docs: add governed OpenAI-compatible endpoint guide

docs/_advanced/governed-endpoints.md

@@ -0,0 +1,79 @@
+---
+layout: default
+title: Governed Endpoints
+nav_order: 6
+description: Route RubyLLM through OpenAI-compatible gateways for centralized policy, audit, and cost controls.
+---
+
+# {{ page.title }}
+{: .no_toc }
+
+{{ page.description }}
+{: .fs-6 .fw-300 }
+
+## Table of contents
+{: .no_toc .text-delta }
+
+1. TOC
+{:toc}
+
+---
+
+RubyLLM can connect to OpenAI-compatible gateways by setting `openai_api_base`. This is useful when a Rails application needs a central control plane for policy, audit trails, model routing, quotas, or cost controls while keeping the same RubyLLM API in application code.
+
+The gateway owns model access and operating controls. RubyLLM still owns the Ruby interface for chat, tools, embeddings, streaming, and Rails integration.
+
+## Example: Tuning Engines
+
+Tuning Engines exposes an OpenAI-compatible endpoint for routing model calls through centralized compliance, control, and cost policies.
+
+```ruby
+RubyLLM.configure do |config|
+  config.openai_api_key = ENV["TUNING_ENGINES_API_KEY"]
+  config.openai_api_base = "https://api.tuningengines.com/v1"
+end
+
+chat = RubyLLM.chat(
+  model: ENV.fetch("TUNING_ENGINES_MODEL", "your-model-alias"),
+  provider: :openai,
+  assume_model_exists: true
+)
+
+chat.ask("Summarize the production controls on this AI request.")
+```
+
+Use a model alias that is enabled in your Tuning Engines catalog. The API key should be an inference key, not a provider secret.
+
+## Rails configuration
+
+For Rails applications, keep the endpoint and key in credentials or environment variables:
+
+```ruby
+# config/initializers/ruby_llm.rb
+RubyLLM.configure do |config|
+  config.openai_api_key = Rails.application.credentials.tuning_engines_api_key || ENV["TUNING_ENGINES_API_KEY"]
+  config.openai_api_base = ENV.fetch("TUNING_ENGINES_API_BASE", "https://api.tuningengines.com/v1")
+end
+```
+
+Then pass your governed model alias when creating chats:
+
+```ruby
+RubyLLM.chat(
+  model: "support-triage-fast",
+  provider: :openai,
+  assume_model_exists: true
+)
+```
+
+## When to use this pattern
+
+Use a governed endpoint when you want one or more of these controls outside individual application code paths:
+
+*   tenant, user, or role-based model access
+*   policy and guardrail decisions before inference
+*   model routing, fallbacks, quotas, or budget ceilings
+*   centralized traces, audit logs, and cost reporting
+*   consistent model aliases across multiple Rails applications
+
+For local servers, LiteLLM, vLLM, Ollama, and other OpenAI-compatible endpoints, use the same `openai_api_base` pattern shown in the [Configuration Guide]({% link _getting_started/configuration.md %}#openai-compatible-apis).