AI SQL Editor
1. Overview
The AI Editor is a powerful functionality within Connecty AI that enables users to review and modify the SQL queries generated by the system. It bridges the gap between AI-generated logic and human oversight, allowing users to intervene directly — whether to refine the output, adjust logic, or simply understand how the query was constructed.
Designed with both technical and non-technical users in mind, the AI Editor offers:
Side-by-side comparison of the current query and the updated alternative.
Visual versioning to track changes over time and maintain clarity over query evolution.
Editable prompt interface, allowing users to tweak their natural language questions and see how the system responds.
Execution controls, enabling users to run and validate updated queries directly.
2. Interface structure
The AI Editor interface is organized into clearly defined sections to help users interpret, validate, and refine queries with ease. Each component plays a distinct role in the review and editing workflow:
Performance metadata panel
Located at the top of the interface, this panel displays key execution metrics such as:
Execution time – how long the query took to run.
Rows returned – total number of data records retrieved.
Column count – number of fields in the result.
Active vs. New version panels
The core of the interface is a side-by-side comparison:
Active panel (left): displays the last accepted or currently running SQL query version.
New Version Panel (right): shows the latest proposed query, either AI-generated or user-edited.
Results section
Located below the query panels, this section presents a tabular preview of the data returned by the current query. It includes:
Scrollable table view
Column headers and row values
Sorting and pagination instruments
It helps users validate if the query returns expected results before finalizing changes.
If results are empty, double-check filters and joins.
Prompt section
This is where users can interact with the AI using natural language. The input box allows users to:
Request changes to the query (e.g., "Add a filter for only active users")
Ask for query optimizations or different aggregations
The system will regenerate a new SQL query based on the prompt and display it in the “New Version” panel.
Combine prompts with SQL knowledge for hybrid editing.
Controls
At the bottom and corner of the interface are key control buttons:
Run: executes the query in the new version panel.
Keep: accepts the new version and makes it the active one.
These controls help view and apply changes made via prompts or manual edits.
Don’t “Keep” until you’ve verified outputs via “Run.”
3. Edit options
The AI Editor in Connecty AI supports two primary ways to modify SQL queries: manual edits and AI-assisted edits using prompts. This flexibility allows both technical users who prefer hands-on SQL control and non-technical users who prefer natural language refinement to efficiently improve their queries.
Manual edits
Users can directly modify the SQL displayed in the New Version panel by clicking into the code editor.
Key features:
Full syntax highlighting for readability
Standard SQL formatting (e.g., line breaks, indentation)
Immediate access to edit filters, joins, aggregations, or table references
Best for:
Users with SQL knowledge
Making small tweaks (e.g., changing a column name, adding a WHERE clause)
Overriding incorrect AI interpretations
Use manual editing when you know the exact change needed.
AI edits
Instead of editing SQL directly, users can enter instructions in the Prompt section using plain language. Connecty AI parses this prompt, interprets the intent, and updates the query in the New Version panel accordingly.
Examples of possible prompts:
“Add a breakdown by region”
“Filter to only show results from the last 6 months”
“Sort by total revenue descending”
Benefits:
No SQL knowledge needed
More intuitive interaction model
Ideal for iterative refinement
Use AI editing to rephrase unclear prompts or to explore new approaches.
4. Query versioning
The AI Editor tracks all changes to queries through a structured versioning system. Each time a query is modified, either manually or through AI, a new version is generated and displayed in a navigable tree format. This allows users to trace the full history of modifications and compare results between versions.
For a full breakdown of Query versioning functionality, refer to the dedicated article.
5. Controls functionality
The AI Editor includes intuitive control buttons to manage query execution and version acceptance. These controls streamline the testing, validation, and finalization of SQL edits.
▷ Run
Executes the currently selected query version.
Can be used to test either a manually written query or an AI-suggested version.
Re-runs the SQL and refreshes the output in the Results section below.
Useful when you've edited a query but want to preview the output before accepting it as the official version.
Use Run early and often during iteration to test assumptions.
✔ Keep
Accepts the selected version as the official query.
Once clicked, this version becomes the latest “committed” logic used for downstream reporting or further editing.
Often used after validating that the AI-suggested or manual changes produce the desired results.
If unsure, leave multiple versions unaccepted and return later for comparison.
Activation logic
Both buttons are enabled only when a change or selection has been made.
For instance, "Run" may be disabled until the SQL is modified or a new version is selected.
"Keep" appears when there is a difference between the current active version and the one being previewed, signaling the need to confirm or discard changes.
These controls give users full flexibility to test and adopt edits safely, making experimentation both reversible and trackable.
Last updated