Six AI Product Principles No One Tells You—And How I'm Applying Them in My Build

The AI strategy chessboard is on fire—75% of AI product launches fail to meet their goals; here's how to avoid being another grim statistic—and here's how I'm building on a real project in AI

This is a special post, because it’s not just sharing product principles in an abstract way. I tear down a bit of what went wrong with one of the most high-profile AI product launch fails in the last year (sorry, Klarna).

And that’s not all—I’m also going to share with you how I’m using and applying these principles with a new project I’m revealing for the first time here: PromptKit. PromptKit is my answer to what I see as one of the biggest gaps in AI today: the lack of high quality prompting tools. We put more and more pressure on our prompts (heck, multi-hundred million dollar startups are built on system prompts these days), and yet we don’t really have a systematic way to test, evaluate, and take care of these prompts like the first class citizens they are.

I’d like to change that, and that’s why I’m working on PromptKit.

I’ll explain below a bit more about the Klarna story, what this tells us about the AI product principles that shape successful launches, and I’ll eat my own dogfood by applying all that to PromptKit at the end of the post. Enjoy!