Route with policy
Configured provider and region filters narrow the eligible routes before scoring. Routing explanations make the decision inspectable.
Products / AI inference
Make the routing
decision explicit.
An Erlang/OTP AI inference switch. Route OpenAI-compatible requests across models and providers using policy, endpoint health, cost and availability.
Engine implemented · Hosted service in development
Discuss your use caseThe inference engine
The self-hosted engine implements routing, provider adapters and fault handling. Its development is distinct from the planned hosted service.
Configured provider and region filters narrow the eligible routes before scoring. Routing explanations make the decision inspectable.
Circuit breakers and failover before response commitment support the request lifecycle. Once a response is committed, failure handling follows that boundary.
Usage accounting, metrics and traces support investigation. Provider adapters connect the engine to OpenAI, Anthropic, Gemini and compatible endpoints.
Current status / 16 October 2026
Implemented
The request-routing engine has been developed, with provider adapters, policy filtering, circuit breakers and routing explanations. This is an implementation statement; live-provider interoperability and production capacity require separate validation.
Planned hosted offering
The hosted direction combines managed BYOK, regional operations, a customer dashboard, spend controls and realtime session routing. Hosted availability is not established by the current engine.
Contact us to discuss current scope and access. A public hosted launch, pricing and service commitments are not offered on this page.
Product enquiries
Talk through fit, current scope
and development status.
Tell us about your use case and the capabilities you need. Product access, support and commercial arrangements are discussed against the current stage of development.
hello@eo.io +44 7846 559692