Build a Local AI Data Analyst for the Stock Market

Overview

Goal — Create an AI data analyst for stock and financial data using Claude Code, Bruin CLI and BigQuery.

Audience — Data professionals, hobbyists, and financial analysts who want to build a local AI analyst.

Prerequisites

• BigQuery project and dataset with data already loaded
• gcloud CLI installed
• Bruin CLI installed
• Claude Code installed
• Bruin MCP added to Claude Code (setup video)

High-level flow

1. bruin init empty and remove the default SQL asset.
2. Authenticate to BigQuery with gcloud.
3. Add and test a BigQuery connection in Bruin.
4. Import database metadata to create source assets.
5. Enhance metadata with bruin ai enhance.
6. Ask a real analysis question.

Steps

1) Initialize pipeline

• Run bruin init empty sp500 or select the empty template and name the folder sp500.
• Remove the placeholder asset or any other empty placeholder files (e.g. rm sp500/assets/empty.sql)

2) Authenticate to BigQuery

• Run gcloud auth login.

3) Add and test a BigQuery connection

• List existing connections: bruin connections list.
• Add a new connection: bruin connections add.
  • Set environment: default.
  • Set name: gcp-default.
  • Choose google_cloud_platform connection and provide credentials. See BigQuery connection options.
  • Provide a service account file, or set use_application_default_credentials to true to use gcloud auth login.
  • If prompted to set as default, answer true.
• Test the connection: bruin connections test --name gcp-default --env default.

4) Import database metadata

• Run bruin import database sp500 --connection gcp-default --schema stock_market.

5) Enhance asset metadata

• Run bruin ai enhance sp500/ using Claude Code.

6) Ask the analysis question

• Prompt: "Which companies had their free cash flow margin improve in the past 4 quarters but saw their stock price decrease more than 10% during the same period?"
• Ensure the agent uses bruin query for data access and returns a table or chart.

Sample outputs

• The analysis prompt and a summary of results.
• A simple table or chart output.

Troubleshooting

• Common BigQuery permission issues.
• Bruin connection errors or missing credentials.
• AI CLI authentication errors for bruin ai enhance.

Helpful links

• Bruin installation
• Bruin MCP
• Bruin MCP + Claude setup
• Bruin core concepts videos
• Import database command
• AI enhance command
• Query command
• Video walkthrough