Getting Started
Use Contact Engine when your product needs more than a single model response. Engine can orchestrate multiple models, compare outputs, verify claims, and return a synthesized result through one API.
Authentication
Authenticate requests with a Contact Engine API key. Keep keys server-side and never expose them in client-side applications.
Quickstart
A minimal request can start with a verified answer. The exact model routing and verification depth are determined by the operation and your configured usage.
API Reference
Engine exposes focused intelligence operations so applications can choose the right depth and cost for each request.
Models
Engine can route work across supported frontier and specialist models. Model selection is an implementation detail when your application requests an intelligence operation rather than a specific provider.
Verification
Verification compares candidate outputs, identifies disagreement, checks evidence where available, and produces a confidence-aware synthesized result.
Research
Research is intended for complex questions requiring deeper analysis, multiple passes, and broader evidence gathering.
Synthesis
Synthesis combines multiple model outputs or source materials into one structured result while preserving relevant distinctions and uncertainty.
Credits
Contact Credits are the usage unit across Contact AI products. Operations consume credits according to intelligence depth and workload.
Rate Limits
Limits depend on account tier, endpoint, and usage profile. Higher-volume applications can request expanded capacity.
Errors
API errors use stable machine-readable codes with human-readable messages. Applications should handle authentication, validation, rate-limit, and provider failures explicitly.
SDKs
Official SDKs can provide authentication helpers, typed requests, retries, and response handling for supported languages.