Get Started with askEdgi
This article introduces the askEdgi workspace and explains how to get started with AI-powered conversations. It describes the available chat modes, interactive guidance, and the basic workflow for submitting prompts and reviewing results. It also explains how to use @Mentions to reference catalog assets and how askEdgi retrieves and analyzes information within the workspace.
Interactive Guidance Section
The workspace offers quick-access learning components designed to promote platform familiarity and operational efficiency.
Watch a Demo
Opens an embedded demonstration of workspace operations, file handling, and analysis execution.
Take a Tour
Launches an interactive walkthrough that highlights key interface elements, buttons, and feature modules.
These options can be initiated at any time and run independently of active workspace operations.
The guidance components operate in a read-only capacity and do not modify workspace configuration or data.
AI Assistant Interface
The AI assistant embedded within the workspace provides a dynamic conversational interface for data exploration, metadata discovery, and analytics generation.
Chat Mode Selection
Discovery
Searches and retrieves catalog information such as datasets, tables, glossary terms, and governance details. The mode focuses on structured exploration and metadata understanding without performing analytical execution.
Analysis
Performs end to end intelligence and execution. The mode interprets intent, retrieves relevant enterprise context, validates dataset relationships, and performs analysis on eligible workspace data. It supports business aligned explanations and analytical outputs such as summaries, tables, or visual insights.
Analysis Mode uses Retrieval Augmented Generation to retrieve enterprise context before execution. Discovery Mode does not use Retrieval Augmented Generation and does not perform analysis.
Continuous Discovery to Analysis Flow
Analysis Mode supports a continuous experience that combines discovery, validation, and execution within a single interaction flow.
Instead of separating guidance and execution into multiple modes, askEdgi:
Understands the business intent behind a question
Identifies relevant datasets
Validates structural relationships
Performs analysis
Explains results in business context
This unified flow reduces ambiguity, improves usability, and accelerates insight generation without requiring manual mode transitions.
Workspace First Execution Boundary
Analysis Mode respects the Workspace as the execution boundary.
Execution rules include:
If tables remain pinned, analysis is restricted to pinned tables
If no tables remain pinned, eligible workspace tables are considered
Additional catalog assets are surfaced only when necessary
Data outside the intended scope is not analyzed
This boundary ensures controlled execution, prevents unintended data usage, and maintains governance alignment.
Deterministic Behavior Principles
askEdgi follows deterministic execution principles to ensure predictable and trustworthy outcomes.
The system:
Executes only when required context is available
Validates structural compatibility before execution
Stops when information is incomplete or unclear
Displays explicit errors without hidden corrections
Avoids automatic retries or silent adjustments
These principles ensure that analytical results remain explainable, repeatable, and governance aligned.
Interaction Process
Select Mode
Choose the appropriate chat mode (Discovery, or Analysis) from the selector below the input field to define the context for the assistant's response.
Context References @mention
The @Mention feature allows users to directly reference specific catalog objects in their prompts, such as catalog tables, files, business glossary terms, data products, tags, live sources, and workspace assets. Instead of relying solely on retrieval searches to identify relevant objects, users can explicitly point the system to the exact assets they want to use.
How It Works
When a user types @ in the prompt, askEdgi displays a list of available objects from multiple categories, including:
Workspace tables
Live sources
Catalog tables
Catalog files
Business Glossary
Tags
Data products
It displays top matching results across all supported asset categories.
Provides category-based filtering to quickly locate the desired object.
Example
A business glossary term called Average Profit defines the formula: Average Profit = Total Profit ÷ Number of Periods
To calculate this metric using a specific dataset, the user can reference both the glossary term and the table directly in the prompt:
Calculate @Average Profit for @Sample Sales Data
The system uses the referenced objects directly instead of inferring intent from a broader catalog search.
Faster Execution
Since the required assets are already specified, the system can bypass extensive retrieval and matching processes. This reduces execution time and speeds up response generation.
Automatic Workspace Inclusion
When a referenced catalog table is not already available in the workspace, askEdgi can automatically add the table to the workspace (subject to user permissions) and use it during execution.
Analysis Mode Support
In Analysis Mode, @Mentions help users perform calculations, transformations, and data analysis on specific datasets and business concepts.
Discovery Mode Support
The @Mention feature is also available in Discovery Mode. Users can reference catalog assets directly to retrieve metadata, descriptions, definitions, and other contextual information.
For example:
“Tell me more about @Average Profit” - The system resolves the referenced glossary term and returns its associated metadata and business definition.
Key Enhancements
Displays top matching results across all supported asset categories.
Supports keyboard navigation using arrow keys and Enter.
Provides category-based filtering to quickly locate the desired object.
Directly resolves user-selected assets during prompt execution for more deterministic and reliable results.
By enabling users to explicitly reference catalog assets, @Mentions provide a faster, more accurate, and context-aware interaction experience in askEdgi.
Submit Prompt
Enter a natural language question or command in the chat input field to specify the requested action or information.
Automatic Processing
The assistant processes the prompt using the selected mode's specialized capabilities and available workspace data to generate a response.
Review Results
The system presents the output in dynamically rendered formats, including visual charts, text summaries, or structured data tables based on query content.
Intent Routing Using Data Stories
askEdgi includes an intent routing mechanism that classifies user requests and directs them to the appropriate response pathway before execution begins. This routing logic is powered by a router prompt stored in Data Stories, ensuring consistent interpretation and deterministic behavior across interactions.
The routing mechanism improves response quality by selecting the most suitable processing path based on the detected intent.
Routing Behavior
For each user request, askEdgi:
Loads the router prompt from Data Stories
Evaluates the request to determine intent type
Routes the request to the appropriate execution pathway
Supported routing paths include:
General documentation and knowledge responses
Analytical execution workflows
Catalog focused exploration scenarios
This approach ensures that requests are handled consistently while preserving governance controls and execution boundaries.
Copyright © 2026, OvalEdge LLC, Peachtree Corners, GA USA
Was this helpful?