Skip to main content
Use this page when you are moving beyond local experimentation into production architecture and rollout.

Main Concerns

ConcernWhy it matters
PerformanceThe in-process design avoids the proxy penalty on deep agent loops
Deployment patternDirect SDK usage, middleware, gateway, or local providers each fit different teams
ReliabilityProduction rollouts need safe defaults, observability, and bounded behavior
OperationsTeams need examples for APIs, telemetry, and deployment setup

Primary Docs

Important Examples