When people imagine industrial AI impact, they often picture robots and fully automated factories. In custom machine building, however, the practical role of AI is subtler and, in many ways, more consequential: it helps engineers save time, organize data and prepare decisions.
Bővíz Botond Gergő, CEO of BBM Engineering Solutions Zrt., explained on the company’s podcast that AI has not taken over engineering design in the custom machine-building sector. Custom machines are bespoke industrial devices built to specific client needs—frequently for large manufacturers—and can represent investments in the hundreds of thousands or even millions of euros, with significant technical and business risk. A distinctive feature of this field is that preparing a price quote is itself an engineering task: offers are backed by research, technology scouting, preliminary cost estimates, component and control selection, and risk analysis. Multiple firms often compete for the same job, meaning much engineering effort never directly pays off.
At these early stages AI can already be helpful. According to Bővíz Botond Gergő, current systems assist with data collection, information organization, scanning catalogs and documentation, and cost estimation. AI can locate relevant solutions, controls, motors, drives or prior examples faster than manual searching.
Nevertheless, present general-purpose AI models cannot replace design engineers. The CEO noted that such models often hallucinate, have limited memory, and can return completely incorrect answers on engineering or computational questions. He recounted that even a relatively simple Python program required several iterations to become functional because the model failed to properly process the full input data.
This underlines the need for domain-trained, validated AI systems. Major engineering software vendors are moving in that direction, but for Hungarian mid-sized companies the deployment form matters. If solutions run only as cloud-based subscription services, they raise data protection and confidentiality concerns: custom machine builders frequently work with client data that cannot be submitted to an external model for training. Real breakthroughs would come from closed, locally deployable AI assistants that firms could train on their own data.
The market currently leans toward cloud subscription models because they are commercially favorable for large providers, but that choice involves trade-offs for clients handling sensitive information.
A concrete case: Bin Picking AI and measurable gains
The podcast discussed Bin Picking AI, a technology enabling a robot to pick objects from a bin or box without individually training the system on every product variant. Previously, camera-based systems had to be taught how each item looked in its many variants—a practically endless task in logistics or food-industry environments.
Siemens’ Bin Picking AI Pro proposes grasping points to the robot based on a trained neural network. BBM Engineering’s role was to convert that development into an industrial solution. Bővíz Botond Gergő said the AI component performed surprisingly well and in certain respects outperformed traditional industrial elements. Such technology opens significant opportunities in logistics and freight handling: if a robot can both identify an object and determine how to grasp, move and handle it, that is an important step toward warehouse automation, parcel processing and potential future drone-delivery systems.
Speeding up administrative engineering tasks
AI can also greatly accelerate administrative parts of engineering work: producing technical drawings, assembling documentation, checking compliance with standards, and preparing maintenance materials are all areas where machines can remove a large time burden from engineers. Creative engineering judgment and responsibility, however, remain human tasks.
Podcast timestamps and topics
Details and timestamps from the conversation:
- What does custom machine building mean? (00:30)
- Why are bespoke industrial machines expensive and risky? (01:33)
- How can AI be used in quotation preparation and data collection? (04:03)
- Why can’t AI yet take over engineering design? (07:37)
- What is Bin Picking AI? (13:56)
- Why could AI be a big opportunity in logistics and parcel handling? (20:09)
- Why might the company shift from bespoke machines toward product development? (22:05)
- Which engineering-administrative tasks can AI accelerate significantly? (28:17)
The podcast’s regular expert guest is Aczél Petra, communication researcher and professor at Széchenyi István Egyetem. The full episode is available on the publisher’s channel.
