“Give me 10…”

Ideas are easy.

And it’s easy to think the first idea is a brilliant, inspired, “ah ha!” moment.

But more often than not, the first idea is mediocre at best—or even terrible or wrong.

Testing assumptions, exploring, and peering around the corners is an essential part of the ideation or design process. But it’s so easy (and often disastrous) to lock-in prematurely on an early direction that can later prove to be wrong.

With AI, “single-prompt” engineering is like those “ah ha” moments. It feels like magic when it works. But, more often than not, it doesn’t.

Your single prompt will not only be wrong but the direction it goes is also wrong. And it can take a while to realize that.

Your AI is so powerful and fast that it can achieve a lot (and burn through a lot of tokens!) very quickly while building something that maybe resembles what you asked for but isn’t what you needed or isn’t fit for purpose.

If you’ve got your AI “assembling” teams of subagents as needed, the wasted time and tokens and effort on the wrong direction while “clanking” away in the background can be immense.

So, here’s the trick: simply tell your AI to “give me 10…”.

Try something like “give me 10 quick mockups or prototypes to help us quickly validate our assumptions, lock in the design, and test the data architecture”.

Or “give me 10 ideas from opposing angles”. Or, “give me 10 typography stacks” or “give me 10 color palettes“ or “give me 10 data tables showing the different relationships or patterns or themes between the datasets”.

Whether you’re prompting an AI or your own team of humans—or you’re tinkering solo—diverging before converging is good practice, always.

It’s easy to think AI is going to get it right on the first shot. But how does it know what you really need or what your tastes or preferences are? How does it know if its “leap of faith assumptions” (as Eric Ries calls them) are the right ones?

It’s quick for AI to mock up 10 prototypes so you can say, “I like the colors in option 3, the typography in option 6, and the layout in option 7”.

You will get you much faster towards your goals than bashing the prompts and going around in circles waiting for it to finish building it’s not-so-inspired-after-all “one-shot” result.

We’re visual beings and it’s easier and more efficient to see a bunch of options and quickly say what we like and don’t like than to attempt to describe it all in long prompts.

Not only will you produce better outputs but you’ll learn more along the way too.