Your policy on top. Our score as the input.
The engine returns a composite score and a recommendation tier. What each tier means for your product - the rate, the limit, the escalation path, the manual review depth - stays entirely inside your loan-decisioning rules.
- Rule engines supportedJSON, DSL, UI
- Rule versions retainedFull history
- Approve / Review / RiskConfigurable
Score first. Rules second. Human review always.
The composite score is deterministic. Your rules are the discretionary layer that turns a score into a decision. Kreedite never replaces your credit policy - it makes it faster to run and easier to defend.
- 01Composite score arrives
Every application returns a 0-100 composite with full per-signal attribution.
- 02Tier assignment
The score is mapped to one of three tiers - approve for review, manual review advised, or high-risk route - using thresholds your team sets per product.
- 03Product rules apply
Interest band, exposure limit, tenor cap, collateral requirement and escalation path are chosen by your rule set - not by the engine.
- 04Human review
A queued underwriter completes the decision. The engine's job ends at recommendation - the lending decision is always yours.
Recommend advancing to underwriter with light review depth. Typical low-friction path for consistent alt-data borrowers.
Routed to full manual review. Reason codes highlight where the signal is weak so the underwriter can focus.
Flagged for a stricter decisioning path. Not an automatic decline - the lending decision remains with your team.
- The engine never auto-approves and never auto-declines - the recommendation is always advisory.
- Rules are versioned. A past decision can be replayed against the rule version that produced it.
- High-risk tier routes to stricter review, not to an automatic no.
Three surfaces, one rule store.
Configure decisioning rules the way that matches your team - through a UI for policy owners, a domain-specific language for underwriting, or JSON for programmatic deployment.
Threshold sliders, tier mappings and escalation paths - built for a credit policy owner who does not want to touch code.
A readable domain-specific language for underwriting teams. Diffs cleanly in version control, reviews cleanly in change management.
Programmatic rule deployment for lending institutions running full CI/CD around credit policy - safest path for institutions with tight change control.
- Every rule change is audited with actor, timestamp and diff.
- A live rule version is always associated with every scored decision.
- Simulations can be run against historical applications before promoting a new rule set.
- Roll-back to a prior rule version is a single operation, not a redeployment.