There’s still a huge amount of work to do before AI

Last week I visited a company in the province of Milan that makes electronic components: the classic SME with a warehouse shed.

We’ve started a consulting engagement with them to modernise some processes and applications built over the years, between Access databases and web software that is now clearly dated.

Some of them go back as far as about 25 years.

Know-how on the shop floor

But what struck me most during the visit wasn’t the software side… it was how much know-how exists inside a manufacturing company of this kind.

Machines developed in-house, processes refined over decades of experience, highly vertical skills and an enormous amount of knowledge accumulated over time. I’m passionate about seeing people who are that enterprising and brilliant.

The problem?

An important part of that knowledge is still in the heads of the founders and a few key people.

There are no fully mapped processes, and not even structured procedures for some fundamental activities, from order management to shipments.

Simply put, some people know how it’s done. And if tomorrow one of those people were missing, in some cases the company would slow down a lot.

In others, it would risk almost stopping.

And I believe this situation is far more common among Italian SMEs than we imagine.

“These things are futuristic for us”

During the meeting we also talked about things that today, for anyone who works in digital every day, are fairly normal.

For example we have a product like LOCRAI, to digitise orders and delivery notes, read incoming documents—including paper ones—extract the data and bring it straight into company systems, without someone having to open the document and retype every single piece of information by hand.

Or make different applications talk to each other, automate repetitive steps and make available information that today is scattered across software, documents and people.

At one point they told us:

These things are futuristic for us.

And that sentence stuck with me.

Before AI

There’s a lot of talk about AI, agents, automation and jobs that could be replaced in the coming years. But when you really walk into companies you also realise something else: there’s still a huge amount of work to do before you even get to AI.

Digitise, map processes, transfer know-how from people into systems.

Modernise software built 20 or 25 years ago.

Eliminate manual tasks that software could already handle today.

Integrate applications that don’t talk to each other. And only then figure out where AI can actually create value.

That’s also why, when I hear that in a few years AI will have automated everything, I have some doubts.

Technology runs extremely fast. Adoption inside companies moves at a completely different speed.

The real opportunity is sitting at the table with these people, listening to them and turning decades of company knowledge into processes, data and software.