21/10/2026
The other day, we heard an interesting story about a manager who built an entire product configurator using Claude Code. And he was understandably impressed with the result. An attractive interface, plenty of functionality, and a working application built much faster than he had expected.
Until he wanted to make a seemingly minor change: replace the color of an element that appeared throughout the application.
What sounded like a simple task turned out to be surprisingly complicated, requiring considerable effort, multiple adjustments, and a lot of tokens. The problem wasn't necessarily Claude Code itself. It was how the application had been structured. With the right architecture, a change like this could have been as simple as updating one variable.
And that's what makes this story interesting.
𝗔𝗜 𝗰𝗮𝗻 𝗴𝗲𝗻𝗲𝗿𝗮𝘁𝗲 𝘀𝗼𝗳𝘁𝘄𝗮𝗿𝗲 𝗿𝗲𝗺𝗮𝗿𝗸𝗮𝗯𝗹𝘆 𝗳𝗮𝘀𝘁. 𝗕𝘂𝘁 𝘀𝗽𝗲𝗲𝗱 𝗼𝗳 𝗱𝗲𝘃𝗲𝗹𝗼𝗽𝗺𝗲𝗻𝘁 𝗱𝗼𝗲𝘀𝗻'𝘁 𝗮𝘂𝘁𝗼𝗺𝗮𝘁𝗶𝗰𝗮𝗹𝗹𝘆 𝘁𝗿𝗮𝗻𝘀𝗹𝗮𝘁𝗲 𝗶𝗻𝘁𝗼 𝗾𝘂𝗮𝗹𝗶𝘁𝘆, 𝗺𝗮𝗶𝗻𝘁𝗮𝗶𝗻𝗮𝗯𝗶𝗹𝗶𝘁𝘆, 𝗼𝗿 𝗹𝗼𝗻𝗴-𝘁𝗲𝗿𝗺 𝗲𝗳𝗳𝗶𝗰𝗶𝗲𝗻𝗰𝘆.
We see a similar challenge when it comes to AI in manufacturing.
We're enthusiastic about AI. We use it ourselves, and we help manufacturers put it to work. But manufacturing involves complex processes, dependencies, and decisions that don't disappear just because AI enters the picture. AI can bring tremendous value when it's combined with solid software, reliable algorithms, and a real understanding of how production works.
Which brings us to something we believe is often overlooked when companies choose an AI partner:
𝗨𝗹𝘁𝗶𝗺𝗮𝘁𝗲𝗹𝘆, 𝘁𝗵𝗲 𝘃𝗮𝗹𝘂𝗲 𝗼𝗳 𝗔𝗜 𝗶𝗻 𝗺𝗮𝗻𝘂𝗳𝗮𝗰𝘁𝘂𝗿𝗶𝗻𝗴 𝗱𝗲𝗽𝗲𝗻𝗱𝘀 𝗹𝗲𝘀𝘀 𝗼𝗻 𝘄𝗵𝗮𝘁 𝘁𝗵𝗲 𝘁𝗲𝗰𝗵𝗻𝗼𝗹𝗼𝗴𝘆 𝗶𝘀 𝗰𝗮𝗽𝗮𝗯𝗹𝗲 𝗼𝗳, 𝗮𝗻𝗱 𝗺𝗼𝗿𝗲 𝗼𝗻 𝗵𝗼𝘄 𝘄𝗲𝗹𝗹 𝘄𝗲 𝘂𝗻𝗱𝗲𝗿𝘀𝘁𝗮𝗻𝗱 𝘁𝗵𝗲 𝗽𝗿𝗼𝗰𝗲𝘀𝘀𝗲𝘀 𝗶𝘁'𝘀 𝗺𝗲𝗮𝗻𝘁 𝘁𝗼 𝗶𝗺𝗽𝗿𝗼𝘃𝗲.
