Add via easy integration

Add a Model via Easy Integration

Connect models from OpenAI, Anthropic, Google, Cohere, or Amazon Bedrock using guided setup.

---

Before You Begin

• Obtain an API key from your provider
• For Amazon Bedrock: Set up IAM role with appropriate permissions (see Configuring Amazon Bedrock)

---

Add a Model (Standard Providers)

Use this method for OpenAI, Anthropic, Google, or Cohere.

1. Go to Models → External Models → Add a model
2. Select Easy Integration and click Next
3. Choose your provider and click Next
4. Select a model from the supported list
5. Enter a Connection name and your API key
6. Click Confirm

The model appears in the External Models list and is ready to use.

After Integration

• Enable/Disable: Use the Inference toggle to control model availability
• Edit: Click the three-dot menu to update credentials
• Delete: Remove the model via the three-dot menu

---

Add Amazon Bedrock Models

Amazon Bedrock requires IAM role-based authentication instead of API keys.

Prerequisites

Create an IAM role in your AWS account that:

• Grants permission to invoke Amazon Bedrock models
• Includes a trust policy allowing Agent Platform to assume the role

For detailed IAM setup, see Configuring Amazon Bedrock.

Steps

1. Go to Models → External Models → Add a model

2. Select Easy Integration → AWS Bedrock → Next

3. Configure Credentials:

   Field	Description
   IAM Role ARN	Full ARN of your IAM role with Bedrock permissions
   Trusted Principal ARN	Pre-populated; Platform’s AWS principal (read-only)

4. Configure Model Details:

   Field	Description
   Model Name	Internal name for identification
   Model ID	Bedrock Model ID or Endpoint ID
   Region	AWS region where the model is deployed
   Headers	Optional custom headers for requests

5. Configure Model Settings — Choose one option:

   Option A: Default (Manual Configuration)

   • Define prompt variables and custom variables
   • Configure request body with variable placeholders ({{prompt}})
   • Test the response and map JSON paths for output extraction

   Option B: Existing Provider Structures (Recommended)

   • Select a provider template (OpenAI Chat Completions or Anthropic Messages)
   • Enable supported features (Tool calling, Streaming, Structured response, etc.)

6. Click Confirm to save, or Save as draft to configure later

---

Model Configuration Options

When adding Bedrock models or using advanced configuration, you can choose between two setup modes:

Default Mode

Manually configure all API components:

Setting	Description
Variables	Define prompt variables and custom variables for dynamic binding
Request Body	JSON payload with {{variable}} placeholders
JSON Path Mapping	Extract output, input tokens, and output tokens from response

Example Request Body:

{
  "prompt": "{{prompt}}",
  "max_tokens": 200,
  "temperature": 0.7
}

JSON Path Mapping:

• Output path: choices[0].message.content
• Input tokens: usage.prompt_tokens
• Output tokens: usage.completion_tokens

Existing Provider Structures Mode

Automatically apply pre-defined schemas from known providers:

Provider Template	API Reference
OpenAI (Chat Completions)	Standard OpenAI chat format
Anthropic (Messages)	Standard Anthropic messages format
Google (Gemini)	Standard Gemini format

Available Features (enable as supported by your model):

• Structured response
• Tool calling
• Parallel tool calling
• Streaming
• Data generation
• Modalities (Text-to-Text, Text-to-Image, Image-to-Text, Audio-to-Text)

---

Related

• External Models Overview
• Add via API Integration
• Configuring Amazon Bedrock
• Supported Models