To plan or not to plan

3 min read

There is a common question that I hear from people. What is the difference between plan mode and fast mode? If AI is so smart, why do I need to bother?

I was not sure about that before either. AI is smart and can figure that out on its own. I have been using AI for rapid prototyping and it works great. But when I was building a new documentation site for our design system, I found that I had to ask AI to do some basic things that I was expecting it to do on its own.

I realised that I was not giving AI enough context and a clear roadmap. It is like the self-driving feature of Tesla. When you activate it, it starts driving on its own on the road you are on and it continues until it reaches the end of the road, but you need to enter a destination where you want to go.

So here’s what I learned.

Fast mode (just get it done)

When I know exactly what I want to do, I get started with fast mode. For example, I am creating a prototype of a date picker with segmented controls in it with some staggered micro-interaction. I prompt it, it gives me a result, I tweak it. It is fast. It is messy. But I can fix the small mess.

Plan mode (the blueprint)

But with my example of building a new documentation site, or when I was building a new Figma plugin, I found I was prompting endlessly. I think this is where plan mode helps. You give it a high-level overview of what you want to build, and it gives you a plan. You can then tweak that plan if it is not to your liking. It helps you see the big picture and what it is going to do. A good plan helps you see the end before you start. It spots problems early. It keeps you sane.

The balance

I think instead of deciding between what to do, you just need to know when to be the architect and when to be the builder.