Claude Code makes the result become the beginning

Claude Code isn't really about code. It's not about Generative AI either.

It's about what happens when the gap between "I have an idea" and "this works" gets much smaller.

For a long time, building software has meant multiple translations. You explain something to a project manager, who explains it to a designer and developer who make it real. This process works, but it is slow and often slightly off.

People usually give up requesting changes before they are genuinely happy with the result.

Claude Code is changing everything.

This is a tool that works with you like a developer. You describe what you want, it makes a first draft, you make requests until you are happy. It designs, develops, adds content, fixes things. Repeatedly. Patiently...

This approach helps you stay in the flow of thinking and making at the same time.

You imagine it and 30 seconds later, you see it.

From idea to thing without the gap

At Visioning Lab, most of the work sits in that awkward middle space:

systems that don't exist yet
tools that are half research, half product
ideas that need to be tested quickly

Normally, this is where things slow down.

You write a proposal. You wait for funding. Most of the time you don't get the funding.

If funded, you write a spec. You hand it over. You wait.

With tools like Claude Code, you can just... try it.

Build a rough version. See how it works. Show it to partners. Adjust. Repeat.

This way you can build a prototype to accompany a proposal. The online tool becomes a way of testing the concept, it is not the end result of the project, and the funding.

How much further can we get if we build the website first and then make it better.

It makes clarity more important, not less

There's a catch.

If it's easier to build things, it's also easier to build the wrong thing.

Claude Code doesn't remove the need to think. It makes it more obvious when you haven't.

The real skill becomes:

knowing what you're trying to do
structuring it clearly
spotting when something is off

In other words, the same skills behind ontology work.

If you can define a system well, these tools become powerful.

If you can't, they just generate noise faster.

Less "development", more "working it out"

This shifts how software gets made.

Less:

long handovers
fixed specs
big builds

messy prototypes
quick iterations
figuring things out by doing

That fits much better with how real projects actually evolve.

For Visioning Lab, this opens up a different way of working:

building small tools directly from research
testing ideas in days instead of weeks
letting domain experts shape systems more directly

It's not about replacing developers.

It's about widening who gets to build, and when.

Claude Code isn't the story. The story is that software is becoming a more immediate medium. Closer to sketching. Closer to modelling. Closer to thinking out loud.

And that changes who can take part.