Release Notes

Data Products

New & Improved

Data Products Integration with askEdgi Workspace

A new feature introduces askEdgi as a delivery mode for Data Products, allowing subscribed users to bring curated data into the askEdgi workspace for analysis and exploration. Data Products appear as organized folders within the workspace while still following the defined governance, storage, and access rules. This functionality helps users discover and use their subscribed data in a single workspace.

askEdgi

New & Improved

askEdgi | Greyed out State During Initialization

In the askEdgi workspace, a controlled frozen state now governs initialization to clearly indicate setup progress and prevent unsupported interactions during system readiness.

The following improvements have been implemented:

• Central messaging communicates that workspace setup is in progress.

• Analysis Mode remains available, while other modes remain visible but temporarily disabled with consistent feedback messaging.

• Navigation, workspace selector, and global actions remain visible but inactive until readiness.

• Existing threads remain readable, while new interactions are restricted during setup.

• Once initialization completes, all features activate simultaneously without abrupt transitions.

askEdgi | Workspace Left Panel Redesign

In the askEdgi workspace, the left panel has been redesigned to reduce visual clutter, improve organization, and support scalable navigation across data and connections. The experience is now more structured, responsive, and consistent across screen sizes.

The following enhancements have been implemented:

• Workspace data is now clearly separated into Imported Objects and Live Sources, distinguishing ingested data from live connections.

• A unified Data Objects search now retrieves results across imported data, live sources, folders, and object names, with clear empty state handling.

• Section-level three-dot menus standardize actions, with controlled options for folder creation, moving objects, and removing items based on section type.

• Cross-section movement is restricted to maintain data integrity between imported content and live sources.

• The left panel now uses a single vertical scroll with a fixed top inline area, ensuring consistent visibility of key actions.

• Responsive behavior has been improved with icon and tooltip support across breakpoints, without truncation or misalignment.

• The previous object limit in the workspace has been removed, allowing unlimited objects to be added from the catalog, uploads, and recipes.

• Global search and navigation have been enhanced to support tables, files, folders, chats, and connections with accurate navigation behavior.

• New Chat placement, sticky upload, and catalog actions improve usability.

askEdgi | Extended File Support in Data Catalog Widget

In the askEdgi workspace, all file types supported for upload can now be added directly from the Data Catalog widget.

The following enhancement has been implemented:

• CSV, PQT, XLSX, and JSON files can be added from the Data Catalog into the workspace.

• File ingestion behavior remains consistent with standard upload workflows.

askEdgi | Thoughts Feature

In the askEdgi module, a Thoughts view is now available for generated answers to enhance transparency and traceability of system responses.

The following capabilities have been introduced:

• A Thoughts option appears for system-generated responses and opens a structured detail view.

• Interpreted intent, detected entities, and execution approach are summarized clearly.

• Internal sources such as catalog entries, glossary references, workspace metadata, and lineage are listed in a structured format.

• Reasoning is presented as a sanitized summary of steps without exposing raw internal reasoning.

• Clear messaging is displayed for cases involving no internal sources, execution failures, or cached results.

askEdgi | Live Query Execution

In the askEdgi module, live query execution is now supported alongside cached execution, enabling direct source analysis with defined governance controls.

The following capabilities have been introduced:

• Connectors can operate in cached or live execution mode based on configuration. Live query execution is currently supported for Snowflake.

• Live connections appear in the workspace with clear indicators and without automatic data ingestion.

• Execution eligibility is controlled through defined rules to prevent unsupported scenarios.

• Mixed execution across live and cached data is blocked with guided resolution.

• Unsupported capabilities for live content remain disabled with clear feedback.

• Read-only access, masking enforcement, and execution logging are applied consistently to maintain governance and security standards.

• Query execution is automatically directed to either the workspace engine or the source system based on the analysis context, reducing unnecessary data ingestion and improving efficiency.

askEdgi | Code Explanation Panel

In the askEdgi code view, a Code Explanation feature has been introduced to provide natural language summaries of generated SQL and Python logic.

The following enhancements have been implemented:

• An Explanation option is available alongside the existing code view.

• A concise human-readable summary is generated automatically when the explanation is opened.

• Explanation supports both SQL and Python outputs.

• Show More and Show Less options are available for longer explanations.

• Explanation refreshes automatically when the underlying code changes.

• Fallback messaging is displayed if explanation generation fails.

askEdgi | Marketplace Recipe Import for On-Prem Environments

In the askEdgi marketplace module, support has been introduced for importing recipes in environments without internet connectivity.

The following capabilities have been implemented:

• Open recipes can be downloaded as structured JSON files containing metadata, configuration, and logic.

• A JSON upload option is available in My Recipes for local import and execution.

• Schema validation ensures structural integrity before import.

• Open recipes remain editable after import, while closed recipes remain executable but non-editable.

• Revoked or republished marketplace versions do not affect previously downloaded local copies.

askEdgi | Masked Column Handling for Connector

In the askEdgi workspace, masked or restricted columns from the specified connector type are now consistently omitted when tables are added.

The following improvement has been implemented:

• Masking and restriction logic now applies uniformly across all connector types.

• Masked columns are excluded from the workspace view regardless of user access level.

askEdgi | Organisation Recipe Management

In the askEdgi Organisation library, a dedicated option has been introduced to remove recipes from the organisation without deleting the original recipe.

The following enhancements have been implemented:

• A Remove from Organisation action is available in the Organisation tab.

• Removing a recipe affects only organisational visibility and does not delete the original My Recipes entry.

• Only authorized roles can remove recipes from the Organisation library.

askEdgi | Retry Upload in My Workspace

In the askEdgi My Workspace module, a Retry Upload option has been introduced for failed object uploads to improve usability and efficiency.

The following enhancements have been implemented:

• A Retry option appears when an upload fails.

• Upload can be retried without deleting and re-adding the object.

• Upload restarts gracefully while preserving user workflow continuity.

askEdgi | RAG Context Discovery and Seamless Continuation

In the AskEdgi discovery engine, a retrieval augmented generation (RAG) architecture has been implemented to enhance contextual data discovery and interaction continuity during workspace asset updates.

The following capabilities have been introduced:

• Context-aware semantic search beyond keyword matching.

• Real-time retrieval to ensure access to current information.

• Context-enriched metadata layer supporting intelligent discovery and recommendations.

• Checkpoint-based flow interruption during asset addition.

• Automatic continuation of the original prompt after workspace assets are added.

• Elimination of manual copy-paste required to resume prompts.

askEdgi | Enhanced Tooltip Context in My Workspace

In the askEdgi My Workspace module, the tooltip display has been enhanced to show full contextual object identification for improved clarity.

The following improvement has been implemented:

• Tooltips now display the complete connection, schema, and object reference for better source identification.

• The update affects only the tooltip display without impacting recipe logic.

askEdgi | AI Enrichment Validation and Progress Tracking

In the askEdgi AI Enrichment workflow, validation and progress tracking capabilities have been introduced to prevent large-scale execution errors and improve transparency.

The following enhancements have been implemented:

• A Validate Prompt preview runs enrichment on a small sample before full execution.

• Full execution requires prior validation to prevent unintended processing.

• Real-time execution progress percentage is displayed during enrichment.

• Hover details provide a function summary, including inputs, outputs, prompt context, and execution statistics after completion.

askEdgi | Analysis Mode Simplification

In the askEdgi module, Smart Mode and Focus Mode previously operated as separate modes, creating ambiguity and requiring mode changes when moving between discovery and analysis.

The following improvements have been implemented:

• Smart Mode and Focus Mode have been merged into a single Analysis Mode that supports discovery and analysis in one continuous flow.

• Discovery Mode remains available for catalog exploration and reference-based responses, while staying separate from analysis-enabled interactions.

askEdgi | Recipe Execution with Explicit Error Handling

In the Recipes module, recipe execution now follows strict and predictable behavior by removing execution time auto correction and enforcing explicit failure visibility. The following updates have been implemented:

• Recipe execution stops immediately when an error occurs and displays the exact error on screen.

• Execution does not retry and does not apply automated fixes during runtime.

• Recipe execution proceeds only when the required workspace inputs are available.

• When required inputs are missing, the flow shifts to a controlled discovery step that surfaces matching catalog assets for explicit confirmation before execution.

• Recipe creation and recipe triggering remain unchanged, while execution behavior becomes more stable and repeatable.

askEdgi | Post Execution Action Framework and Take Action Enhancements

In the Post Analysis module, the Post Execution Action Framework now enables configuring recommended actions from analysis outputs and executing them through the Take Action experience after review and approval.

The following updates have been implemented:

• Analysis result tables support Configure Action alongside existing table actions with mandatory field validation.

• Configure Action appears during analysis on a thread and supports updates after saving.

• Take Action appears after the recipe is saved and executed when tabular results are generated.

• Action execution initiates a job and displays the job identifier with a clear completion summary for successful and failed entries.

• Supported actions include Create Data Quality Service Desk Ticket, Assign Tag on Data Asset, and Post Question on Question Wall.

• Action review remains tied to a single execution and refreshes on re execution.

askEdgi | Question Wall Immediate Auto Acceptance Option

In the Question Wall module, an immediate auto acceptance option is now available in askEdgi to support instant acceptance when required.

The following updates have been implemented:

• Auto Accept SLA configuration includes a zero-hour option to enable immediate acceptance.

• When the Auto Accept setting is configured to zero hours, the response is automatically accepted after a reply is provided.

• Notifications trigger as expected after automatic acceptance.

• Existing configuration options remain available and unchanged.

askEdgi | Intent Routing Using Data Stories

In the Prompt Routing module, a router prompt stored in Data Stories now classifies intent and routes requests to the appropriate agent flow for more consistent responses.

The following updates have been implemented:

• The router prompt loads from Data Stories and evaluates intent for each request.

• General documentation intent routes to a general response path powered by curated Data Stories.

• Analytical intent routes to an analytical response path, including a catalog-focused route when applicable.

• Downstream execution logic remains unchanged, while routing becomes deterministic and explainable.

RAG | Relationship Lookup Tool for Better Table Selection

In the Discovery and Analysis module, relationship context is now retrieved on demand through a dedicated lookup tool to improve multi-table reasoning without inflating stored context.

The following updates have been implemented:

• Relationship details load only when needed for candidate tables during analysis planning.

• Relationship context remains bounded to the requested scope to avoid unnecessary expansion.

• Relationship information supports better validation of related table selection and execution planning.

Platform Libraries | Remove Unused Dependency from Core Library

In the Platform Libraries module, an unused dependency is now removed to reduce maintenance overhead and keep the library footprint clean.

The following updates have been implemented:

• The unused dependency reference is removed from the build configuration.

• Related references are cleaned up to remove dependency.

• Build stability remains verified after removal.

Advanced Jobs

The latest release introduces a set of new advanced jobs that allow users to update the data specific to modules or features using advanced algorithms.

Name

Description

RAG Knowledge Store Builder

This advanced job automatically creates and maintains a governed knowledge base for AI-powered search and agentic workflows. It crawls trusted documentation sources, extracts structured content such as text, images, and media references, and intelligently chunks the information for optimal retrieval. The job generates meaning-based representation and indexes the content into Elasticsearch to enable accurate, context-aware Retrieval-Augmented Generation (RAG). This ensures up-to-date, high-quality knowledge is available for intelligent assistants and AI applications within the platform.

System Settings

The latest release introduces new system settings that enhance user control over the application's behavior.

Key	Description	Impacted Modules
enable.googlechat.interactive.card.notification	This system setting controls whether Google Chat interactive card notifications are enabled for the Question Wall module whenever a question is assigned to them or on subsequent replies. The recipient can respond directly in Google Chat without logging in to the application. Parameters: The default value is false. If set to true, users receive an interactive card in Google Chat. If set to false, the notification sent to Google Chat is a static (non-interactive) card. The recipient can view the information, but cannot reply from Google Chat. To respond, they must log into the application.	Notifications
googlechat.credentials.path	This system setting specifies the file system path to the Google Chat Service Account API credentials JSON file used to authenticate and authorize communication with Google Chat. Parameters: The default value is empty. Provide the absolute file system path to the Google Chat Service Account credentials JSON file stored on your application server.	Inbox & Notifications
lineage.build.transformations	This system setting enables the extraction and identification of the specific transformation code segments that contribute to lineage. Parameters: The default value is false. If set to true, during lineage building, the transformation code responsible for lineage is extracted, mapped, and highlighted in the Data Catalog. If set to false, lineage is built without extracting or storing the information about the location of the transformation code. Note: Enabling this setting may increase lineage build time due to additional code parsing and extraction.	Lineage

---

Copyright © 2026, OvalEdge LLC, Peachtree Corners, GA, USA.