From 45 % to 0 % mistakes: How a German AI solves the biggest problem in industry

Instead of months of testing, just a few clicks: How intelligent software perfectly configures factories right from the start

Imagine a very complex machine in a factory, for example, one that paints car parts or coats microchips. This machine has many “controls” and “buttons” (parameters), such as temperature, pressure, speed, duration, voltage, etc.

The problem that needs to be solved

The big problem for factories is: How do you perfectly adjust all these controls to achieve the best result?

The best result can mean: the highest quality (e.g., a flawless paint finish), the least waste, the lowest energy consumption, or the fastest production time.

The traditional method is trial and error. You test a setting, look at the result, adjust a knob a little, test again, and so on. This is extremely expensive, takes a very long time, and wastes a lot of material.

MachOptima's solution

MachOptima replaces this lengthy “trial and error” process with a smart shortcut using artificial intelligence (AI).

Their approach can be described as a cycle

1. Input: A factory sends a few test runs with different settings (input parameters) to MachOptima.
2. Analysis: MachOptima's AI software looks at the results of these tests (e.g., "With setting A, the paint thickness was 95% perfect", "With setting B, only 60%").
3. Intelligent prediction: Instead of simply guessing the next setting at random, the AI ​​calculates which new settings are most likely to lead to an even better result. It learns from the limited data how the controls interact.
4. Optimization: The AI ​​suggests a new, promising setting to the factory. A new test is performed, and the cycle begins again – but it gets much, much closer to the perfect result much, much faster.

The most important thing about this

• Non-invasive: They do not rebuild the factory. They simply provide the “perfect recipes” for the existing machines.
• Data-efficient: You don't need thousands of test runs. Your AI is so smart that it can learn from very little data. This is crucial, since every test run in a factory is expensive.

Success Stories: What does that mean in practice?

• For Bosch (microchips): MachOptima helped optimize the settings for microchip coating, reducing the defect rate from 45% to virtually zero (0.003%). Savings: 85% in time and costs.
• For Mercedes-Benz (car painting): MachOptoma provided the perfect settings for painting car bodies. This not only saved on testing but also reduced production downtime. Savings: 80% in time and costs.
• For the Max Planck Institute (Simulations): MachOptima helped to make complex computer simulations much more accurate by finding the right settings from 9.8 million possible options – and doing so with a fraction of the testing effort otherwise required.

Specialized AI optimizes complex manufacturing processes in record time

MachOptima uses a special AI to find the “perfect setup recipe” for complex industrial processes in a flash, saving companies enormous amounts of time, money and materials.

They are therefore not a typical AI company that develops chatbots like ChatGPT, but rather use a specialized form of AI (machine learning for optimization) to fundamentally improve real, physical processes in industry. Their strong ties to cutting-edge research (Max Planck Institute) give them a technological edge.

About MachOptima – Research meets Industry

MachOptima is a spin-off project of the Max Planck Institute for Intelligent Systems and part of MAX!mize, the official start-up incubator of the Max Planck Society.

The goal: to translate groundbreaking research into industrial applications. Using AI-powered technology, MachOptima helps companies make production processes more efficient, precise, and error-free.

MachOptima combines scientific excellence with practical benefits – for a flawless industry of tomorrow.

Dr. Richard Hagl

I would be happy to serve as your personal advisor.

MachOptima Interim Manager

LinkedIn