Ubuntu Score

Inspiration

Imagine contributing faithfully to a savings circle for ten years, never missing a payment, yet being told you have no credit history.

Across Africa, millions of people demonstrate financial discipline through stokvels, savings circles, and community lending groups. Month after month they save, contribute, and build trust within their communities. Yet when they apply for formal credit, that history is often invisible.

We asked a simple question:

What if community trust could become financial identity?

Ubuntu Score was born from that idea.

Rather than replacing traditional credit systems, Ubuntu Score helps bridge the gap between informal community finance and the formal financial sector by transforming verified community savings behaviour into transparent, explainable evidence that lenders can understand, while ensuring individuals remain in control of when and with whom their financial identity is shared.

What We Built

Ubuntu Score is an explainable financial identity platform.

It converts verified community savings behaviour into an Ubuntu Score and loan-readiness assessment while ensuring every official outcome remains transparent, explainable, and grounded in verified evidence.

The platform includes:

An explainable deterministic scoring engine that generates an Ubuntu Score from verified contribution history.
Automated processing of CSV, Excel, and WhatsApp-style contribution ledgers.
Human verification before any contribution can influence a member's financial identity.
A Financial Identity workspace showing contribution history, readiness, explanations, audit history, and downloadable reports.
A dynamic Readiness Gauge that visualises financial progress.
A What-if Simulator that allows hypothetical future scenarios without changing official records.
Consent-aware sharing designed for future institutional partnerships.
An advisory machine learning model that forecasts future contribution consistency trends without influencing official scores.

Our MVP currently models 34 members with hundreds of verified monthly contribution records, processing multi-member CSV, Excel, and WhatsApp-style ledgers independently while maintaining complete audit trails for every verified update.

How We Built It

Ubuntu Score combines deterministic decision logic with responsible AI.

The platform was developed using Next.js, React, TypeScript, Tailwind CSS, and Python.

At its core is an explainable scoring engine that evaluates:

Contribution consistency
Payment completion
Membership duration
Recent savings behaviour
Verification confidence

These verified signals produce an Ubuntu Score between 0 and 850, together with an explainable readiness assessment.

Community ledgers uploaded as CSV, Excel, or WhatsApp-style exports are automatically converted into structured contribution records. Every extracted record enters a human verification workflow before it can update a member's contribution history, score, readiness, timeline, audit log, or report.

To complement this deterministic engine, we trained an advisory Random Forest model using scikit-learn. We selected Random Forest because it performs well on tabular behavioural data while providing feature importance insights that help us understand which behavioural signals most influence its forecasts.

The model is trained on reproducible synthetic behavioural data and forecasts whether contribution consistency is likely to improve, remain stable, or decline using signals such as payment gaps, contribution volatility, contribution amount trends, and partial-payment patterns.

The machine learning model is intentionally advisory.

It never determines a member's Ubuntu Score, changes a readiness band, or replaces human judgement.

Challenges

The hardest challenge was not building AI, it was deciding where AI should stop. Financial identities affect real opportunities. We believed no automated system should change someone's financial profile without accountability. That led us to design a human-in-the-loop workflow where automation accelerates processing, but verified evidence and human review remain responsible for every official outcome.

Another challenge was explainability. Many credit models cannot clearly explain why a decision was made.Ubuntu Score takes the opposite approach. Every score can be traced back to verified monthly contributions, transparent calculation rules, audit logs, and supporting evidence.

Finally, we continually refined the user experience so the application feels like a professional financial platform rather than a technology demonstration.

What We Learned

Building Ubuntu Score reinforced an important lesson:

Responsible AI is not about replacing human judgement. It is about making human decisions better informed, more transparent, and more accountable.

We learned that trust in financial technology comes from explainability, consent, governance, and evidence just as much as it comes from technical accuracy.

Perhaps our biggest lesson was this:

Technology should recognise the trust that communities have already built, not replace it.

What's Next

Today, Ubuntu Score demonstrates how verified community savings can become an explainable financial identity.

Tomorrow, we see it becoming trusted infrastructure connecting community savings groups, SACCOs, microfinance institutions, banks, and fintech platforms across Africa.