Agentic BI ? Use Cases & Examples

Practical workflows and prompt examples for getting the most out of Knowi's Agentic BI. These examples use the in-product Dashboard Agent, but the same capabilities are available through the MCP Server for external AI tools.

For setup and feature overview, see Agentic BI and AI Tools.

Build a KPI Dashboard from a Question

Start with a business question and let the agent build a dashboard for you.

Step 1 ? Ask what you want to see:

"Create a sales performance dashboard with revenue trends, top products, and regional breakdown"

The agent identifies relevant datasets, generates queries, selects chart types appropriate for each metric, and builds a multi-widget dashboard. You get a named, saved dashboard in seconds.

Step 2 ? Refine individual widgets:

"Change the revenue trend to a line chart"
"Add a filter for date range"

Step 3 ? Add missing metrics:

"Add a widget showing average order value by month"

Step 4 ? Generate a summary:

"Summarize this dashboard"

The agent produces a written narrative covering all widgets ? useful for executive briefings or Slack updates.

Set Up Automated Reporting

Turn a dashboard into a recurring email report in one conversation.

"Email this dashboard as a PDF to [email protected] every Monday at 8 AM"

The agent creates a scheduled report with the specified recipients, format, and frequency. To adjust later:

"Change the sales report to run every Friday instead"
"Add [email protected] to the weekly report"
"Pause all reports for this dashboard"
"Show me all scheduled reports"

You can also test a report before committing to a schedule:

"Send a test run of the weekly sales report"

Create Monitoring Alerts

Set up alerts that notify you when your data changes.

Threshold alerts:

"Alert me if daily revenue drops below $5,000"
"Notify me when error count exceeds 100 in any hour"

Data freshness alerts:

"Alert me if the orders dataset hasn't been updated in 12 hours"

Managing alerts:

"Show me all active alerts"
"Pause the revenue alert"
"Delete the error count alert"

Explore and Analyze Data

Use natural language to query data without writing SQL.

Simple questions:

"What were total sales last quarter?"
"Show me the top 10 customers by revenue"
"How many support tickets were opened this month?"

Comparisons:

"Compare revenue this quarter vs last quarter by region"
"Which product categories grew the most year over year?"

Trends:

"Show me the trend in monthly active users over the past year"
"What day of the week has the highest order volume?"

Specifying a dataset:

If you have multiple datasets and the agent picks the wrong one, specify it explicitly:

"From the Product Sales dataset, show me top 10 customers by revenue"

Share Dashboards Externally

Generate shareable URLs without navigating menus.

"Create a shareable URL for this dashboard"

The agent generates a secure link that anyone can view without logging in. Useful for sharing with stakeholders who don't have Knowi accounts.

Export Data in Any Format

Export from the chat without clicking through export dialogs.

"Export this dashboard to PDF"
"Export the revenue widget to Excel"
"Download this data as CSV"

Supported formats: CSV, Excel (XLSX), PDF, and PowerPoint (PPT). When exporting a dashboard to Excel, each widget becomes a separate tab.

Rearrange a Dashboard Layout

Reorganize widgets without drag-and-drop.

"Move the revenue chart to the top left corner"
"Make the customer table take up the full width"
"Swap the pie chart and the bar chart"
"Stack all the KPI cards in a single row at the top"

Find Things Across Your Workspace

Search across all your assets when you can't remember where something is.

"Find dashboards about customer churn"
"Which datasets have shipping data?"
"Search for widgets showing NPS scores"

Results include clickable links for quick navigation.

Configure Dashboard Filters

Set up dashboard-level filters through conversation instead of the filter builder.

"Add a date range filter to this dashboard"
"Set the region filter to West Coast"
"Add a customer filter with values from the orders dataset"

Tips for Writing Effective Prompts

Be Specific About What You Want

Vague prompts produce vague results. Compare:

Instead of...Try...
"Show me sales data""Show me total revenue by product category for Q1 2026"
"Make a dashboard""Create a customer support dashboard with ticket volume, average resolution time, and satisfaction scores"
"Set up an alert""Alert me if daily order count drops below 50"

Name Your Datasets

When you have many datasets, help the agent pick the right one by including the dataset name:

"From the Shopify Orders dataset, what is the average order value this month?"

Use Follow-Up Prompts to Refine

You don't have to get everything right in one prompt. The agent maintains conversation context, so you can iterate:

"Show me revenue by region" ? "Break that down by month" ? "Change it to a stacked bar chart" ? "Export to Excel"

Ask for What You Want to Do, Not How

Describe the outcome, not the steps:

Instead of...Try...
"Run a SQL query that groups by month""Show me monthly trends"
"Create a widget with chart type bar""Add a bar chart of revenue by category"
"Schedule a report with frequency weekly""Email this dashboard every Monday at 9 AM"

Let the Agent Clarify

If your request is ambiguous, the agent will ask clarifying questions rather than guessing. Answer its questions to get better results:

You: "Show me the sales data"
Agent: "I found 3 sales datasets: Product Sales, Regional Sales, and Online Sales. Which one would you like to explore?"
You: "Product Sales"