I was at Half-Price Books last week, standing in the computer science aisle. It is a strange feeling, looking at thousands of books filled with knowledge you spent twenty years acquiring, and realizing the AI in your pocket already knows all of it.
Programming languages. Network architecture. Database design. Algorithm optimization. Framework patterns. Deployment strategies. Books on top of books, shelf after shelf, all teaching people how to build software. And AI can now implement the vast majority of what is in those books faster than any human who ever read them.
I am not saying this as a developer who feels threatened. I am saying it as someone who realized the thing I was proud of knowing is now a commodity. And commodities are not worthless. They are just no longer what you compete on.
The thing AI cannot do
Here is what happened the next day.
A client called. They have a SaaS product that processes data from industrial equipment. Their customers use it to track maintenance schedules, part inventories, and compliance deadlines. Something was wrong with the reporting module. The tables were not joining correctly. The charts were pulling stale data.
An AI could have written the SQL query to fix the join. An AI could have rewritten the chart component. An AI could have optimized the database indexes. But an AI would not have known which of those things to do first, or second, or whether to do any of them at all.
That is because the problem was not technical. The problem was that the schema had been designed three years ago by someone who did not understand how maintenance teams actually use the data. The tables were structured around the equipment hierarchy, not around the workflow of the people who needed the reports. The correct fix was not a better query. It was restructuring the schema to match how the business actually operates.
That is judgment. And AI cannot do it.
What commoditization actually changes
Think about what happened when electricity became a commodity. Before widespread electrification, factories competed on who had the better water wheel or steam engine. After electrification, nobody competed on power generation. They competed on what they did with the power.
The same thing is happening with software implementation.
For years, businesses that needed custom software had two options. Option one was hiring a big agency that charged a premium for the expertise of writing code. Option two was hiring a junior developer and hoping they could figure it out. Both options were expensive in different ways. The agency was expensive in dollars. The junior hire was expensive in mistakes, delays, and technical debt that accumulated quietly until it broke something important.
AI changes the economics of both options. The code itself is now cheap. What is expensive is knowing what to build, in what order, with what tradeoffs, for which business outcomes.
This is not a subtle shift. It is a fundamental reordering of what a software development company actually sells. We used to sell the ability to write code. Now we sell the judgment to decide what code is worth writing.
The bookstore moment
Here is what I actually felt standing in that aisle.
I felt relief.
Not because the knowledge is worthless. Because it is not. But because for the first time in my career, the bottleneck is not how fast someone can type. The bottleneck is how well someone understands the problem.
I have worked on projects where the development team spent weeks building features nobody asked for. I have worked on projects where the technical architecture was beautiful but the business logic was wrong. I have worked on projects where the code shipped on time and solved the wrong problem.
None of those failures were caused by a lack of programming skill. They were caused by a lack of business judgment. And AI does not change that equation at all. If anything, AI makes it more dangerous. When code is cheap, bad ideas ship faster.
The value of a software development partner in 2026 is not the ability to implement. It is the ability to stop you from building the wrong thing, to sequence the right things in the right order, and to make the thousand small decisions that separate software that works from software that creates more problems than it solves.
What this means if you are buying software
If you are a business owner who needs custom software, the commoditization of implementation is the best thing that has happened to you in a decade.
The thing you have always paid for (writing code) is getting cheaper. The thing you have always struggled to find (someone who understands your business well enough to translate it into software) is still scarce. But now that developer's time is spent on judgment instead of syntax.
That is the shift. The developer who spent 80 percent of their time implementing can now spend 80 percent of their time thinking. And that changes what you get for your money.
You do not get more code. You get better decisions.
The books are still there
I bought one book at Half-Price Books. It was not a programming book. It was a book about how manufacturing companies made the transition from craft production to mass production in the early 20th century.
The parallel is surprisingly direct. Before mass production, every product was built by a craftsman who had to know everything. After mass production, factories could produce goods at scale. The craftsmen did not disappear. They became the people who designed the factories, improved the processes, and made the decisions about what to produce.
The programming books on those shelves are the craft manuals of a previous era. The code they teach is already written by AI. But the judgment they represent, the accumulation of pattern recognition, the intuition for what breaks and why, the hard-won sense for when a simple solution is the right one and when it is not, none of that is in the books. And none of it is in the AI.
That is still in the developer. It is just no longer buried under syntax.
Book a discovery session to talk about what your business should build, not just what it can build.