Demo Scenario

An AI analyst on top of the data warehouse

The agent answers managers’ data questions: it queries the warehouse, explains the deviation and attaches the query it used.

This is a demo scenario: how we would build the solution. Names and data are illustrative — it is not a specific client’s story.

01Task

Task

Every manager’s question about data turns into a task for an analyst that takes days.

Constraints

  • Read-only: the agent never changes data.
  • Every number comes with the query that reproduces it.
02Solution

Solution

  1. 01

    Mart and metric descriptions as the agent’s context.

  2. 02

    A read-only SQL tool with volume limits.

  3. 03

    A report with a chart and the attached query.

03Architecture

Architecture

Architecture: An AI analyst on top of the data warehouse
  1. Manager → Agent runtime
  2. Agent runtime → Metric catalogue, SQL tool · read-only, Report + query
  3. Metric catalogue
  4. SQL tool · read-only → Data Warehouse
  5. Data Warehouse
  6. Report + query
Integrations
Data Warehouse · Metric catalogue · Shared drive
Stack
LLM · Tool calling · SQL · PostgreSQL · Python
04What it demonstrates

What it demonstrates

Demonstrates: data answers with a verifiable query and read-only access for the agent.

No data warehouse yet?

Data WarehouseDataZone