Last month I sat with the operations lead of a German Mittelstand manufacturer. Fifteen people on the floor per shift. SAP handles financials. A MES from the early 2010s tracks production. A separate SCADA system watches the machines. Salesforce covers the commercial side. And somewhere in the middle of all of it, there is one guy. He has been there since 2007. He knows exactly which Excel file to open when the numbers fall apart.
He is the orchestration layer.
And I say this with deep respect for him. He is brilliant at what he does. But a company should never depend on one person to hold everything together. That is a fragile setup. Everybody in that room knew it. Nobody said it out loud until I did.
So let me say it here: what this manufacturer needs, what most manufacturers need, is a technical orchestration layer. A system that does what this one person does. Every day. Every shift. Without getting sick, without going on holiday, without retiring.
The real number behind the chaos
How many applications does the average enterprise actually run? Depends who you ask. Okta’s 2025 data, based on real login activity, says 101. Zylo, tracking SaaS subscriptions, says 275. MuleSoft lands at 897, though they sell integration software, so take that number with some healthy skepticism.
For a manufacturing plant, the number that matters is usually somewhere between twelve and thirty. ERP, MES, SCADA, PLM, QMS, CMMS, WMS, CRM, ITSM, BI tools. Plus whatever the last digital transformation project left behind that nobody dares to turn off.
Most of these systems work perfectly fine on their own.
But they were never built to work with each other. MuleSoft found that only 29% of enterprise applications are actually integrated. The rest? Manual handoffs. People copying numbers from one screen into another. Decisions based on whoever last saved something to a shared drive.
A Harvard Business Review study tracked workers across three Fortune 500 companies. They toggle between applications 1,200 times per day. That burns almost four hours a week just reorienting after each switch. Roughly 9% of their working time, gone. And that is office workers. On a factory floor, where a wrong number in the wrong system means a missed shipment or a failed audit, the cost goes far beyond time. It hits trust. It hits customers. It hits the bottom line.
Where the hours actually disappear
I keep asking the same question when I talk to operations people: what happens when your ERP says one thing and your MES says another?
The answer is always the same. Someone walks over to someone else. They open two systems side by side. They compare. They make a judgment call. They move on. Nobody logs this. Nobody tracks the twenty minutes it took. It is just how things work.
Multiply that by every shift. Every handoff. Every order confirmation that requires cross-referencing two sources of truth that were never designed to agree with each other.
That is the orchestration gap. Your systems work. They just work alone.
The AI that industry actually needs (and the one it keeps buying instead)
Here is where I have to be honest about something that most people in this industry will find uncomfortable.
The instinct right now is to throw “AI” at everything. And I understand why. The marketing is loud. The demos are impressive. The pressure from the board is real.
But let me share what the data actually says.
MIT’s NANDA initiative studied hundreds of enterprise AI deployments. Only about 5% of integrated GenAI pilots achieved measurable impact on the P&L. Five percent. RAND’s research on AI project failure puts broader rates above 80%. The reasons come up again and again: misaligned expectations, wrong tool for the job and infrastructure gaps.
In manufacturing, I believe the numbers are worse. Because manufacturing does not need surprise. Manufacturing needs repeatability.
Think about it this way. Would you put a system that gives a different answer to the same question every time you ask it in charge of your quality process? Your supply chain reconciliation? Your audit documentation?
I would hope the answer is a clear no.
The distinction that matters here is simple. Probabilistic AI generates plausible outputs that vary each time. It is built to be creative. Deterministic execution gives you the same output for the same input. Always. Full audit trail. Zero hallucination. Zero creative reinterpretation of your production schedule.
For a factory, that second one is everything.
Connect what you have. Stop replacing what already works.
The pattern I see at almost every manufacturer we work with goes like this: someone proposes a platform migration. Rip out the old system. Drop in a new one. Twelve months minimum. Eight figures. The promise is consolidation.
The reality? You end up with a shiny new system that still fails to talk to the three old ones you could never switch off because somebody built a critical process on top of them years ago.
I have seen this story play out a dozen times. It always ends the same way.
The better path is orchestration. Instead of replacing systems, you connect them. An orchestration layer sits above your existing platforms and coordinates what they already do well. Your ERP stays. Your MES stays. Your SCADA stays. The orchestration layer makes them work together. Deterministically. Auditably. Automatically.
This idea is simple in principle. What changed is that it is now feasible at the speed and reliability that manufacturing actually requires. But only if you build it on deterministic AI. The kind that does the same thing right, every single time.
Why this conversation belongs in Europe
I am writing this ahead of Hannover Messe 2026 and there is a reason this topic hits harder here than in San Francisco.
Europe’s regulatory environment with GDPR, the AI Act and data sovereignty requirements is often framed as a burden. And sometimes, honestly, it is.
But for industrial AI, it is the competitive advantage.
Data sovereignty means your orchestration layer will never silently route production data through a US hyperscaler. Good. Audit requirements mean every automated decision must be traceable. Good. Quality standards mean you can only ship something that works reliably. Good.
These constraints force you to build things that are actually solid. European industry has been doing this for decades. Precision engineering. Process discipline. Quality systems that hold up under pressure. The orchestration layer is the same philosophy applied to software. Software that reads between the tools. Just like that one guy on the floor does, but at scale, around the clock, across every system.
The companies that figure this out now will have more than a productivity advantage. They will have a compounding one. Because once your systems are coordinated, once the gaps are closed, every improvement you make to any single system benefits the whole operation.
The question nobody is asking at Hannover Messe
Every booth will demo AI this year. Every panel will discuss AI. Everyone will hear the word until it loses all meaning.
But here is the question I would actually like someone to answer honestly:
How many manual handoffs happen in your operations every single day?
It is the boring question. The unsexy one. But when you actually sit down and count, it explains exactly where your time and money are going.
That is the conversation I want to have in April.
Because at INXM, we are already building the answer.
Now let me show you what this looks like when it is real.
Imagine you are a buyer at a large enterprise. You wake up on a Tuesday morning. Coffee in hand. You have 180 suppliers under your responsibility. Contracts, quality scores, delivery performance, risk profiles, incoming goods receipts and open issues. All of it spread across SAP, your supplier portal, your quality management system, your email inbox and probably three or four Excel files that only you understand.
You have no idea that three of your suppliers are running out of contract today.
But INXM does.
INXM runs through all 180 suppliers. Every single day. It collects the data about their situation, their problems and their challenges. It pulls incoming goods receipts. It checks the contracts. It reviews quality scores, delivery reliability, the full performance picture and every risk signal in the system.
INXM knows that yesterday, three contracts were about to expire. It has already prepared everything you need for the call this morning. The supplier performance summary. The open quality issues. The delivery track record. The full negotiation preparation tailored to each supplier sitting ready for you.
And it has aligned everything in SAP. Checked. Prepared. Ready for you to review and go.
This is what orchestration looks like when it actually works. In a fast moving market, with processes and systems of record that no single human can manage alone anymore. INXM does that work. Quietly. Reliably. Every morning before you finish your coffee.