.28/month charges, even though both are reading the same shared score. You're paying to maintain your own connection, not to compute the score again.
Usage
$1.00 / score check
Charged each time you request a score — at origination, on a refresh, or when a returning lender checks an existing identity through consent reuse. Flat rate, every confidence tier.
Running high account volume? Ask about volume-based rates.
FAQ
Frequently asked questions
They start at a neutral 500 score — no evidence either way, not assumed good or bad. As sources connect and history accumulates, the score moves up or down from there.
Each borrower has one persistent credit identity, not a separate profile per lender. But no lender sees another lender's data or a shared score without that customer's explicit, per-lender consent — an account match alone never grants access.
A one-time $4,500 onboarding fee gets you live, then two ongoing lines: $0.28/month per credit account — that's per lender-to-borrower link, not per borrower, so two lenders connected to the same person means two charges — plus $1 per score check. Borrowers never pay.
Most teams are hitting the sandbox API within a day. A pilot with your own applicant sample typically runs a few weeks before a go-live decision.
Now Piloting
Bring your own thin-file portfolio.
Score applicants a bureau can't reach. We're running a hands-on pilot with our next 20 lenders — bring a real batch of applicants, and we'll show you the scores against your outcomes before you commit anything.
Ready to Get Started?
Request a pilot, or bring your risk team a question first.