Skip to main content
The Workbook Agent enables conversational analytics that take the need for technical know-how out of data analysis.

​
Workbook Agent

This feature requires enabling the Workbook Agent setting in AI settings.
The Workbook Agent can help you create and refine workbook queries using natural language. You can ask questions like: What was the total revenue for the last three months? How many new users signed up last month? How many support tickets are submitted on weekends? The Workbook Agent will either provide a text response or generate a query in the workbook.

​
Generating queries

In a document draft, click the ✨ (three stars) in the left side navigation. A panel will open next to the field picker with some tips about how to use the feature, as well as a text box to ask your first question. Workbook Agent panel In addition to generating queries, the Workbook Agent can also:
  • Iterate on the questions you ask to provide a more refined result. For example: 
    You: How many orders did we have in California?
Omni Agent: 1,005
You: How about in just the last 30 days?
Omni Agent: 126
    Note: Closing the Workbook Agent will delete the text history.
  • Add totals to results, such as adding a column total that sums a Sale Price measure
  • Create new measure fields, such as a Sale Price Average
  • Attach images by pasting from your clipboard to provide visual context (for example, pasting a screenshot of a chart you want to analyze) To attach an image, copy it to your clipboard and paste it directly into the Workbook Agent text input. The image appears as a file chip, ready to be sent with your message.
    The Workbook Agent > File uploads AI setting must be enabled to use this feature.
Along with the Workbook Agent in the workbook, you can also use the Generate an AI query API to interact with it. You can then use the response to run the query with the Run a query API.If your model has query_all_views_and_fields enabled and you want to query views outside of topics via the API, set the queryAllViews parameter to true.

​
Creating filters

Make filtering easier by having the agent create filters for you using natural language. You can describe a filter like Only include orders created on a Saturday or Exclude customers with a status of deactivated and the agent will automatically apply the filter to the query. To add an AI filter to your query:
  • Use the Workbook Agent window, or
  • Click the starred funnel icon next to the Options button in the Results tab: Workbook AI filter button

​
Summarizing query results

This feature requires enabling the Workbook Agent setting in AI settings.
In addition to helping you build queries, the agent can also summarize the results of the query for you. For example, you can ask the agent to:
  • Describe any anomalies in the data
  • Identify insights into trends
  • Detail the next steps you should take
To get started, you can enter a prompt directly in the Workbook Agent or click the chat bubble near the Options button: Summarize results button Clicking the Summarize results button will automatically enter a Summarize these results prompt into the Workbook Agent. The agent will typically respond with an overview of the data in the query, some observations, and potential analysis questions: Query summary from clicking the Summarize results button
Hourly Issue Distribution AnalysisKey Insights
  • Peak Hours: Issues are most frequent between 8-14 hours (PT), with a maximum of 1,600 issues at 9 AM
  • Response Time Trends:
    • Fastest median response times (2-3 minutes) occur during 8-16 hours (PT)
    • Slowest response times are late night/early morning (22-23 hours), reaching ~80-97 minutes
Visualization Recommendations
  • What are the busiest hours for issue creation?
  • How does response time correlate with issue volume?
Potential Actions
  • Optimize support staffing during peak hours (8-14 hours)
  • Investigate reasons for slow response times during late night/early morning
  • Consider implementing automated first-response mechanisms during low-staffing periods
If you want to enter an initial prompt that’s more specific, use the Workbook Agent. In the following example, the agent was given the following prompt: Can you please summarize these results and include recommendations for improvement? Query summary with recommendations
Hourly Issue Analysis SummaryKey Insights
  • Peak issue volume occurs between 8-14 hours (PT), with a maximum of 1,600 issues at 9 AM
  • Lowest issue volumes are during late evening/early morning (19-23 hours)
  • Median response times vary significantly across hours
Response Time Observations
  • Best response times (2-3 minutes):
    • 9-12 hours (PT)
    • 14-16 hours (PT)
  • Worst response times:
    • 22-23 hours (PT): ~80-97 minutes median response
    • 18 hours (PT): ~7.7 minutes median response
Recommendations
  • Investigate staffing during peak hours
  • Optimize support coverage for 22-23 hour window
  • Understand why 9 AM has highest issue volume
Potential Action Items
  • Adjust support team scheduling
  • Implement automated first-response mechanisms
  • Analyze root causes of high-volume periods
To make a summary part of a dashboard, use the AI summary visualization.

​
Generating forecasts

The agent can generate statistical forecasts for time-series data. Refer to Generating forecasts with AI to learn more.

​
Generating visualizations

The agent can also generate visualizations based on your query results. Refer to Generating visualizations with AI to learn more.

​
Getting help

If you’re stuck and can’t remember how to do something in Omni, just ask Blobby. Questions like “How do I do [thing]?” will prompt Blobby to search the official Omni docs and provide you with an answer, all without leaving your Omni workflow. You can also directly tell Blobby to search the docs when researching the answer to your question.
Working in an embedded context? If you have the Hide Omni watermark setting enabled to provide a fully white-labeled experience, the AI doc search feature will respect it. Omni doc links will not be returned in chat, even if explicitly requested.

​
Improving AI results

Queries and filter prompts submitted to Workbook Agent may require some tweaking to get the best result. Refer to the Optimizing your models for AI guide to learn how to curate your models and include context to achieve this.