AI Data Assistant (AIDA) #

Starburst’s AI Data Assistant (AIDA) is a conversational analytics assistant. AIDA accepts natural language queries and translates them into SQL, then executes those queries against the data sources configured in your Starburst Galaxy account. AIDA returns results in formats that best suit your query, such as summaries, observations, or grids.

AIDA lets you query data without writing SQL. It is designed for all users, from analysts and data engineers to business stakeholders who do not have a SQL background.

Requirements #

To use AIDA, you need:

• Existing data products (curated datasets) that you want to analyze. Consider enriching your data products to get better answers from AIDA.
• A role that has been granted the privileges to access a cluster and query the underlying schema of the data product/datasets that you want to analyze.
• A license that includes AIDA. If you cannot access AIDA in your Galaxy account, contact Starburst support.

How AIDA works #

When you ask AIDA a question, it:

1. Interprets your intent. AIDA uses a large language model (LLM) to understand what you’re asking, with the data product that you selected as context. If your question is unclear or incomplete, AIDA asks for clarification before proceeding.
2. Generates and executes SQL. AIDA generates SQL queries against the data specified in your data product. Depending on the complexity of your question, AIDA might show you the generated SQL for review before running it.
3. Returns an answer. AIDA executes queries against live data and presents results in a readable format (for example, a summary or data grid) based on the nature of the query. Depending on the persona you select, AIDA might return executive or technical summaries, key observations, implementation warnings, or recommended next steps. Responses reflect the current state of your data. AIDA does not cache results. Double-check all responses provided by AIDA.

AIDA uses a chat-style interface that supports follow-up questions, query refinements, and requests for AIDA to explain its reasoning or the SQL it generated. This lets you explore data iteratively, without requiring you to restart the session.

AIDA can access the full set of data sources configured in your Starburst Galaxy account, regardless of where the underlying data resides. However, AIDA relies on data products to provide the structure and context required to generate accurate queries. Before you can use AIDA, you must select a cluster and a data product.

Personas #

AIDA supports three personas. Each persona tailors its responses and level of detail to suit different user roles. To view the specific prompt used for each persona, click the rule icon next to the Persona menu.

The following sections describe the executive, analyst, and data engineer personas.

Executive #

Provides high-level summaries tailored to executives and decision-makers.

• Focuses on business insights and trends.
• Omits technical detail unless explicitly requested.
• Presents concise bullet points for quick understanding.

Analyst #

Offers detailed analytical summaries suitable for analysts and data scientists.

• Includes statistical analysis and relationships in the data.
• Adds contextual information and potential implications.
• May include suggestions for further exploration.

Data engineer #

Provides technical summaries tailored to engineers.

• Focuses on structure, data quality, and metadata.
• Includes schema details, cardinality, and patterns.
• Highlights potential data issues or anomalies.

Accessing AIDA #

When AIDA is live in your Galaxy account, you are notified upon login and prompted to select either AIDA or Galaxy as your future home page. If you select Make AIDA default login, you will be taken directly to AIDA upon future logins. If you select Continue with Galaxy, the Galaxy home page will remain your home page, but you can switch to AIDA at any time from the role selection menu.

Using AIDA #

1. In the AIDA UI, select a persona from the Persona menu to control the format and detail level of AIDA’s responses. Click the rule icon next to the Persona menu to review the specific prompt used for that persona.
2. From the Cluster menu, select a cluster. Click the rule icon next to the Cluster menu to see the cluster details.
3. From the Data product menu, select the data product that you want to analyze. To see the datasets contained within a data product, or an overview of the data product, click the View data product link.
4. Enter a question or prompt for AIDA in the text box. For example, you can:
   • Query specific metrics or trends (“What are our top-selling products?” or “Show me revenue by region”).
   • Explore available data (“What tables do we have?” or “What customer data is available?”).
   • Run specific analyses or create reports.
5. Press Enter or click the send submit button.

Suggested prompts #

AIDA can suggest prompts for you to run. These are inherited from the Description field provided in any usage examples that were created for the data product.

Chat history #

Use schedule Chat history to review past AIDA conversations, including the persona active during each session. Chat history shows the steps that AIDA performed, including any tool calls and SQL queries executed during the conversation. Chat sessions are deleted after 90 days.

AIDA chat sessions are tied to a specific user, not a role, such that only the user who initiated the chat session can view its history. The exception is users assigned a role with the Manage Security privilege, who can view all AIDA chat sessions across all users.

Manage chats #

Each conversation appears in the schedule Chat history.

• To rename a chat, hover over it and click the more_vert options menu, then select Rename.

• To delete a chat, hover over it and click more_vert options menu, then select delete.

• Use the search search bar to find previous chats by name.

The following table describes the icons used in the AIDA chat dialog: