Rover and Post-Skeuomorphic Apps

Sumeet Singh has a nice piece over at Every (and if you aren't a subscriber, what, actually, are you waiting for?) "Two Ways to Win in the Post-software Era" makes the case that the only companies that will win are ones that either a) make the models better in some way, or b) do things that you could not do before AI.

It's b) that's particularly interesting/relevant to us at Rover (we're not working in the 'make the models better' field) because it mirrors the question we've been asking ourselves every day since we started: How does Rover take advantage of this new tool to do something that was completely impractical before?

Here's Singh:

The [startups] that will survive will leverage models’ unique, nuanced properties to invent new workflows that were not technically possible before. I call these “post-skeuomorphic apps.”

Skeuomorphism is the trap of assuming that because you have new technology, it should look like what came before. Early mobile apps fell into this pattern constantly. They replicated the physical world: trash can icons that looked like actual garbage bins, the beer-drinking app that was popular with the first iPhone. But they weren’t exploring what our phones could uniquely do.

The apps that won broke this trap entirely. Uber didn’t digitize the taxi dispatcher’s desk. It asked: What becomes possible when everyone has a phone in their pocket that knows where they are? The phone became a remote control for your life, as investor Matt Cohler has said—for food (DoorDash), for rides (Uber), for groceries (Instacart). They didn’t adapt existing workflows. They invented new ones.

In general, this tracks with how I think the AI company world will shake out (though I do have some beefs with the examples; the beer-drinking app, for instance, used tools that in fact the phone could uniquely do; Uber started as a black car service that was pretty much exactly the digitized taxi dispatcher until they realized the potential of the phone in pocket). But if, as Singh says, AI is every bit as revolutionary as the iPhone, then what becomes possible for companies working to build in the social/content space?

For us, a few things come to mind. One is certainly tailored, multimodal answers (get an answer to any question you have in whatever form you prefer). AI is well suited to this because of it's ability to make content completely liquid--"Tell me this story as a podcast!" "Now do it as an infographic!" "OK now write it as the Swedish Chef!"

Another is better recommendations (find things I will be interested in based on the contents of the things). This one is interesting because it points to a way past the algorithmic distribution feed that has dominated social media for the past decade. An AI engine that shows you things that you like based on the content of the thing itself, rather than the signals around that thing (It's similar to things you like before, or people like you liked this thing, etc) is potentially revolutionary.

But the one I find most compelling is providing human context at scale.

What does that mean? It means using AI to help facilitate conversations by providing information and guidance based on the shared history between the participants. Our goal with Rover is an AI that helps you ping your friends and have an (audio, video, or text) discussion about something with them. Rover in the background should know when to chime in, when to hold back, and what further things to recommend for you all to explore.

This is a hard problem! Many actual humans struggle with knowing when to butt into a conversation, when to lay back, and when to offer information or correction. But we all know the people who are really good at this—the perfect host, the great moderator, the friend who just gets people. What's common to all of these is that they all have deep knowledge about a subject and the human connections to it (the host who remembers who everyone's husband is, and who got too drunk at the party last year; the conference MC who has done her homework and knows all the right questions to ask) combined with a strong situational awareness about the dynamics between people.

We're really excited about this challenge in particular, and are so encouraged to see people at OpenAI and Continua start to build in this space as well; it's a strong validation that we're heading down the right path.