AI Chat

Overview

AI Chat has been added to QueryPie. Users and administrators can perform various tasks in QueryPie through natural language. It is connected to the built-in MCP (Model Context Protocol) server in QueryPie ACP and supports Tool calls. The following are examples of tasks users and administrators can perform through AI Chat by access control area.

Database (DAC)

Feature	Description
View DB connections	Check the list of database connections that I can access
Run SQL	Run a single query (up to 1,000 rows)
Session query	Run multiple queries that require large result retrieval or transactions
Workflow request	Request new DB access, schema permissions, or SQL execution approval
AI Preview	Validate schemas and analyze expected results without actually running SQL

Server (SAC)

Feature	Description
View servers/roles	Check accessible servers and assigned roles
Select role	Activate a role for access to a specific server
Shell session	Connect to a server through SSH/SFTP and start a shell
Run commands	Run Linux/Windows commands in a started shell session
Web terminal integration	Send commands to a terminal session opened by the user in the web UI

Kubernetes (KAC)

Feature	Description
View clusters	Check the list of accessible K8s clusters
Manage resources	View/create/update/delete Pods, Deployments, Services, and more
Analyze clusters	Diagnose cluster status and report issues
Check events/metrics	View node metrics and cluster events

Admin

Feature	Description
Manage access control	Grant/revoke DB/server/K8s roles, policies, and permissions
Manage connections/clusters	Register and update DB connection information, servers, and K8s clusters
Manage users	View user lists and activate/deactivate users in bulk
Audit logs	View SQL execution history, server access history, command audits, and K8s request audits
Data masking/tagging	Manage sensitive data tags and masking policies

Other supporting features

Feature	Description
Schema validation	Check actual table column names and types before writing SQL
Chart visualization	Convert query results to bar, line, pie, or area charts
YAML policy validation	Automatically validate YAML formats for SAC and KAC
Usage guide	Reference official specification documents and run automatic validation

Using AI Chat

To use AI Chat, an administrator must configure it in advance so users can use the feature.

Loading the chat window with the Ask AI button

Click the Ask AI button on the right side of the top menu in QueryPie to open the chat panel.
The chat panel slides in from the right. You can adjust the panel width by dragging the resize handle at the right edge.
You can use the following features in the header area.
- Conversation list button : Opens a dropdown that shows the automatically generated conversation list.
- New conversation button : Click the pencil icon button to start a new conversation.
- Tool list : Click the wrench icon button to check the list of MCP Tools that AI can use, such as Database, Server, and User.
- Full-screen toggle button : Switches the chat panel to full screen.
- Settings button : Click the gear icon to refresh the MCP Tool list.
- Close : Click the button to close the chat panel.
Enter a question in the chat input field at the bottom and send it.

Model selection : Check the currently used LLM model and switch to another model within the same Provider.
Context information for the page you are currently viewing is entered automatically, such as a Database detail page or SQL Editor.
When you switch pages, the context changes and a warning message is displayed.

AI generates an answer to the question. If needed, it calls MCP Tools and provides answers by referencing QueryPie data.
- While a response is being generated, a spinning indicator is displayed at the bottom, and you can click the stop button to stop generation.
- AI responses can include various artifacts, such as SQL code and charts.

AI Preview
When you click the AI preview button in the DAC Web editor toolbar, the AI chat window shows the expected execution result for the query. You can check the target schema structure, expected row count, expected performance, and potential issues without running the query.

AI responses are for reference only. For important tasks, make sure to verify the results.

Managing Conversation History

You can view and manage previous conversations.

Click the hamburger menu (☰) button in the AI Chat panel header to open the recent conversations dropdown.
In the dropdown, you can view and select up to five recent conversations, and you can start a new conversation with the New Chat button.
The View All button appears at the bottom of the dropdown only when there are six or more conversations. Click it to open the conversation list sidebar.
When the conversation list sidebar is open, you can use the following features.
- Select conversation : Click a conversation in the list to display its contents.
- Edit conversation title : When you hover over a conversation item, a more button (…) appears on the right. Select Rename and edit the title.
- Favorites : Use the more button (…) menu that appears when you hover over a conversation item to add it to or remove it from favorites.
- Delete conversation : Use the more button (…) menu that appears when you hover over a conversation item and select Delete to remove the conversation.
- Load more : The Load more button appears at the bottom of the list only when there are additional conversation pages. Click it to load older conversations.

The current version does not provide conversation search.