From Builders to Information Architects

While working through some old “stuff” last weekend, I came across an article summarizing a Claude developer conference.  One statistic stood out. During one presentation, approximately 40% of the developers were able to ship clean code directly from Claude. These weren’t full-scale enterprise applications, but they were producing clean, usable code with minimal manual intervention.

That is a remarkable milestone!

To put it into perspective, it feels as though AI, and Claude in particular, has advanced more in the past six months than Synon 2E did during its first three years. The pace of improvement is extraordinary.  Whenever a new code generation technology appears, developers naturally focus on the code itself. They inspect every line, question every design decision, and point out every inefficiency.

“It’s obscure.”

“There’s too much bloat.”

Those criticisms aren’t wrong. They’re simply part of the adoption curve. We’ve seen this pattern before with every major advance in software development, from assemblers to fourth-generation languages, from code generators to modern frameworks.

Eventually, something changes.

Instead of asking whether the generated code is perfect, we begin asking whether it solves the business problem correctly.

For me, that is the exciting part.  I can finally see light at the end of the tedious, laborious web development tunnel.

As AI continues to mature, our greatest value will no longer come from writing every line of code ourselves. Our value will come from understanding the business problem deeply enough to model it clearly; creating a shared understanding that users, developers, and AI can all agree upon.

When the model is correct, the implementation increasingly becomes an engineering exercise rather than an act of craftsmanship.

I believe our profession is entering a significant transition.  We are moving from being builders of software to becoming architects of information systems.  Our success will be measured less by how elegantly we write code and more by how accurately we understand, model, and communicate complex business problems.

The future belongs not to those who write the most code, but to those who can create the clearest models of reality.