Skip to main content

🛠️ Settings

🌟 Introduction​

Configure your AI agent effectively by following these steps:

  1. Provide a detailed prompt for generating output data from your input dataset.
  2. Define clear evaluation criteria for judging the quality of generated outputs.
  3. Select the relevant input columns for your agent.

Task Description​

Describe the task that the agent needs to accomplish. This should be a clear and concise statement of the objective.

Example:

Based on user task, get a list of properties based on location, choose the top 2 most expensive ones, and send the result via email to show the address, price, and highlighted image.

Input Prompt Template​

Detail how the agent should generate output data from the input dataset. Use specific instructions and include any necessary variables.

Example:

Find the top 2 most expensive houses in Palo Alto, CA, and send the result to email charisliao@berkeley.edu. Ensure the email is sent, and the image is included as a circle thumbnail, not a link.

Reference Column​

Select the data column that will be used as the reference for evaluation. This helps in assessing the accuracy and relevance of the generated outputs.

Example:

price

Result Format​

Define the standard options for output data. If there are no restrictions, this can be left blank. Ensure that the format aligns with the expected output.

Example:

3750000

Since our example is a price, the result format is a number.

Tools​

Select the tools that can be invoked when the multi-agent is running. This allows for additional functionalities to be integrated into the agent's workflow.

Example:

find_properties_list, send_email

Knowledge​

Select the knowledge resources that can be accessed when the multi-agent is running. This ensures the agent has the necessary information to perform its tasks effectively.

Best Practices 🎯​

  • Be Clear & Specific: Clearly state what you want the agent to accomplish.
  • Use Data Smartly: Ensure all necessary variables are included and correctly formatted.
  • Keep Scoring Simple: Use straightforward evaluation criteria to assess output quality.

Common Issues & Quick Fixes 🔧​

  1. Bad Outputs? → Add example formats in the prompt.
  2. Poor Scoring? → Use numerical scores instead of words.
  3. Missing Info? → Ensure all needed variables are included in the prompt.

Need help? Check our example templates or contact support!