When Building Is Cheap, What Is Hard to Copy?
AI has made building your own tools cheap enough to sit on top of the products you already pay for. What that means for anyone selling software, and what stays hard to copy.
Andrew Yardley, engineering leader, hands-on with AI.CareerMore posts
The software products most teams rely on still have a purpose. People use them because they are trusted, and because so many others have adopted them. I use plenty of them myself. But I have noticed a change in how I work around them: more and more, I build small tools of my own on top of them, because no single product does everything I need in one place.
Building is no longer the expensive part
A few years ago, writing your own tool to glue several services together was a project. You needed time, a reason big enough to justify it, and someone to maintain it afterwards. With AI in the loop, that cost has dropped sharply. A tool that pulls several feeds into one view, sorts them the way you think, and uses AI to help make sense of them is now something you can put together quickly and keep improving.
Job Scout is a small public example of mine: a self-hosted API and dashboard for roles, with AI-assisted screening built in. It is not a product. It is a tool shaped exactly to how I wanted to work, and building it was the easy part.
Once you can build your own, you start to drop the platforms
The interesting step comes later. When the tool you built has better integration and better foundations for what you actually do every day, the platforms underneath start to look optional. I can see that happening in my own habits. I keep adding to what I have built, and at some point some of the services underneath will simply not be needed.
That is fine for me as a user. It is a harder question for anyone selling software.
Good ideas are easier to copy than ever
If building is cheap for me, it is cheap for everyone. A good product idea on its own is no longer much protection, because someone can rebuild the parts they need, or the whole thing, faster than ever before. I expect to see more of this: open alternatives to established products, and teams quietly replacing tools with things they made themselves.
So if you are making products now, I think you have to move very fast and stay ahead. Speed helps, but it is not enough on its own.
What stays hard to copy
The part AI cannot generate for you is specialist data, insight and information that is unique to you. That is what makes a product more than a set of features.
My view is that the strongest products from here will use AI to build fast, and then hook what they build into something that is not available anywhere else: the data you have gathered, the insight you have earned, the knowledge of a particular domain. The features can be copied. That cannot.
Where this leaves me
I am still working this out. What I am confident about is that "build or buy" has shifted. Building is cheaper than it has ever been, so the real question is what you have that nobody else can rebuild. That is where I would put the effort.