AI Provider Configuration

Configure connections to AI services for content generation, including API keys, model selection, and rate limiting.

Overview

AI Providers are configured through the ai.provider model. Each provider represents a connection to an AI service with associated credentials and settings.

OpenAI Setup

OpenAI is the default and recommended provider for the AI Workflow Framework.

Step 1: Get API Key

1. Visit platform.openai.com
2. Sign in or create an account
3. Navigate to API Keys section
4. Create a new secret key
5. Copy the key (it won't be shown again)

Step 2: Configure in RebusAI

# Via RebusAI shell
provider = env['ai.provider'].create({
    'name': 'OpenAI GPT-4',
    'provider_type': 'openai',
    'api_key': 'sk-proj-xxxxx',  # Your API key
    'model': 'gpt-4-turbo-preview',
    'active': True,
})

# Or update existing provider
provider = env.ref('ai_project_workflow.openai_provider')
provider.api_key = 'sk-proj-xxxxx'

Step 3: Verify Connection

# Test the provider connection
provider = env['ai.provider'].search([('provider_type', '=', 'openai')], limit=1)
result = provider.test_connection()
print(result)  # Should return True if successful

Azure OpenAI

For enterprise deployments, Azure OpenAI provides additional security and compliance features.

provider = env['ai.provider'].create({
    'name': 'Azure OpenAI',
    'provider_type': 'azure_openai',
    'api_key': 'your-azure-api-key',
    'api_base': 'https://your-resource.openai.azure.com/',
    'api_version': '2024-02-01',
    'deployment_name': 'gpt-4-deployment',
    'active': True,
})

Anthropic Claude

Anthropic's Claude models can be used as an alternative to OpenAI.

provider = env['ai.provider'].create({
    'name': 'Anthropic Claude',
    'provider_type': 'anthropic',
    'api_key': 'sk-ant-xxxxx',
    'model': 'claude-3-opus-20240229',
    'active': True,
})

Model Selection

Different models offer varying capabilities, costs, and performance characteristics:

Rate Limiting

Configure rate limits to stay within API quotas and manage costs:

provider.write({
    'rate_limit_rpm': 60,       # Requests per minute
    'rate_limit_tpm': 150000,   # Tokens per minute
    'max_retries': 3,           # Retry on rate limit
    'retry_delay': 5,           # Seconds between retries
})