Instruct

The Instruct module in the Great Wave AI Studio is designed to provide detailed guidance on how you want your AI agents to operate. Here’s a comprehensive guide on how to navigate and utilize this feature to tailor your AI agents according to specific needs and preferences.

Step 1: Select the Agent Mode

  • Use-Context - The Agent will only use the Index and therefore your context to respond

  • Context-And-LLM - The Agent will use a combination of your context and its own knowledge to respond

  • LLM-Only - The Agent will only use the LLMs knowledge to respond

Step 2: Select Index

  • If using context, select the Index you would like the Agent to use to formulate its response

Step 3: Select the Language Model

  • LLM Selection Dropdown: Begin by choosing the appropriate Language Large Model (LLM) for your agent from the dropdown menu. Options include GPT-3.5 and GPT-4o.

Step 4: Set General Instructions

  • General Instructions Input Box: Use this box to define the broad tasks and behavior expectations for your agent.

Step 4: Advanced Options

  • Prepending Text: Here you can prepend an input with a defined set of text which will be used for retrieval. This is useful if you want the agent to target a specific area of your context.

  • Labeling Context: You have the flexibility to label the information derived from the Grounding feature however you prefer—be it "training data," "source data," or even "my brain." This helps in aligning the agent’s use of this data with how you conceptualize it within your operations.

  • Context Utilization Choice: Finally, decide how your agent should use the context. You can set the agent to rely solely on the Grounding context or allow it to supplement this with its built-in knowledge base. This setting is crucial for determining the balance between using curated data and the agent's own generative capabilities.

Save Finalizing Settings

  • Refresh Agent: After configuring all the instructions and settings, save your changes and Refresh Agent and conduct tests to see how the agent performs under various scenarios. This testing phase is vital for ensuring the agent operates as intended and meets your operational requirements.

Last updated