Navigating to Your AI Agent

1. Access the Agents Page:

   • From the Main Navigation bar, click on "Marketing Apps".
   • Select "Agents" to open the AI Agents dashboard.

2. Find Your Agent:

   • Use the Favorites section for quick access.
   • Page through the list or use the Search Agents field.
   • Apply filters to view only your Agents or filter by permission access.

3. Select an Agent:

   • Click on the desired AI Agent from the list.
   • Once selected, you can begin interacting with the Agent using its capabilities.

📘

All conversations with an AI Agent are unique to each user. You can interact with an Agent that a peer also uses, but you will not see each other’s conversations.

Best Practices for Interacting with AI Agents

1. Ask Clear Questions

Be specific about your request based on the Agent’s function. Example queries:

• “Identify the top 5 campaigns with the highest ROI in the last 30 days.”
• “Summarize ad performance data by channel for this month.”
• “Tell me what budget adjustments you’d recommend for Coca Cola’s campaigns based on the last 30 days.”

2. Understand the Output

Review the response carefully. If the output is incorrect or incomplete:

• Ensure your question is clear and precise.
• Confirm that the Dataset provides the necessary context.
• Refine the Agent’s prompt if needed.

📘

Iteration is normal—adjust the Agent’s prompt multiple times to refine its output.

3. Data Limitations

• AI Agents can analyze over 100,000 rows but can only return up to 100,000 rows.
• If the response is truncated, adjust your prompt to narrow the scope.
• If an Agent attempts to return over 100,000 rows, it may display null values or incomplete results.
• Look for an expandable link: “100,000 Results collected from [Dataset]” to confirm data retrieval limits.

4. Take Action Based on Insights

• Use the Agent’s insights to make data-driven decisions.
• Adjust advertising budgets, campaign strategies, or reporting based on AI-generated recommendations.

5. Provide Feedback for Improvement

• AI Agents improve through continuous iteration.
• If results aren’t accurate, tweak the prompt, adjust data inputs, or provide more detailed examples.

Example Interactions

Scenario 1: Monitoring Budget Overspend

Question: “Which campaigns overspent their budget by more than 10% last month?”

Agent’s Output: A table listing campaigns, overspend amounts, and potential reasons.

Scenario 2: Summarizing Campaign Performance

Question: “Provide a summary of ad performance across all channels for Q3 for Coca Cola.”

Agent’s Output: A report highlighting total spend, ROI, and top-performing campaigns.

Best Practices for Interaction

• Be Specific: Detailed questions lead to more relevant responses.
• Iterate Frequently: Regular refinement ensures optimal Agent performance.
• Leverage Visualizations: Enable the Code Interpreter for visual outputs like charts and graphs.

Now that you understand how to interact with an Agent, let’s explore how to edit and fine-tune existing Agents.