Your New Business App: Built with Popsicle Sticks and Glue
By: Ethan Wooley, IntegraMSP TAM & Field Engineer
You may have seen a rise in “applications” made by people you didn't know were software engineers. The truth is, theym may not be. They may just be someone who put together some good prompts and asked Claude to build an app that does “fill in the blank.” This person may be your friend, co-worker, or, even WORSE, your boss. The difference is, your boss might actually make the company use it. Concerned? You should be.
Now, building apps with AI isn't all bad. I built one myself! One of my hobbies is selling on eBay, and I wanted a faster way to compare items that had sold with items still listed to determine what I could sell my own items for and how quickly they might sell. My good friend Claude and I spent hours conjuring up an app that would do just that. After all the prompting, testing, and fault-finding, I had the brilliant idea: “Why don't I see if this app already exists?” Sure enough, the app I envisioned and strived to create already existed — with more features than I could have ever prompted Claude to build myself.
While I enjoyed creating the app “myself,” the product already existed. More importantly, it was SUPPORTED. This is the canyon that exists between an application constructed by a “vibe coder” and one developed by professional software engineers. Is the software your company relies on something Joe from accounting conjured up with AI? (I hope you get that promotion, Joe.) Or does it come from a trusted software vendor with a support team, development standards, security practices, and people responsible for maintaining it?
My app doesn't interact with sensitive data. It looks at eBay.com. That's it. But what does your homebrewed software interact with? Internal databases? Servers? Employee credentials? Client data? If so, I hope your business has good insurance. The bigger concern is whether the person who built it even knows what secure and stable software is supposed to look like. Something can function perfectly during a demo while being held together by popsicle sticks and glue underneath.
That's not to say we shouldn't use AI to build things. AI has made software development more accessible than ever, and there is tremendous value in experimenting, learning, and even creating solutions to problems that don't have an existing answer. But before turning an AI-built experiment into something your business depends on, it may be worth looking at what already exists. Many established software companies are already incorporating AI into their products, and what you're paying for isn't just the application. You're paying for the developers, security, testing, updates, support, and maintenance behind it.
AI has dramatically lowered the barrier to building software. It hasn't lowered the standards that good software needs to meet. Building something yourself can be a great way to experiment, learn and explore what AI can do. Sometimes the experience alone makes it worthwhile. But when that software becomes responsible for your business, your employees, or your clients' data, the question shouldn't just be, “Can I build this with AI?” It should also be, “Should I?”
