PgAdmin 4 9.13 Adds AI Assistant Panel to Query Tool
PgAdmin 4 9.13 AI Assistant Integration
PgAdmin 4 version 9.13 adds an AI Assistant panel to its Query Tool, enabling SQL generation from natural language input when AI services are configured. This feature appears as a tab within the SQL Editor panel alongside existing History and Scratch Pad tabs.
Query Tool Features
The Query Tool remains a feature-rich SQL environment that allows:
- Execution of arbitrary SQL commands and ad-hoc queries
- Editing of updatable SELECT query result sets
- Display of current connection and transaction status
- Export of output panel data to CSV files
- Review of SQL execution plans in text, graphical, or table formats (similar to explain.depesz.com)
- Viewing analytical information about SQL statements
- Opening multiple Query Tool instances in individual tabs
Workspace Layout Changes
Version 9.13 introduces a Workspace layout that provides a distraction-free environment for the Query Tool. Key changes include:
- Welcome page opens by default when accessing Query Tool workspace
- All Query Tool and View/Edit Data tabs open within the Query Tool workspace
- Ability to connect to ad-hoc servers not registered in Object Explorer
- Connection parameters include: Server Name, Host name/address, Port, Database, User, Password, Role, and Service fields
- Connect & Open Query Tool button launches the tool with provided server details
This differs from the classic UI where users must connect to a database server and navigate to the database node before using the Query Tool.
Panel Organization
The Query Tool features two main panels:
- Upper panel (SQL Editor): Contains tabs for entering/editing queries, History tab for session queries, Scratch Pad for text snippets, and the new AI Assistant tab
- Lower panel (Data Output): Displays query result sets, execution plan information, server messages, and asynchronous notifications
If the Scratch Pad is closed, it can be reopened by right-clicking in the SQL Editor or other panels and adding a new panel.
📖 Read the full source: HN AI Agents
👀 See Also
aco-system: An Entire Company OS for Claude That Writes User Stories, Breaks Tasks, Reviews PRs
A Reddit user shared how aco-system turned a single GitHub issue into a fully validated PR with tests — driven entirely by Claude. Includes user story generation, task breakdown, secret checking, and PR review.
Cowork vs. Claude Chat: Document Extraction Accuracy Comparison
A developer tested Claude.ai chat and Cowork on extracting data from 140+ page financial PDFs using identical prompts. Chat produced institutional-grade results with self-correction and zero errors across 150+ data points, while Cowork fabricated reconciling line items, reversed unit counts, and had prior-year column contamination.
Altimate Code: Open-Source Agentic Data Engineering Harness
Altimate Code is an open-source harness that provides deterministic data engineering tools for AI agents, addressing issues like hallucinated SQL and missing schema context. It includes column-level lineage, SQL anti-pattern detection, and dbt integration, with benchmarks showing 74.4% performance on ADE-bench.
AI Subroutines: Deterministic Browser Automation with Zero Token Cost
rtrvr.ai's AI Subroutines let you record browser tasks once as callable tools that replay inside the webpage context with auth propagated for free, eliminating LLM inference costs and non-determinism for repetitive tasks.