Deploy an AI Analyst to Bruin Cloud & Slack

Deploy the analyst

1) Initialise the Bruin project

• Run bruin init empty <pipeline-name>. If the current folder is already git-initialised, this creates <pipeline-name> unless you pass --in-place.
• If the current folder is not a git repo, Bruin creates a bruin/ folder first and then creates the project and pipeline inside it.
• For more context, see Bruin project docs and video walkthrough.

2) Build the pipeline

• Use Claude to extract stock data from Yahoo Finance and Wikipedia.
• Build assets that clean and join the data.

3) Enhance metadata

• Run bruin ai enhance across the assets.
• See AI enhance command for flag options.
• Confirm descriptions, column metadata, quality checks, and lineage look correct.

4) Add the repo to Bruin Cloud

• Go to Bruin Cloud, open Team settings, then Projects.
• Add the repo to the workspace.
• Enable the pipeline and run it.
• Confirm backfill runs and daily schedule work as expected.

5) Create the AI agent in Bruin Cloud

• Go to Agents and create a new agent.
• Select the repo and pipeline.
• Add Slack credentials.
• Name the agent and select the target Slack channel.

6) Add agent instructions

• Create an AGENTS.md file in the project root with pretext, context, rules, and instructions.
• Require bruin query for all data access, and use --dry-run while testing.

7) Test in Bruin Cloud UI

• Ask a few questions to verify the agent can query the data.
• Confirm it can self-correct when the first query is not correct.

8) Test in Slack

• Mention the agent in a Slack channel.
• Ask a stock market question.
• Open the generated SQL to validate the logic.
• Request a PDF report and confirm it is generated.