Get your data in order before you buy another AI tool

You have been told to use AI. By your board, by your bank, by every vendor with a demo. So you tried. The chat tool explained month-end beautifully. Nothing was booked.

That is not a failure of the tool. It is what happens when a brilliant brain is connected to nothing.

What the mess costs you every month

Look at how a simple question gets answered in your company today. “Why did costs rise last quarter?” Someone exports a spreadsheet from the accounting system. Someone else pulls the shop numbers. A third person knows which customer is really which, because the CRM has her twice. Two days later you have an answer, built on figures nobody can trace back to a record.

Now add an AI tool on top. It reads the same three spreadsheets. It answers fast and with confidence. It is wrong in the same places your spreadsheets were wrong, and you cannot see where.

This is the quiet cost. Not the licence fee. The hours spent exporting and retyping. The decisions made on stale numbers. The risk that the confident answer is the wrong one. Every new tool bolted onto the same mess multiplies it.

AI changes nothing fundamental

Here is the part that should calm you down. A sound business still runs on four things:

• Serve the customers. Understand what they need and deliver it well.
• Run the processes well. Efficiency and good use of resources still decide the bottom line.
• Decide well. Timely decisions on solid insight.
• Keep innovating. The market moves; so must you.

AI does not change those rules. It changes how well you can keep them. It gives deeper and faster insight than a person can reach alone, and it does routine work so people can do the work that needs them.

So judge every AI idea by one test: which of the four does it make you better at? If the answer is none, skip it, however impressive the demo.

The right question in an uncertain time

A detailed five-year plan for AI is impossible. The technology will not look the same in five years. It barely looks the same in five months.

The useful question is a different one. What can we do now that helps us today and also prepares us for a future we do not yet know?

The answer is almost always the same. Get your data in order.

Your data is the fuel

AI models are rocket engines. Powerful, and useless without fuel. Your data is the fuel, and no AI solution will ever be better than the data under it.

Putting the data foundation in order means connecting and cleaning your master data, your CRM, your ERP, your accounting data, and the files around them, so every figure is traceable. It helps you twice.

It helps now. Reporting becomes reliable. Business intelligence gets sharper. Sales and marketing can segment on facts. None of this needs AI.

It helps later. When you want AI to predict customer churn, optimise stock, or answer “why did costs rise” with the records behind the answer, the data is already in place.

What changed: AI builds the data foundation

Here is the new part, and the reason an old verdict on a data project no longer holds.

Structuring, cleaning, and enriching data used to be a grinding IT project. Many failed. Many cost more than they returned. That has changed, because AI itself is very good at exactly this work:

• Reading unstructured text in emails, PDFs, and reviews and turning it into fields.
• Cleaning, standardising, and de-duplicating customer records across systems.
• Enriching thin product data with categories, specifications, and sales arguments.

AI now does much of the structuring and cleaning that made these projects manual. If a data project was shelved three years ago as too expensive, it is worth reopening.

One step to take this week

Pick one question your company asks every month that takes a spreadsheet to answer. Write down which systems the answer touches. That list is the first pipeline, and the first page of your data foundation.

If you want a second pair of eyes on that list, that is what a conversation with us is for.