Data Explorer

The Data Explorer skill allows users to ask natural language questions about a dataset directly within Chat. Max translates the question into SQL, executes the query, and returns a human-friendly response—along with transparency into how the result was generated.

Overview

Use the Data Explorer skill to:

• Ask questions in plain English about a specific dataset.
• Automatically generate SQL queries behind the scenes.
• View and understand the resulting data and query.

Key Features

• Natural Language to SQL: No SQL knowledge required. Just ask your question in Chat.
• Answer Transparency: View the final SQL, the results as a DataFrame, a visualization, and an English-language explanation of the SQL.
• Consistent with Explorer View: Shares the same logic and generation pipeline as the Explorer view within a dataset.

How to Use

1. Initiate Chat: Open a conversation with Max.
2. Ask a Data Question: Type a question about your dataset (e.g., "What were sales by region last quarter?").
3. Skill Activation: Max uses the Data Explorer skill to query the appropriate dataset.
4. Review Results: Max presents:
   • A Visualization of the data
   • The DataFrame of results
   • The SQL that was executed
   • An Explanation of what the SQL is doing

Improving Results with Examples

To improve future responses, helpful examples can be saved—but only from the Explorert area of the datase. In that view, you can re-run questions and mark strong results as examples for training.

Limitations and Configuration

• Limited to one dataset per skill—cross-dataset analysis is not supported.
• Requires the dataset to be built and associated with the assistant or skill in use.

Configuring Local Max Stats for Data Explore

1. Import the Data Explorer Skill
   Copy/import the Data Explorer code skill into your system
   Create a new system and give it an icon
2. Set Up Database Connection
   Go to database connections and add a new connection
   Choose "local folder" as the connection type
   Give it a name and description (e.g., "local usage data")
   Create the connection
3. Configure Tenant Settings
   Go to Admin → Tenant Configuration
   Set the "local max stats connection" to point to your database connection
   This will pull data from the question browser every 30 minutes
4. Wait for Data Population
   The system automatically updates every 30 minutes
   Check your database connection → Resources to see if the "max fact chat answers" file appears
   The file will show a recent modification timestamp when ready
5. Create Your Dataset
   Create a dataset using the auto-generated file as the source
   Point it to your local folder connection
   Set appropriate date ranges based on tenant usage history
   Alternative: Import a pre-configured dataset with calculated metrics already set up
   Using Data Explorer
6. Configure API Access
   Ensure one of your API keys has the SQL box checked in API config
   This enables SQL generation functionality
7. Start Exploring Data
   Go to the dataset → Explore section
   Ask natural language questions about your data
   The system generates SQL queries and visualizations automatically
8. Build Knowledge Base
   When you get good results, use "Add to Examples"
   This saves the question-answer pair for future reference
   You can modify the trigger phrase before saving
   The system uses these examples to improve future responses
9. Advanced Usage
   Copy SQL queries from good results
   Paste them back into Explorer for exact reproduction
   Use the examples to train the system for specific output formats
   Set the number of examples to include in prompts (1-5 recommended)
   Key Features
   Automatic visualizations (non-deterministic - may vary each time)
   SQL generation based on natural language queries
   Example-based learning to improve consistency
   Integration capability with existing assistants or Claude
   Real-time data updates every 30 minutes from usage statistics
   The system allows users to ask questions like "Who is the biggest super user?" or "Show me negative feedback count by user by date" and get automated SQL queries, data tables, and visualizations in response.