Skip to content

Scoring and decisioning engine

Sentinel

Policy rules plus model scores, combined into one explainable credit decision for every application that enters the rail.

What it does

01

Policy first, model second

Hard rules from your credit policy run before any model: sector exclusions, exposure caps, document completeness. Failures return instantly with the failing rule named.

02

Scores you can defend

Each score ships with its drivers ranked: which signals moved it, in which direction, by how much. The decision log keeps every version.

03

Challenger in shadow

Run a new scorecard against live volume in shadow mode before it makes a single real decision.

Score drivers
Cashflow strength
Bureau history
Card & POS trend
Sentinel · decision detailLIVE
0.71score
Approve

Probability of default 3.1% · suggested limit PKR 4.2M

14 of 14 policy rules passed
Score drivers ranked and logged
Routed to Book A · working capital

Illustrative product view.

In and out

What goes in
Signals from DocuMind (bank, card, POS, e-commerce)
Bureau and registry data
The application package from Lift
Your credit policy, expressed as rules
What comes out
Score and probability of default
Approve, decline, or refer decision
Suggested limit and profit rate
Ranked reason codes for every decision

Where it sits in the stack

Sentinel sits between signal extraction and routing. It consumes DocuMind output and bureau data, and hands a scored, reasoned decision to orchestration. It also runs standalone behind your existing loan origination system.

Security and compliance

Sentinel logs every rule evaluation and score input to an immutable decision trail, retained for the regulator’s review window. Models never train on another institution’s data.

Full security and compliance posture