A long time ago (way back in the 90’s), at the beginning of the dot-com era, I wrote some software that became Claris HomePage. Claris was a company attached to Apple, that built desktop applications.
The internet was just taking off at the time. It was very clear that it was like the current moment in some ways: there was obviously a lot of disruption, economic opportunity, and things to be built for a new paradigm. Claris very much wanted to have an entry, as did everyone else at the time.
Claris looked at the new paradigm through the lens of what it knew. It was a desktop app company, building desktop software (as was pretty much everyone at that moment - the internet had really just barely started), so they looked at the internet, decided web pages were the central and important feature, and decided the right thing to do was to build a desktop web page creator/editor.
This wasn’t wrong, exactly - you couldn’t argue that it had no value, and people did use it. But it wasn’t really very right either - it wasn’t eBay, Craigslist, Google, Amazon, Facebook or any of a number of other attempts to understand the new model more natively. It was a footnote, but you couldn’t really point at it in the moment and say that it was “wrong”. It just wasn’t very “right”.
This is happening again with AI, no doubt. There are lots of places where someone is adding a bit of AI to an existing application or workflow. They’re better - again, not completely wrong - but they’re not really fully transformative either.
It’s hard to be in the middle of a paradigm shift like this. It’s stressful and uncertain to find new things to do, so once you’ve found one thing that makes sense, it’s easy to want to attach to it and not look further. But that strategy can easily lead you to do something that isn’t wrong, but isn’t right, either.
(I’ll be taking the week off next week for spring break)
Interesting perspective! Do you have any advice for developers on how to strive toward more transformative solutions rather than just adding AI as an additional layer to products? Any thought experiments or the like that would help with product improvements?