Chat2DB

Ask questions in plain English and get SQL, charts, and insights instantly with AI-powered data exploration.



Few Notes on AI Usage within DataPallas & Data Privacy

Your data remains private and safe — even when using all the features you'll see below.

For the longer version, read




Athena -> Data Exploration / Analysis

So, let's get started to meet Athena

From the top menu, open Processing → Explore Data & Build Dashboards and click Start — Chat2DB runs inside this app. (It's also reachable from the Chat2DB tab of any database connection.)


Processing → Explore Data & Build Dashboards — the tab from where you start the Chat2DB app, with the tab, app, and Start button highlighted

First step is to provision the DataPallas AI Crew assistants and for this you'll need to provide an API key.

Which AI vendor? For the best results, use OpenAI and/or Anthropic — their top models are the best-in-class option for powering DataPallas AI Crew assistants. The screenshot below shows every AI vendor DataPallas supports, with OpenAI and Anthropic at the top.


That said, every Chat2DB session on this page (and most AI interactions throughout these docs) was created using a much smaller and more affordable model — deliberately, to show what's achievable without breaking the bank.


Now imagine what Athena could do powered by OpenAI or Anthropic — or better yet, just wire it up and see for yourself.


Settings — API Provider: all AI vendors DataPallas supports, with OpenAI and Anthropic at the top, followed by Google Gemini, Ollama, and OpenAI-compatible providers

On the top right of the DataPallas AI Hub app provide your API key


DataPallas AI Hub — configure your AI vendor API key

Mind the Give db_query tool to Athena checkbox which is OFF by default.

Keep it OFF

(and you'll still be able to perform all the activities you'll see below)


Update AI agents settings — Give db_query tool to Athena, OFF by default

Click 'Update Agents' and wait a few minutes for the provision to complete.


AI Crew provisioning complete

All agents are provisioned - Athena is today's star.

Athena's mission: DataPallas Guru & Data Modeling/Business Analysis Expert. I help the users master DataPallas, design data models, write SQL, and architect business and reporting solutions.


DataPallas AI Crew — Athena, Hephaestus, Hermes, and Apollo

Select the existing DuckDB sample connection

Session 1: First Encounter


Chat2DB — Athena draws the database ER diagram from the opening request
Chat2DB — a top-customers question answered with SQL and a ranked results table
Chat2DB — the SQL stays visible; you always see exactly what was run
Chat2DB — when the user runs out of ideas, Athena suggests directions to explore
Chat2DB — a suggested question answered, with the result turned into a chart
Chat2DB — the full SQL behind that answer, one click away
Chat2DB — a second question, answered again as a chart
Chat2DB — the user steps away, to continue later

Persistent memory across sessions. All DataPallas AI Crew members have built-in persistent memory — they remember people, projects, past conversations, and even self-improve over time. It doesn't matter whether you come back after 5 minutes, 5 hours, days, or even months — the agents pick up right where you left off. The only way to reset an agent is to check the Force recreate option when updating agents, which completely wipes and recreates them from scratch. For this session, the app was deliberately stopped and restarted to verify that Athena genuinely remembers previous interactions.


Chat2DB — on returning to an empty chat, Athena recalls what was explored
Chat2DB — session one wraps up

Session 2: Charts and Deeper Analysis


Chat2DB — connecting to the Sales Warehouse sample
Chat2DB — Athena maps the warehouse as a star schema
Chat2DB — total revenue
Chat2DB — total sales
Chat2DB — the biggest customers, ranked
Chat2DB — revenue concentration (the 80/20 Pareto)
Chat2DB — the quarterly revenue trend
Chat2DB — revenue by product category
Chat2DB — a product-level breakdown
Chat2DB — the chat has grown long: 'the wall' that leads to organizing the same queries on the Data Canvas