AI Configurations
Configurations are the core building blocks of ConfigureMyAI. Each configuration defines an AI assistant with specific behavior, knowledge, and capabilities.
What is a Configuration?
A configuration combines an AI model with customization settings to create a specialized AI assistant. Think of it as a "preset" that defines:
- Which AI provider and model to use
- The AI's personality and role (via system prompt)
- Rules and restrictions for responses
- Reference data and training materials
- Model parameters like temperature and max tokens
Creating a Configuration
- From the dashboard, click New Configuration
- Fill in the basic settings:
- Name: A descriptive name (e.g., "Customer Support Bot")
- Description: Optional notes about the configuration's purpose
- Select your AI provider settings:
- API Key: Choose from your saved credentials
- Model: Select from available models for that provider
- Configure behavior (optional but recommended):
- System Prompt: Define the AI's role and personality
- Rules: Set boundaries and restrictions
- Training Data: Provide reference information
- Adjust model parameters:
- Temperature: Control randomness (0 = precise, 2 = creative)
- Max Tokens: Maximum response length
- Click Create Configuration
Configuration Settings
Basic Settings
| Setting | Description |
|---|---|
| Name | Unique name within the workspace |
| Description | Optional description for documentation |
| Active | Toggle to enable/disable the configuration |
AI Provider Settings
| Setting | Description |
|---|---|
| API Key | The credential to use for API requests |
| Model | The specific AI model (e.g., gpt-4o, claude-sonnet-5) |
Model Parameters
| Parameter | Default | Description |
|---|---|---|
| Temperature | 0.7 | Controls randomness. Lower = more focused, higher = more creative (0-2) |
| Max Tokens | 1024 | Maximum tokens in the response (1-128000) |
Advanced Settings
For advanced use cases, you can override the default API endpoint and model name:
- Custom API Endpoint: Override the provider's default endpoint URL
- Custom Model Name: Specify a custom model identifier (e.g., for fine-tuned models)
These settings are found in the "Advanced Settings" section when editing a configuration.
Statistics & Analytics
Each configuration tracks:
- Number of chat sessions
- Total API calls (via endpoints)
- Tokens used
- Last activity timestamp
View statistics on the configuration's detail page in the sidebar.