From d9d037b9d070d0ecd20d67edfc5ea19169757c2d Mon Sep 17 00:00:00 2001
From: VC <bazooka720@gmail.com>
Date: Sat, 30 May 2026 10:47:52 -0400
Subject: [PATCH] Document Tuning Engines provider config

---
 README.md         |  8 ++++++++
 docs/providers.md | 21 +++++++++++++++++++++
 2 files changed, 29 insertions(+)

diff --git a/README.md b/README.md
index 78bafbc9..74e2a0bd 100644
--- a/README.md
+++ b/README.md
@@ -123,10 +123,18 @@ development:
     service: "Ollama"
     model: "llama3.2"
 
+  tuning_engines:
+    service: "OpenAI"
+    access_token: <%= Rails.application.credentials.dig(:tuning_engines, :api_key) %>
+    base_url: "https://api.tuningengines.com/v1"
+    model: "gpt-4o-mini"
+
   ruby_llm:
     service: "RubyLLM"
 ```
 
+`tuning_engines` is an OpenAI-compatible gateway configuration for Rails teams that want centralized model routing, policy controls, audit logs, traces, approvals, and cost visibility without changing ActiveAgent code.
+
 ## Features
 
 - **Agent-Oriented Programming**: Build AI applications using familiar Rails patterns
diff --git a/docs/providers.md b/docs/providers.md
index add6f66a..db19300d 100644
--- a/docs/providers.md
+++ b/docs/providers.md
@@ -54,6 +54,27 @@ GPT-4o, GPT-4.1, GPT-5, and o3 models. Two APIs available: Responses API (defaul
 
 **Choose when:** You need reliable, high-quality responses with strong reasoning. Vision support and structured output work well. Azure OpenAI compatible.
 
+#### OpenAI-compatible gateways
+
+The OpenAI provider also accepts a custom `base_url`, so Rails apps can route ActiveAgent generations through a governed OpenAI-compatible endpoint. For example, [Tuning Engines](https://www.tuningengines.com/) can sit underneath ActiveAgent to centralize model routing, policy controls, audit logs, traces, approvals, and cost visibility:
+
+```yaml
+production:
+  tuning_engines:
+    service: "OpenAI"
+    access_token: <%= Rails.application.credentials.dig(:tuning_engines, :api_key) %>
+    base_url: "https://api.tuningengines.com/v1"
+    model: "gpt-4o-mini"
+```
+
+Then use it like any other generation provider:
+
+```ruby
+class SupportAgent < ApplicationAgent
+  generate_with :tuning_engines, model: "gpt-4o-mini"
+end
+```
+
 ### [OpenRouter](/providers/open_router)
 **Best for:** Multi-model flexibility, cost optimization, experimentation