Deep Data Agent

An intelligent data analysis assistant with dual-mode capabilities for code and SQL workflows.

Overview

Deep Data Agent provides two specialized AI copilots:

• Data Agent: Code-first assistant for Python data analysis, visualization, and file management
• SQL Agent: Natural language to SQL with automatic schema understanding and query validation

Key Features

• Dual-mode chat: Switch between Code and SQL modes based on your task
• Live streaming: Real-time responses with inline code, SQL, and visualizations
• RAG integration: Teach the agent about your data for context-aware answers
• Auto-visualization: Charts generated automatically from your data
• Multi-datasource selection: Connect and switch between multiple databases in one session
• Session history: Conversations persist across sessions
• Dynamic model discovery: Automatically discovers available AWS Bedrock models in your region
• Test connection: Validate datasource connections before running queries
• LLM auto-detection: Defaults to Claude 4.5 Haiku on Bedrock; falls back gracefully to configured providers

Using Deep Data Agent

Code Mode

In Code mode, the Data Agent helps you:

• Write and execute Python code in a secure environment
• Analyze data files in your workspace
• Generate visualizations and downloadable artifacts
• Build data pipelines and transformations

Example prompts:

• “Load the sales.csv file and show me a summary”
• “Create a bar chart comparing revenue by region”
• “Clean this dataset and export it as Parquet”

SQL Mode

In SQL mode, the SQL Agent helps you:

• Query databases using natural language
• Explore database schemas automatically
• Generate and validate SQL queries
• Visualize query results with charts

Example prompts:

• “Show me the top 10 customers by order value”
• “What’s the monthly trend for product sales?”
• “Join orders and customers tables to find inactive users”

Connecting Your Data

Databases

Connect to your databases through the admin panel:

1. Click the Settings icon in Deep Data Agent
2. Select Datasources
3. Add your database connection details
4. Click Test Connection to validate before saving
5. Select the active datasource from the sidebar to switch between sources mid-conversation

Supported databases: PostgreSQL, MySQL, SQLite, and other SQL-compatible databases. Schemas are cached automatically to speed up subsequent queries.

Documents (RAG)

Teach Deep Data Agent about your documentation:

1. Open the RAG section in settings
2. Upload documents or point to a directory
3. The system indexes content for semantic search
4. Context is automatically included in relevant queries

Supported formats: PDF, Markdown, text files, code files.

Tips for Best Results

• Be specific: “Show sales for Q4 2024” works better than “Show me some data”
• Iterate: Ask follow-up questions to refine results
• Use context: Reference previous results with “Using that data…”
• Request formats: Ask for “a table” or “a chart” explicitly

Capabilities

Documentation