GPT is a horizontal product if I'd ever seen one. It's so broadly applicable that you almost forget to use it as a possibility, big and small.
I use Roam as a note-taker, and the experience of getting text that I'm copying from the web into the note-taker with proper formatting is the biggest source of friction. As a field, we can't get copy and paste right at all. The best solution we can come up with is a "Paste" and a "Paste Special".
I've had to install a browser extension that filters out HTML formatting and just let me copy text-as is. Because I tire of removing carriage returns.
Today, I wanted to cut and paste this thread on high agency product managers into my notes. But I was loathed to either invoke unroll thread or do it by hand. And I remembered that GPT-3 would be able to handle it easily. 💡
I guess there's two thoughts here.
First, just like the smartphone killed off a lot of stand-alone devices, GPT will kill off lots of stand-alone apps in small niches. Whether OpenAI will be that platform remains to be seen. Will 3rd party prompters provide "apps" for consumers to use? Or are "apps" finally easy enough to build that user-agency will finally burst forth?
Second, if it's the latter, there's a lot of room for finding these little bits of workflow in existing tools, and being able to construct these on the fly. But I wonder if it'll happen because product survive because they address an entire job-to-be-done, rather than just one part of the entire problem. When it's the latter, we call them "just features". I'd be curious how the unit economics of this is going to work out.