Chat with Data

You can chat with data in your MySQL database using MySQL Studio AI conversations. These conversations are similar to ChatGPT‑style conversations and are represented in MySQL Studio as chat files. Conversations persist in your project so you create new chats or resume an existing chat at any time.

Start or Continue a Conversation

Open the Project View and select a chat file (or create a new one). Use the chat input to enter a message and interact conversationally with MySQL AI. You can start a new inquiry or follow up on earlier responses for an ongoing, intelligent conversation.

Start Over

Reset your session at any time by using the Start Over button located in the top right corner of the conversation. This clears your conversation history, allowing you to begin a new topic with a fresh context.

Access Vector Stores

In any conversation, click the Vector Stores button to open the Vector Stores panel. This shows all vector stores currently available to MySQL AI for retrieval‑augmented answers.

Select Vector Stores for a Conversation

Use the checkboxes in the Vector Stores panel to choose which vector stores you want to include for this conversation. Select one or more vector stores for targeted searching; if all stores are unchecked, MySQL AI will rely solely on its built‑in model for responses.

Loading a Document into a Vector Store

Note

Project data files are separate from files located in the secure file priv space. Files located in the secure file priv location can only be accessed via the database or SQL commands, whereas project data files are only accessible from within MySQL AI and MySQL AI Python notebooks. See Projects and Content for details.

To load a new document into a vector store:

1. In a conversation, click Vector Stores, then select Add (+).
2. Choose a file from list of files. Note that the list of files will differ depending on which product is in use.
   1. MySQL AI will list files from the secure file priv location. The current user must possess the FILE privilege.
   2. OCI will list files located using the specified Object Store.
3. Select the vector store schema and table where the document should be loaded. A new vector store will be created if needed.
4. Select Load to begin ingestion.

Copyright (c) 2025, Oracle and/or its affiliates.