We don’t replace lenders. We help them see.
AfriScore is a B2B credit-scoring engine. It turns mobile money, utility and airtime data into explainable scores for people banks can’t read. The lender keeps the decision and the risk.
Kenya pilot with Jiweze Improved Productivity Ltd
2025 production pilot in Kiambu County, run alongside the lender’s existing system over 60 days.
previously unbanked people approved
- loan applications scored
- average processing time
- auto-approval rate
- observed default rate (60-day pilot)
- infrastructure uptime
Figures from the AfriScore × Jiweze pilot report, 2025. Accuracy compares predictions with Jiweze’s historical approvals.
From a working engine to a lending layer
Built
A Python ML pipeline with calibrated scoring and SHAP explanations, trained on synthetic data from real mobile money distributions.
Proven
A 60-day production pilot in Kenya, run next to Jiweze’s own process. 700+ unbanked people approved at an average loan of about US$40.
Next: CEMAC
In discussions with a credit union in Cameroon. Real-data calibration comes after the first partner signs.
Then
UEMOA and the wider continent, one calibrated market at a time.
How a score gets made
- 1
Data in
Mobile money, utility payments and airtime behaviour, through one API.
- 2
Score out
A calibrated score, risk tier and default probability.
- 3
Reasons attached
Every score lists what pushed it up or down.
- 4
Lender decides
Your policy, your pricing, your balance sheet.
Where do you want to start?
I lend
Run AfriScore beside your current process at no cost and see how it would have performed on your own portfolio.
Apply for the pilotI want to see my own profile
Enter transaction behaviour and see exactly what drives the result.
Run the demoI build products
Read the endpoints and try a request from the browser.
Open the API referenceSee what it could do for your portfolio
Approval Optimiser
Simulate your uplift
Set your lender profile to tailor the estimation to your scenario.
ROI Calculator
Measure your uplift
Estimates based on industry benchmarks for alternative-data credit scoring. Actual results may vary based on portfolio composition.
Built for regulated environments from day one.
Data sovereignty
Scoring can run within your own infrastructure. No borrower data leaves your environment unless you choose a hosted deployment.
Regulatory alignment
Architecture is designed with CEMAC and regional data protection frameworks in mind from the outset, not retrofitted later.
Explainable by default
Every score ships with feature-level explanations. No black-box decisions — lenders and borrowers can both see why.
Built for every stakeholder
Digital lenders and credit providers
Add explainable scoring through one API. Run the free pilot beside your process.
- A score in well under a second
- Reasons with every score
- Free shadow-mode pilot
- From $0.25 per scoring request
Banks, MFIs and credit unions
Run it beside your credit policy, with evidence your committee can review.
- An auditable, stress-tested model
- Hosted or on your own servers
- Your committee keeps every decision
- Free shadow-mode pilot
Individuals
See what drives a credit profile, in plain language.
- A score in under a second
- Reasons you can read
- What to change to improve it
- No bureau file needed
Developers
Add scoring through the API. Try it in the console before you write any code.
- Endpoints and a try-it console
- Integration guides and field reference
- Hosted API or Docker
- Pay per score
Where AfriScore works
Cameroon
Home baseWhere we are based, and the CEMAC market we design for.
Kenya
Pilot marketA 60-day pilot run beside a lender’s own process.
Building or lending in an underbanked market?
We’re accepting a limited number of pilot institutions. No cost, no commitment — just a real evaluation against your own portfolio.