Build and Improve an Agent with AI
Use the AI Agent Builder to work out what you want your agent to do, or to improve an agent that you already have. The AI Agent Builder is a chat assistant next to the agent form. It proposes a configuration, and you review and refine it.
Note
The AI Agent Builder gives you a draft, not a finished agent. Review and edit the proposed instructions yourself, and test the agent in the preview window before you share it.
Before You Start
You must have the Agent Studio capability assigned to you with the AI Agent privilege enabled.
The AI Agent Builder suggests only datasets that you have permission to access.
A dataset must be prepared for insights before the AI Agent Builder can review it. The AI Agent Builder can't prepare a dataset for you yet. If a dataset isn't prepared, the AI Agent Builder gives you a link to the dataset's Data Insights tab. Opening the tab starts the preparation. To learn more about the tab, go to Dataset Details Page.
Create an Agent with the AI Agent Builder
Open Agent Studio in Alteryx One.
Select New Alteryx Agent.
Select AI Agent Builder.
Describe what you want the agent to help with and the kinds of questions people will ask. The AI Agent Builder asks follow-up questions if something is unclear.
If the AI Agent Builder tells you a dataset isn't prepared for insights, open the link it gives you. The Data Insights tab starts preparing the dataset when it opens, and preparation continues in the background. When you're notified that insights are ready, return to the AI Agent Builder and follow the prompts.
Review the datasets that the AI Agent Builder suggests. Tell it which ones fit and how the agent should use them.
Review the proposed name, description, instructions, datasets, and starter prompts. Edit anything that doesn't match what you need.
Tell the AI Agent Builder to apply the changes. It fills in the agent form.
Review the form, make your own edits, and select Save.
Use the preview window to test the agent.
Applying changes only updates the form. Your agent isn't saved until you select Save.
Improve an Existing Agent with the AI Agent Builder
In Agent Studio, open the agent that you want to improve and select Edit.
Select AI Agent Builder.
The AI Agent Builder reviews the agent's current instructions, datasets, and starter prompts, and asks questions about anything that is unclear or missing.
If the AI Agent Builder tells you a dataset isn't prepared for insights, open the link it gives you. The Data Insights tab starts preparing the dataset when it opens, and preparation continues in the background. When you're notified that insights are ready, return to the AI Agent Builder and follow the prompts.
Answer the AI Agent Builder's questions to clarify what you intend.
Review the proposed changes. If it suggests another dataset, check that it fits.
Tell the AI Agent Builder to apply the changes, review the form, and select Save.
Test the updated agent in the preview window.
Tips for Better Results
Describe the real questions that your audience will ask, not just the topic.
Include how you analyze the data. Agents don't reliably choose an analysis approach on their own.
Answer clarifying questions specifically. Vague answers lead to vague instructions.
Use the suggested replies to explore a dataset and get ideas for analysis.
Treat proposed instructions as a starting point, and edit them.
Limits
The AI Agent Builder currently works with datasets. It doesn't support other asset types yet.
The AI Agent Builder can't prepare datasets for insights yet. Opening a dataset's Data Insights tab starts the preparation.
The AI Agent Builder doesn't test your agent. Use the preview window.