MES in 2026: What You Need to Know (and What You Can Build Yourself)

Few topics cause as much confusion as MES. Manufacturers know they “need to do something with MES,” but what that actually means and how to get started, opinions vary wildly. Time to set the record straight.

In this edition I walk you through what MES really is, which four functions you need at minimum, what routes exist to implement it, and why I believe open technology has become a serious alternative. Want to go deeper afterward? Read my complete MES guide for metalworking.

MES is not a product, it is a function

This may be the most important insight I can share: an MES is not a software package you buy and install. It is a function in your factory. The function of live navigation through your production.

The analogy I like to use: your ERP is your route planner. It tells you what needs to be where, and when. But once you are on the road, you need navigation, something that tells you where you are right now and what your next turn should be. That is MES. It is the difference between having a plan and knowing what is happening on the shop floor at this moment.

That nuance matters, because it means you do not necessarily need to buy an expensive system. You need functions. And you can fill those in different ways.

The four core functions to start with

If you had to choose today where to begin, these are the four functions I would recommend to every metal company. I call them the “Core Four.”

1. Real-time order status

If there is one thing you should sort out today, it is this: a dashboard where you can see the status of every order in a single click. Which department it is at, how many orders are running simultaneously, which are on time and which are late. It is 2026, if you do not have this digitally yet, start here. No Excel, no whiteboard, no walking around the shop floor asking how things are going.

2. Daily fine scheduling

The MRP run in your ERP is too coarse for day-to-day reality. What you need is fine scheduling: daily rescheduling so operators know what has priority. And when something changes, a machine goes down, a rush order comes in, you can reschedule immediately without the entire schedule falling apart. I have written about the art of simple production planning before.

3. OEE measurement

Overall Equipment Effectiveness remains the strongest indicator you can have. Availability times performance times quality, three numbers that tell you exactly where your production stands. My experience is that the vast majority of companies have no clear picture of this, while the insights it provides are enormously valuable. You only know where you are losing time once you measure it.

4. Downtime registration

Low barrier, immediate registration. Not filling in a form after the fact, but capturing why a machine is down at the moment itself. Only then can you spot patterns and drive structural improvement. This is one of those functions that sounds simple but makes a huge difference once you have the data.

Four routes to MES

There is no single right way to implement MES. I see four common routes, each with its own pros and cons. Which one fits best depends on your situation, your IT maturity, and your ambition.

ERP extension (think Ridder, MKG, Bemet), You stay with one vendor, integration is direct. The downside: limited real-time capabilities and little room for customization. For companies that want to keep it simple this can be sufficient, but you will hit its limits eventually. In my overview of 7 essential digital factory software tools I discuss how these systems fit into the bigger picture.

Specialist apps (think Propos, Azumuta, WeldEye), Deep in one domain, strong at what they do. But you end up with multiple vendors and have to manage integration yourself. That sounds manageable until you have five running side by side.

Machine software (think TRUMPF Oseon, BySoft), Strong connectivity with that brand’s machines. Fine if you have a homogeneous machine park, but as soon as you have multiple brands it gets difficult. You are locked into an ecosystem.

Open technology with a Unified Namespace, Maximum flexibility and scalability. You build on open standards and connect everything via a Unified Namespace. The downside: it requires technical knowledge. But the barrier is dropping fast, and for companies that are serious about scaling with AI and agents this is the strongest option in 2026. More on how a UNS solves your integration problems in this edition about escaping spaghetti integrations.

ERYXON is going open source

This is what I am personally most excited about: ERYXON, an MES system I developed myself using AI coding tools, is going open source.

What is included: real-time order status per department, operator assignment to specific parts, issue tracking with photos, simplified capacity planning with dynamic rescheduling, a PDF viewer and 3D viewer in the browser, and API connections for UNS integration. It covers the “Core Four” and goes further on several points.

The model is simple: installing locally is free. The hosted version may become paid to cover server costs, but the code stays open. No vendor lock-in, no surprises.

Why this matters now

The tools to build something like this yourself matured in 2025. With AI coding tools like Claude Code, Lovable, and Cursor you can build working applications in a short time. ERYXON is proof of that, the first version was up and running within a week.

That does not mean you have to build everything yourself. But it does mean that dependence on expensive, inflexible systems is no longer necessary. The choice has grown, and the barrier to getting started is lower than ever.

Want to go deeper into MES? My complete guide to MES for metalworking covers everything from the basics to implementation.