Industrial parts · analytics

The chart that was never on the dashboard, asked for in chat

The BI system only shows what somebody built in advance. Anything deeper meant a request to a data engineer and a wait.

Client: Spare parts & industrial equipment · Germany · $15M revenue · 100 staff
Scope: MCP connection → query → chart and report
Surface: Claude.ai chat
Role: Solo — design, build, hand-over

An analytics question answered directly against the operational database.

§ 01 / Problem

Analytics that only exist if someone built them first

Management opens the BI system and the dashboard they need is often not there.

Getting a view that had not been built in advance meant asking a data engineer to produce it — slow, and expensive in a scarce resource.

So deeper questions simply went unasked, and decisions were made against whatever dashboards already happened to exist.

§ 02 / Solution

Delivered inside Claude.ai over MCP, so there is no new tool to learn.

setup

The manager adds the MCP connection in Claude.ai once.

chat

They ask an analytics question in plain language, in the chat they already use.

analyse

The agent queries the database, gathers the data, builds the charts and produces a report.

§ Built with

Claude APIMCPPostgreSQLERP / CRM data