Trouble Viewing Images? Right-click on any image and select "Open in new tab" to view a larger version. You can also zoom in using Ctrl + Mouse Wheel for easier readability.
Summary
The Bridgit AI Knowledge Editor provides a centralized workspace for managing the information, rules, examples, and configuration settings used by Bridgit AI for a connection profile.
Administrators and advanced users can use the Knowledge Editor to define schema information, business terminology, reporting examples, query history, memories, and AI processing behavior. These settings help Bridgit AI better understand data structures, interpret user requests, and generate more accurate responses.
Note: The Knowledge Editor is currently labeled as a Beta feature and may continue to evolve in future releases.
Accessing the Knowledge Editor
The Knowledge Editor can be accessed from a connection profile within VDM.
To open the Knowledge Editor:
- Locate the desired connection profile.
- Right-click the connection profile.
- Select Edit Knowledge (Beta).
Knowledge Editor Overview
The Knowledge Editor is organized into multiple tabs that manage different types of knowledge and configuration information used by Bridgit AI.
Each tab serves a specific purpose and contributes to how Bridgit AI interprets requests, understands database structures, and generates responses.
Schema Discovery
The Schema Discovery tab defines how Bridgit AI understands the structure of a database and the relationships between database objects.
Schema Discovery includes several areas used to describe and maintain schema knowledge.
Notes
The Notes section stores schema conventions, business context, caveats, and additional guidance that can help Bridgit AI interpret database structures and reporting terminology.
Discovery SQL
The Discovery SQL section defines the metadata query used to retrieve schema information.
This information can include:
- Schemas
- Tables
- Columns
- Data types
- Descriptions
- Keys
- Additional metadata
Discovery SQL helps Bridgit AI understand the structure of the connected database.
Preferred Join SQL
The Preferred Join SQL section defines authoritative join relationships between database objects.
These relationships can be imported and used to improve relationship discovery and query generation.
Join Rules
The Join Rules section stores profile-specific relationship rules that help Bridgit AI determine how tables should be joined when generating reports and queries.
Pipeline Instructions
The Pipeline Instructions tab contains profile-specific operating guidance for Bridgit AI.
Pipeline Instructions can be used to define:
- Business rules
- Naming conventions
- Output expectations
- Workflow guidance
- Query handling preferences
These instructions help influence how Bridgit AI processes requests and generates responses for a specific connection profile.
Schema Vocabulary
The Schema Vocabulary tab stores profile-specific terminology, aliases, and business language.
Schema Vocabulary can be used to map business terminology to database objects and improve schema matching.
Examples may include:
- Alternate field names
- Industry-specific terminology
- Business-friendly aliases
- Common user terminology
This helps Bridgit AI understand requests that may not directly match physical database object names.
Example Reports
The Example Reports tab stores reusable report patterns and examples that Bridgit AI can reference when generating similar reports.
Examples may include:
- Example prompts
- Report descriptions
- SQL templates
- Finished Report examples
- Visualize examples
- Expression examples
Providing examples can help establish reusable reporting patterns and improve consistency.
Successful Queries
The Successful Queries tab maintains a history of queries that have successfully executed and returned results.
This information can help:
- Identify commonly used query patterns
- Improve query reuse
- Support future report generation
- Provide reference examples for similar requests
Successful queries can also be promoted into Example Reports when a query pattern should be intentionally preserved.
Repair Lessons
The Repair Lessons tab stores validated correction records from previously repaired queries.
Repair Lessons help Bridgit AI learn from successful corrections and avoid repeating known issues when generating future queries.
This area can be used to maintain knowledge about:
- Query corrections
- Schema-related fixes
- Known problem patterns
- Successful remediation techniques
Memories
The Memories tab stores persistent information that should remain available across sessions.
Memories may include:
- Business preferences
- Naming conventions
- Organizational standards
- Profile-specific instructions
- Reusable contextual information
Memories provide a way to preserve important information that Bridgit AI should consistently consider when processing requests.
Agents
The Agents tab contains configuration and instructions used by the various AI agents that support Bridgit AI processing.
Examples include:
- Intent Agent
- Answer Agent
- Schema Answer Agent
- Schema Context Planner Agent
- Schema Lookup Concept Agent
- Query Planner Agent
- Query Summary Agent
- SQL Builder Agent
- SQL Repair Agent
- Memory Answer Agent
These agents work together to interpret requests, understand schema information, generate queries, and provide responses.
Health
The Health tab provides diagnostic and validation information for the Knowledge Editor.
This area helps administrators identify potential issues such as:
- Missing schema information
- Missing retrieval entries
- Missing or stale embeddings
- Validation warnings
- Knowledge maintenance recommendations
The Health tab can be used to monitor the overall status of the knowledge stored for a connection profile.
Notes
Note: The Knowledge Editor is intended primarily for administrators and advanced users responsible for configuring and maintaining Bridgit AI knowledge for a connection profile.
Note: Some functionality within the Knowledge Editor may continue to evolve while the feature remains in Beta.
Article Summary
The Bridgit AI Knowledge Editor provides a centralized workspace for managing the information, rules, examples, and configuration settings used by Bridgit AI. Through areas such as Schema Discovery, Pipeline Instructions, Schema Vocabulary, Example Reports, Successful Queries, Repair Lessons, Memories, Agents, and Health, administrators can help improve how Bridgit AI understands data and responds to user requests for a connection profile.
Comments
0 comments
Please sign in to leave a comment.