Configure and use complex tasks

🚧

Note: This feature is available exclusively for users on Premium and higher subscription plans.

1. Select the agent you want to configure and click on "Agent Settings".

2. Within Agent Settings, locate and click on the "Intelligence" tab.

3. Click on "Complex Tasks" to enable this feature for your agent.

4. Scroll down to the "Generate Responses From" section in the Intelligence tab.
5. Select the preferred response source:

• My Data Only (Recommended): The AI uses only your uploaded content for responses.
• My Data + LLM: The AI combines your data with its general knowledge base to generate more creative and diverse responses.

6. "Select Your AI Model". Choose the model that best fits your agent's needs:

• GPT-4o
• GPT-4 Turbo
• Claude 3.5 Sonnet

7. After configuring the desired options (e.g., enabling Complex Tasks, customizing responses, and selecting the AI model), scroll to the bottom of the page. Click on the "Save Settings" button to apply your changes.

How Complex Tasks Works

• The agent breaks down user queries into five sub-queries, isolating key components to better understand the user's intent.
• A vector database is queried for each sub-query individually to retrieve the most relevant data.
• The retrieved contexts are merged into one, providing comprehensive insights.
• This process adds a slight latency of 1-2 seconds due to the enhanced processing.

When to Use:

• Best suited for complex, multi-dimensional queries where advanced insights are required.
• Ideal for projects where the system needs to anticipate user needs and provide in-depth responses.
• Avoid enabling for simple queries, as it might add unnecessary latency without significant benefits.

Key Benefits of Complex Tasks

• Enhanced Understanding: The agent can handle multi-dimensional queries by breaking them into smaller, manageable parts.
• Improved Response Quality: Merging results from sub-queries ensures the response is thorough and contextually relevant.
• Forward-Thinking AI: Anticipates user needs, delivering insights beyond the immediate query.