Build an AI Data Analyst
Go from zero to a working AI analyst that understands your database, speaks your domain language, and answers real business questions.
Before you start
- Bruin CLI installed
- A data warehouse with data already loaded (BigQuery, Redshift, ClickHouse, or Postgres)
- An AI coding tool installed (Cursor, Claude Code, or Codex)
Create a Bruin Project
Initialize a Bruin project — the local workspace that will hold your database metadata, quality checks, and agent context.
Connect Your Data
Add your data warehouse as a connection so Bruin can reach your tables and the AI agent can query them.
Build Your Data Context
Import your database schema and enrich it with AI-generated descriptions, quality checks, and tags — so the agent actually understands your data.
Set Up Your AI Agent
Connect Bruin to your AI coding tool via MCP so the agent can read your data context and query your warehouse.
Analyze Your Data
Create an AGENTS.md with domain-specific context, then put your AI analyst to work answering real business questions.
Improve Your Context
Go further with a Bruin Glossary for shared business definitions and external MCP servers like Notion or Confluence to pull in even more domain knowledge.
What you'll build
By the end of this module you will have an AI agent that can connect to your data warehouse, understand the shape and meaning of your data, and answer natural-language business questions with real SQL queries — no dashboards, no tickets, no waiting.
The core setup takes about 45 minutes (Steps 1–5). Step 6 covers optional ways to push your context even further. Each step builds on the last, so we recommend going in order, but you can jump to any step if you already have parts in place.
How it works
Bruin CLI creates a local project that maps your database schema into lightweight asset files. Those files carry column descriptions, data quality checks, and domain context that AI agents can read. Bruin MCP then bridges the CLI to your AI coding tool so the agent can query your data directly.
The result: an AI that doesn't just guess at SQL — it knows your tables, understands your business terms, and runs validated queries against your actual warehouse.