Db2 for i MCP Server

For business teams

New questions for the ERP data you already have

If your ERP runs on IBM i, years of orders, customers, items and invoices are already there. This project lets your team ask that data new questions in plain language, through an AI assistant, alongside everything you use today.

Ask the docsSend this to your IBM i team

It works with what you already have

Most companies with an ERP on IBM i get their answers in one of three ways. Each does its job well, and each was built with care by people who know the business.

ERP screens and reports

The system your team works in every day, with the reports that were built for it.

Excel over ODBC

Workbooks that pull straight from the ERP, shaped by someone who knows exactly which fields matter.

A warehouse or BI tool

Dashboards and reports on data that has been modeled for the questions the business asks most.

Then there is the question that comes up on a Tuesday afternoon and doesn't have a report yet. That is where this fits in. It doesn't replace any of the above. It gives you a way to ask the one-off question straight away, against live data, and it reuses what your existing reports and queries already know.

The kind of questions people ask

Your team types a question. The assistant finds the right tables, writes a read-only query, and answers with the numbers.

Sales

  • “Which customers ordered less this quarter than last?”
  • “What is still open on our ten largest orders, and when is it due?”

Purchasing

  • “Which purchase orders are late, and from which suppliers?”
  • “Which items have we bought from more than one supplier this year, and at what prices?”

Service

  • “Which service orders have been open longest, and for which customers?”
  • “Which items come back for repair most often?”

Manufacturing

  • “Which work orders are behind schedule this week?”
  • “Which components are short for the orders due next month?”

Finance

  • “Which customers have invoices more than 60 days overdue?”
  • “How do this month’s sales compare with the same month last year, by product group?”

Your definitions, not the model's guesses

Every company has its own meaning of “open order”, “active customer” or “on time”. Your IBM i team can write those definitions down once, in a plain text file: what a status code means, how two tables join, which query answers “late purchase orders”.

The assistant then uses your definitions instead of guessing. The logic that already lives in your reports and Excel queries is a good place to start.

How business definitions work →

Why your IBM i team can say yes

  • Read-only at three levels. A read-only connection, a check that only allows queries, and a time limit that stops long queries on the IBM i.
  • Everyone signs in as themselves. People use their own IBM i user, so they see exactly what they can see today.
  • You choose what is visible. Only the libraries you allow, with sensitive columns hidden or masked.
  • Everything is logged. Every request is written to an audit log.
  • The data stays on your IBM i. Nothing is copied into a new database.
Security details →

Good to know

Someone sets it up once

Your IBM i team installs and configures the server. After that, people only add a link in their assistant and sign in.

It is not your financial reporting

Official figures still come from the reports your finance team relies on. This is for exploring and for the questions in between.

Better notes, better answers

The assistant is only as good as what it knows about your tables. A few definitions make a big difference.

Next step

Ask the documentation assistant anything about setup, security or what it can do. Or send the summary on the right to your IBM i team. It is free and open source, with nothing to buy.

Ask the docsRead the blog