This is a Shadow/Demo Mode. All data is simulated for evaluation only.
PAYKX
Deterministic safety layer for payments
A pre-execution GO / DEGRADED / NO-GO verification layer that helps both human operators and autonomous AI agents make safer payment decisions — built for high-friction corridors and agentic flows.
Same inputs always return the same GO / NO-GO. Auditable, rule-based scoring — no black-box model in the decision path.
Idempotent, safe retries
An idempotency_key makes agent retries safe: a repeated call returns the original result instead of acting twice. Mismatched reuse is rejected.
Fails closed
No key, an unknown corridor, or thin data never silently passes. Unsupported corridors return DEGRADED — the system says when it doesn't know.
Honest by design
What it is
A verification / risk layer agents call before paying
Deterministic, rule-based scoring from your inputs
Live for the UK ↔ Nigeria corridor
What it isn't (yet)
Not connected to live rails — shadow mode on simulated data
No ML model and no on-chain settlement in the decision path
Not a payment processor — it decides, it doesn't move funds
ValidatedFCA Digital Sandbox
Tested on synthetic transaction data — April 2026.
April 2026
Sandbox Run
71%
DEGRADED
29%
GO
✓Live API aligned with sandbox behaviour
BenchmarkReproducible Engine Benchmark
Scored 750,000 synthetic SEPA-style transactions using our production scoring engine. Results are fully reproducible via committed code in the repository.
63.1%
GO
36.2%
DEGRADED
0.7%
NO-GO
0.83
Avg Score
Results generated using committed code in the repository (seed: 42).
Ready for Production Access?
Unlock higher limits, production support, and priority assistance with the authenticated API.