AI in industry – does it really work?
Artificial intelligence in industry is no longer just a topic for conferences, presentations, and futuristic visions. Increasingly, one very specific question is being asked: does AI actually work in real production processes—where deadlines, costs, and accountability for decisions truly matter?
With this question in mind, we went to the MIT Universe AI Summit in Warsaw—not for inspiration, but for answers.
During the event, together with Christian Ulstrup (GSD at Work LLC), we conducted a live AI test.
Without prepared data or a predefined scenario, and in front of a live audience, we took a real process from Mikrostyk—the RFQ (request for quotation) process—and recreated it in real time using AI.
Case study: from 8 days to 8 minutes
Typically, the quotation preparation process in a manufacturing company like Mikrostyk takes 8 to 10 days (and sometimes longer).
It includes:
- analysis of customer documentation
- selection of technology
- preliminary tool design
- cost calculation
- feasibility analysis
- risk assessment
- preparation of the commercial offer
During the AI workshop, the same input data was processed in… 8 minutes.
Within that time, AI generated:
- a complete cost calculation
- a proposed production technology
- tool design assumptions
- market analysis
- predicted customer feedback
- and even ready-to-use marketing content.
Does this mean AI will replace engineers?
No. And it’s very important to say this clearly. The tool created during the workshop was a demo. It demonstrated the potential, but at the same time clearly highlighted the limitations—the accuracy of calculations without expert involvement at every stage is still too low to make business decisions based on it.
The quoting process in manufacturing involves dozens of variables, technological experience, an understanding of material behavior, tooling and process risks, and nuances that simply don’t exist in the data. AI does not fully “understand” this yet.
Key takeway: AI is no longer about “if,” but “how”
What we saw changes the perspective. AI itself is no longer the most interesting part.
The biggest challenge is its implementation.
Some companies are still talking about AI, while others are already building a real competitive advantage with it—and the difference between them is no longer measured in percentages, but in time.
What does this mean for manufacturing companies?
Even in such niche areas as metal stamping, progressive tool design, and complex technical cost calculations, AI is starting to find real applications.
And faster than most of the industry expects.
It will not replace experts, eliminate risk, or make decisions for the organization—but it can radically shorten analysis time, support decision-making, and transform the way technical and commercial teams work.
Step by step—or miles behind
The biggest risk today is not implementing AI, but ignoring it. And even if expert supervision and validation are still required, and the tools are not yet perfect, the direction is clear: processes that take days today will take minutes tomorrow.
That’s why we’re no longer asking whether AI works, but:
how quickly we can start using it in practice.
The MIT Universe AI Summit didn’t provide a single definitive answer, but it delivered something far more valuable—real tests, real data, real conclusions.
AI in industry works. But only when we stop treating it as a curiosity and start treating it as a tool.