Dutch machining publication Vraag & Aanbod interviewed Frerotech, a Dutch manufacturer, about its use of CloudNC's CAM Assist software. A translation of their original article follows below, with permission.
--
Frerotech, based in Elsloo in the Netherlands, is exploring how artificial intelligence can assist in programming milling machines. The company is testing CAM Assist software from CloudNC, which automatically generates CNC programs based on AI analysis. The first results are promising — but true craftsmanship remains essential.
Smarter programming with AI
Since the summer of 2025, Frerotech has been working with CAM Assist from CloudNC, integrated into Autodesk Fusion. The software analyses 3D geometries, tools, fixtures, and machine data, and automatically proposes a machining strategy.
“For some parts, the program came out perfectly the first time,” says Stefan van Bergen, mechanical engineer and sales manager at Frerotech, about the initial tests. “We didn’t have to change a thing. But there were also cases where we still had to make major corrections. Even then, it saves time — because correcting is always faster than programming from scratch.”
Frerotech is currently using the AI software on a trial basis.
“We don’t see it as a replacement for the programmer, but as a tool,” says Van Bergen. “Let the AI handle the repetitive work — surfacing, rough milling, drilling — so that the machinist can focus on the critical details.”
From VISI CAD to Fusion
The move to Fusion came with some hurdles. Frerotech traditionally works with VISI CAD/CAM, a lesser-known package that is popular among mould makers. “We still use VISI,” Van Bergen explains. “But to run CloudNC we had to switch to Fusion, which means that people need specific training.”
Although Fusion is much cheaper, the transition wasn’t only financial. “It also requires a different mindset,” says Van Bergen. “For older machinists that takes some getting used to; younger ones adapt more easily.”
A university intern compared several AI-CAM solutions last year. Based on functionality and development speed, Frerotech chose CloudNC. “Since July 2025, our first machinist has been working with it,” Van Bergen adds. “He’s young, enthusiastic, and learns fast.”
From zero to 95 percent time savings
The test results vary widely. “We have parts where the program is 100 percent correct right away,” says Van Bergen. “And others where perhaps only half is usable.” According to the measurements, Frerotech saves between 60 and 90 percent of programming time, with peaks of up to 95 percent. “That only applies to programming time, not to machine time,” Van Bergen emphasizes.
The biggest gains occur in standard operations. “An experienced machinist knows when a material has internal stress or how a wall will react,” he says. “You only see how a rolled material behaves once you start cutting. The AI can’t sense that. But for all those repetitive steps, the software does a great job.”
The role of data and database quality
Success depends heavily on the database feeding the AI. “The most important thing is that your foundation is perfect,” notes Van Bergen. “Your tools need to be entered with the correct speeds and feeds. The fixtures and clamps have to match reality one-to-one. When that’s set up properly, the results improve immediately.” Frerotech has noticed that every CloudNC update brings progress. “During our testing phase, several updates were released — and we immediately saw significant improvements.” The current version 2.0 of CAM Assist features a more intuitive interface and support for 3 + 2-axis machining. “It’s still a beta version with some teething problems,” Van Bergen says. “But we have good contact with CloudNC; they respond quickly to feedback and are open to suggestions.”
Handling sensitive data
One important consideration is that CAM Assist operates in the cloud. The software sends geometric data to a server for calculation, but according to the supplier, never complete models. “They say only partial geometries are uploaded, so that no full product can be reconstructed,” Van Bergen explains. “Still, we remain cautious. For customers under NDAs we don’t use it yet. For now, we only apply it to less sensitive parts.”
From testing to production
Frerotech will continue testing in the coming months. By early 2026, the first machinist should be using CloudNC for customer parts on a regular basis. “The plan is to expand gradually,” says Van Bergen. “My expectation is that within a year several machinists will be working with it — but we’ll have to see how it develops. Once the setup is ready, the machinist can get to the machine sooner and produce more. That’s where the real gain lies.”
Precision company with a broad foundation
Founded in 2001, Frerotech employs twelve people. Its roots are in mould making for the thermoform industry — the tooling used to produce plastic packaging trays. That market has become smaller, but Frerotech still delivers worldwide, including to Brazil and Saudi Arabia. Today the company focuses on monoparts — single precision components — for sectors such as the ASML supply chain, aerospace, medical technology, and machine building. Frerotech doesn’t supply ASML directly, but works for Tier-1 module builders.
Because ASML components require extreme cleanliness, Frerotech built a small cleanroom, or as Van Bergen calls it, a grey room. “There’s a flow cabinet that allows us to work up to ASML Grade 2 cleanliness,” he says. “There we assemble clean inserts before the full products are further cleaned.” The machine park consists entirely of Hermle milling machines, equipped with various automation systems from BMO, Hermle, and a cobot. “We can load the machines on Friday evening with multiple products, and by Monday morning they’re done,” Van Bergen says. “That’s how we make full use of our capacity.”
The balance between AI and craftsmanship
For Frerotech, adopting AI is not a revolution but a logical step in further automating the workshop. “We’re a small company,” says Van Bergen. “Our machinists do everything — programming, setup, milling, measuring. If AI can shorten programming time, we gain valuable hours at the machine.” Still, human skill remains indispensable. “The real feel for material and precision will always come from the machinist,” he concludes. “AI can assist, but craftsmanship makes the difference.”
