How to Get AI Going in Manufacturing
by Chris Ryan

This article was originally published in German in KI Kompass Magazin (Issue 01 – Dec. 2025 / Jan. 2026, WEKA Business Media AG).
Artificial intelligence (AI) may be on everyone's lips, but it's far from being in every factory hall. On the contrary: according to various studies, the adoption of AI-powered technologies in tech companies is stalling, especially in manufacturing. Due to a lack of strategy, only a few companies have dared to take the supposedly difficult first step. But it doesn't have to be a big leap. Manageable automation of inefficient, manual processes can serve as door openers.
AI in Swiss Manufacturing
Unlike in the consumer goods industry, where AI already frequently drives central business processes such as quality control through AI-powered image analysis, AI is still in its infancy at most Swiss technology companies.
While there is a broad field of possible applications – from process analysis to machine control to virtual customer service via chatbots and much more – many places lack the necessary access to AI know-how, not least due to inadequate IT infrastructure and data foundation for AI.
Manufacturing: A Problem Child
Manufacturing companies in particular struggle with implementing AI. According to a 2024 study by ETH Zurich, Swissmem, and Next Industries, more than half of the 200 Swiss tech companies surveyed have not yet considered AI in manufacturing. Only about a quarter of companies are planning or already conducting tests, or are already applying AI in manufacturing.
In addition to lacking AI expertise and IT infrastructure, the study authors point to the comparatively difficult availability of data in the manufacturing environment as a cause. Furthermore, around 75% of the surveyed companies have no AI strategy. There seems to be a lack of orientation on where and how to approach the topic in a business-appropriate way.
How to Get Started with AI?
The study authors recommend starting with the question "What problem do we want to solve with AI?" The Smart Factory Navigator (SFN) is well-suited for answering this question. The approach, developed through many years of research, compares business areas with development stages (degree of digitalization) and offers action options for 44 defined application fields.
This makes the current state of digitalization, which includes AI usage, visible – and areas and processes with the greatest optimization potential tangible. A company-specific action plan can be defined on this basis. This plan specifies which task or process can gain the most efficiency through AI.
Example: Automated Order Processing
For a Central Swiss CNC contract manufacturer with around 30 employees, the SFN analysis revealed that the greatest potential for efficiency gains lies in order processing. The AI-optimized process should address the following weaknesses:
- Orders in different formats require manual adjustment steps
- The transfer of relevant data to the ERP does not happen automatically
- The transmission of information to production also does not run directly
iosys was commissioned to automate order processing using an AI-powered solution. Based on a detailed analysis of the existing, largely manual processes, the project team paved the way for unified processing of orders in heterogeneous formats.
Three Steps in the New Workflow
-
Data Transformation: Data from different order formats undergoes extraction and is brought into a uniform format through transformation.
-
Validation: The prepared data is validated using an AI language model. The validated data reaches the ERP via a newly created database or programming interface.
-
Production: In the ERP, each order receives an order number, which is automatically forwarded to production along with all relevant information.
A combination of Data Engineering, expertise in handling AI language models, and programming skills enabled the deployment of the new automated order processing system. AI-powered solutions in manufacturing typically require multidisciplinary skills. Industry and company-specific process understanding can serve as valuable accelerators in developing the solution.
Efficiency Gains Through AI
At the Central Swiss CNC manufacturer, the new AI-powered solution reduced the time required for processing orders from around 20 hours to two hours per week within three months. This naturally comes with remarkable cost savings, not least due to a significant reduction in errors through automation.
Equally valuable is anchoring AI in manufacturing in general:
- Building AI Know-how: Through a low-threshold, quickly deployed solution, management and staff approach the topic.
- Acceptance Through Efficiency: The experienced benefits of usage reduce fears and increase acceptance alongside efficiency.
- Focus on Strategic Tasks: Employees relieved of monotonous, repetitive tasks have more time for demanding, innovation-promoting activities.
Contrary to the widespread AI hype-hyperactivism, especially in the SME environment when using AI for manufacturing processes: It's not about one big leap. Small, safe steps aligned with the company culture also lead to the goal.
Data as the Foundation
The example of the Central Swiss CNC manufacturer illustrates that even small, traditional manufacturing companies can open up to AI in a business-appropriate way. The key is to involve employees from the start, quickly demonstrate the benefits through suitable projects, and reduce fears while building acceptance.
Once the initial hurdles are removed, follow-up projects have better chances of success. Among the recent advances in AI development, so-called AI agents that independently learn and make decisions are generating much euphoria. Despite all the enthusiasm, the risks must not be lost from view:
- Security: Interacting AI solutions bring great benefits. However, they can also create new security gaps.
- Corporate Culture: The introduction of AI solutions can lead to discontent among employees. Proactive communication can mitigate this.
- Expectations: Innovation leaps only succeed step by step based on existing or externally acquired competencies.
Considering these risks, the "AI intensity level" will always vary depending on the company, environment, type of manufacturing, etc. Regardless, capturing central manufacturing data will be crucial not only for developing new solutions but also for optimizing existing ones.
How AI Can Succeed in Manufacturing
- Define the answer to "what problem do we want to solve with AI"
- Focus on manageable, unpopular, and inefficient processes
- Get employees on board early
- Visualize and communicate benefits
- Capture data purposefully for further AI steps
Learn more about our solutions for manufacturing digitalization: Our Services | Get in Touch
Added 2026: which AI categories pay off for a manufacturing SME today, and under what condition each one does, is set out in Which AI Tools Pay Off for Manufacturing SMEs?. How order intake runs stage by stage today is set out in AI Agents in Order Intake. How such building blocks fit into a whole architecture is shown in The Model Comes Last.
Source: KI Kompass Magazin, Issue 01 – Dec. 2025 / Jan. 2026, WEKA Business Media AG, Zurich. References: ETH Zurich Study, HSG Servitization Study.