Skip to main content
Secondary Model Routing adds a secondary model that comes into play when the primary can’t respond. Built on the Model Library, it serves two purposes at once: business continuity and data security. The routing decision happens automatically and transparently, based on the primary–secondary mapping defined in the Model Library — no extra configuration or code changes on your side.

​
Failover on error

When an error condition is detected on the primary model call — a timeout, service outage, provider-side error (5xx), or rate limit — the system automatically routes the request to the predefined secondary model. A temporary or permanent outage at a single model provider no longer stops the end-to-end service.

​
Local routing on sensitive-data detection

The same mechanism protects sensitive data. When PII or PHI is detected in a request or an output, the system can automatically route the request to a secondary local model instead of the external/cloud primary:
  • Sensitive data is processed by a model hosted within your organization’s boundaries, never transmitted to a third-party provider.
  • This prevents data-leakage risk and adds an assurance layer for compliance with KVKK / GDPR and similar regulations.
One feature, two benefits: uninterrupted service during a model’s outage, and in-house processing of sensitive data when PII/PHI is detected — addressing continuity and compliance at the same time.

Model Library

Define the primary–secondary model mapping.

Data Protection

The PII/PHI detection that triggers local routing.

High Availability

Infrastructure-level resilience.

Compliance

Data-residency and regulatory alignment.