ENGERA
InsightsSecurity
Log inBook a demo
InsightsSecurityLog in
ENGERA

Make AI work in engineering projects.

Platform

  • Overview
  • Analyst
  • Actions
  • Editor
  • Planner
  • Documents
  • Wiki
  • Memory
  • Registers
  • Skills
  • Connectors

Industries

  • Machinery
  • Shipbuilding
  • Large projects
  • Engineering services

Implementation

  • Our approach
  • Pilot and rollout
  • Team enablement

Company

  • About
  • Careers
  • Partners
  • Insights
  • Security
  • Platform
  • Industries
  • Implementation
  • Insights
  • Security
  • About
  • Careers
  • Partners
© 2026 EngeraImprintPrivacyStatus
deen
Insights · Field report

Everyone on their own: why AI in companies does not yet work together

More than half of companies say they use AI. With one example we show the difference between using AI and actually implementing it.

16 November 20264 min read

Everyone is in

In 2024, one in five German companies with at least 20 employees used artificial intelligence. In 2026, 57% say they do.1 IT has rolled out Microsoft Copilot, software teams code with AI, and sales writes texts faster than before.

Yet in conversations with managing directors we often hear the same question: why does so little change in the company itself?

Everyone on their own

Whoever ticks “Yes, we use AI” in a survey usually means something specific. In conversations and in rollout projects with our customers we keep seeing the same picture: everyone uses AI on their own, at their own desk.

To illustrate, we picked an example from plant engineering. A supplier sends the order confirmation for a pump by mail. The buyer uploads it to a chat assistant and has it compared with the purchase order. After that, things go on as always.

Figure 1
At the desk: AI helps one person
  1. Order confirmation arrives by mailAs a PDF in the buyer's inbox
  2. Chat assistant comparesThe purchase order is copied in from the ERP
  3. New delivery date into the ERPTyped in by hand
  4. Project manager informedBy mail, if someone notices
  5. Assembly schedule adjustedAt the next scheduling meeting

AI works on itPersonData for the rest of the project

Example from plant engineering, for illustration.

The time saved in this one step is real. The next step still runs without AI: because the purchase order sits in the ERP and the assistant has no access to it, because the assembly schedule is an Excel file on another drive, because every system works on its own.

This is what most AI rollouts look like today: Microsoft Copilot, Gemini or similar assistants. Single steps get faster. The AI is not yet part of the workflow, and the core process stays as it was.

Who carries the rollout

In many companies AI sits with IT. That is natural: licences, security and systems live there. But where AI is meant to go into the work itself, it takes the people who know that work: how a purchase order runs, who signs off what, where the data really lives.

Where AI actually arrives in the company's processes, we see this pattern. Leadership shows that the topic matters. The business carries the rollout, close to day-to-day operations. IT makes it possible; it is the enabler.

That matches a McKinsey study: almost three quarters of AI high performers have fundamentally redesigned their workflows. Among the others it was a quarter.2

The same process with AI in the process

Let's look at how the same process would run if it were digital end to end, with AI working wherever it helps.

Figure 2
In the process: AI works throughout
  1. Order confirmation arrivesCaptured straight from the inbox
  2. Checked against the orderDate, quantity, price and specs against the ERP
  3. Deviation reviewedPurchasing accepts or objectsDeviation with its reason
  4. Delivery date in the ERPUpdated without retypingCurrent delivery date
  5. Assembly planning informedThe same day, with the impact on the plan
  6. Supplier ratedAcross all projectsOn-time rate per supplier

AI works on itPersonData for the rest of the project

The same process as above, for illustration.

The difference lies less in speed than in what the process produces. Every deviation is recorded with its reason, every delivery date in the ERP is current. A delayed delivery triggers the next step by itself: assembly planning learns about it the same day instead of at the next scheduling meeting.

And with every order the company learns. After a few projects the data shows which supplier keeps its dates and where buffers are needed. The next quotation plans with these numbers.

Two things have to come together for this. The work has to be digital so AI can reach the data. And AI has to store its results where the next steps will find them. That takes AI connected to the ERP and the project files that fits into the processes, and a person who signs off.

What to do now

  1. 01

    Show from the top that it matters

    Leadership makes clear that AI is a priority and gives teams time and backing for it. Which process changes is decided by those who know it.

  2. 02

    Let the business lead

    Those who know the process carry the rollout. IT makes it possible.

  3. 03

    Pick a whole process

    From intake to sign-off, with the systems it touches.

  4. 04

    Think beyond the single use case

    Consider which steps come next and which future use cases will build on the same data.

Sources

  1. Bitkom: For the first time, a majority of companies use AI (German), 14 Sep 2026
  2. McKinsey: The state of AI in 2026: On the road to ROI, 25 Aug 2026

Read on

  • GuideWhere your AI runs: three paths and their terms
  • EssayNot wrong, but that's not how we do it here
  • AnalysisThe future of engineering work

Which challenge is keeping
your team busy right now?

Tell us, and in the demo we show you how Engera makes it easier.

Book a demoSee Engera in action