In-house
The "we'll do it ourselves" path is tempting — it looks like the cheapest option and it gives you full control. It works if someone in the company understands both the process and the tools — someone able to build the solution and with the time to finish it and keep it running. In most mid-sized manufacturing companies, that person either doesn't exist, or is the most overloaded person in the organization.
The most common failure scenario looks like this: an enthusiast from the IT department or a talented engineer starts building something after hours, a promising prototype appears, and then the project dies. There wasn't enough time, day-to-day duties buried the priority, or the author left and no one knows how to maintain it, because the knowledge lived in one head and was never written down. It's not a question of talent — it's that an implementation treated as a side project gets only leftover attention.
Software house
Hiring a specialized firm solves the problem of time and technical competence. You get a team that will build the solution professionally and on schedule. It works well for projects with a clearly defined, stable scope — when you know upfront what needs to be built.
The weakness shows up in two places. The first is distance from the process: a software house knows the technology, but not your company, so the first weeks go into explaining what work in the department actually looks like — and it's easy to end up with a solution that's technically correct but misses how things really run. The second is dependency after go-live: when the application is built outside the company, every change and every extension goes back to the vendor, and the company is left with a tool it can't touch itself. The better the implementation, the stronger this dependency becomes.
Off-the-shelf software
Buying a ready-made system — with an AI module, prediction, an assistant — looks like the fastest path. Sometimes it's the right call: if a mature product exists that covers your process without stretching it, buying can be more sensible than building anything at all.
The problem is that a mid-sized manufacturing company rarely fits a ready-made product. Processes tend to be specific, data is scattered across spreadsheets and emails, and an off-the-shelf system requires the company to adapt to its logic — not the other way around. That ends with using a fraction of the features while paying for the full subscription, or with costly customization that's closer to a software project than to a shelf purchase. I wrote about why smaller, tailored solutions beat a large system in the guide on microapps.