how to leverage AI to empower product

Six years ago I tried to build a simple React page for a product prototype.
Graphs, charts, components, libraries - the whole thing was completely
overwhelming. Through a ton of pain and suffering, I got a basic prototype
together for my developers to reference.

Generative AI didn't exist yet. But the reason I pushed through was simple: I
wanted to show the devs exactly what I wanted. Not a requirements list, not a
static mockup - the real look and feel.

Fast forward to today, and I can build the entire front end of an application on
our framework in a few days. Every click, every hover effect. The time it takes
to go from an idea in my head to something a developer can actually see has been
cut down immensely.

Here's the approach that gets me there.

1. Learn your own framework first

Before you touch an AI tool, understand the ground you're building on - your
framework, your component library, your API schemas. You don't need to write
production code. You need to know enough to tell when something is right, wrong,
or going to be a headache for your devs down the line.

2. Describe the vision, not just the requirements

The whole point is to show, not tell. Instead of handing over a bullet list,
describe the actual experience: what's on the screen, what happens on a click,
how it should feel. That's the part a requirements doc always loses.

3. Let an AI agent do the fast iteration

This is where it gets fun. Whether you use ChatGPT or Claude, treat it like a
tireless pair-programmer. Ask good, specific questions, give it your framework
and schemas as context, and iterate quickly - tweak the layout, adjust a hover
state, restructure a component - until the prototype matches what's in your head.

4. Keep your developers in the loop

You're building a reference, not shipping to production. As you iterate, check in
with your devs: what's feasible, what complicates the codebase, what should be
done a different way. Their input keeps the prototype grounded in reality.

5. Build a real relationship with the tools

None of this works if the technology feels like a black box you're fighting
against. Stay curious, stay a little technical, and keep working with it. Do
that, and turning your vision into something tangible stops being daunting - it
becomes one of the best parts of the job.