Real-time scoring
Every request to the /v1/score endpoint returns a decision and its drivers in milliseconds. Ready for origination flows.
CRI is a credit + churn risk API that scores any customer in real time and returns an explainable decision — APPROVE, REVIEW or REJECT — ready to wire into your loan origination, CRM, or core banking flow.
Every request to the /v1/score endpoint returns a decision and its drivers in milliseconds. Ready for origination flows.
The engine resolves APPROVE and REJECT on its own. Only the ~30% borderline cases reach a human analyst for manual review.
Every decision ships with the drivers that caused it. Audit, compliance, and analyst onboarding in a single response.
A single hybrid score combines default risk and churn risk. Approve customers who pay and who stay.
Your system (CRM, core banking, loan originator) sends a JSON payload with the customer's data. The API responds in milliseconds.
XGBoost scores default (0.70) and churn (0.30). SHAP extracts the drivers.
A JSON response with the decision, the score, and the top drivers that explain it.
Click Score new to watch the full flow: a request comes in, the models run, and the decision lands in the table with its explainable drivers.
| Customer | Credit Risk | Churn Risk | Hybrid Score | Decision |
|---|
No SDKs, no new infrastructure. Your system sends the customer's data and gets back the decision and its drivers. It fits any flow: loan origination, account opening, portfolio scoring, collections.
// illustrative — reference architecture, not a public endpoint // POST https://api.example.com/v1/score { "customer_id": "CUST-0042", "features": { "income_monthly": 85000, "debt_ratio": 0.34, "employment_years": 3.2, "credit_history_years": 5, "product_usage_score": 0.72, "complaints_last_90d": 0, "tenure_months": 48 }, "locale": "es" }
// 200 OK · 42ms { "customer_id": "CUST-0042", "decision": "APPROVE", "hybrid_score": 0.21, "credit_risk": 0.18, "churn_risk": 0.28, "drivers": { "credit": [ { "feature": "Historial limpio", "impact": -0.13 }, { "feature": "Ingreso estable", "impact": -0.09 } ], "churn": [ { "feature": "Antigüedad: 4 años", "impact": -0.07 } ] }, "model_versions": { "credit": "v1.0", "churn": "v1.0" } }
# illustrative — reference architecture, not a public endpoint curl "https://api.example.com/v1/score" \ -H "Authorization: Bearer $NAPOLI_API_KEY" \ -H "Content-Type: application/json" \ -d '{ "customer_id": "CUST-0042", "features": { "income_monthly": 85000, "debt_ratio": 0.34 } }'
We stand up a sandbox in your environment, tune the models on your historical data, and you start a pilot in weeks — not months.