How This Team Plans To Find The World’s Top AI Experts
Are you one of them?
Evernote, the uber-popular note taking app described as “a place for individuals and teams to capture, nurture, and share ideas in any form,” is a billion dollar company and recognized as one of the first “unicorns” of tech. Phil Libin, c0-founder and former CEO of Evernote, grew the company from zero to over 100 million users. Libin is also a former managing director of General Catalyst, one of Silicon Valley’s premier venture capital firms (investor in companies like airbnb, KAYAK, Jessica Alba’s The Honest Company, Snapchat, and Warby Parker), and he’s now the founder of AI startup studio, All Turtles.
Libin remains first and foremost a “product guy” who is obsessed with creating practical products that solve real problems. During his 20+ years in the startup ecosystem, Libin realized that the traditional venture capital funnel is too narrow for most products. His solution — create an environment where product geniuses can focus solely on the product and have support for everything else.
Let’s get into the interview, shall we?
Please share a little bit about yourself.
I am a computer nerd. I got my first computer when I was ten or eleven. Most of my friends had computers and were programmers and that’s what we all did, so I self-identified as a geek for a long time. Programming was the main thing I did for most of my childhood. I started a few companies right out of college with some college friends, and I’ve been starting companies with more or less that same group of people ever since.
My third company was Evernote. I was the co-founder and CEO of Evernote for nine years. I’m really proud of the stuff we built over there. We actually started working on some of these AI concepts inside of Evernote.
Afterwards, I became a VC and joined General Catalyst to invest in other companies, most focusing on bots and AI. We made a bunch of super cool investments there, and then just recently we announced our new project which is an AI startup studio called All Turtles.
“Programming was the main thing I did for most of my childhood.”
When did you become interested in AI and when did you consider it something you wanted to make part of your life’s work?
I’ve been interested in it since I was a little kid. Probably from sci-fi movies. I’m a big sci-fi nerd. I read I, Robot by Asimov and watched a lot of Star Trek, and I became really intrigued with this idea of, “What if we could design an intelligence?”
I think if you ask most people who are studying computer science in college, there are two broad categories they want to work on. It’s either AI or video games. I definitely wanted to work on both, but I didn’t start taking AI seriously until Evernote. Inside of Evernote, we were really trying to build practical applications of AI, and we thought that the “A” shouldn’t stand for artificial. We thought of it more as augmented intelligence.
The challenge at Evernote was making it so that the computer + the person was much smarter than the person by themselves. The question was, “How do you augment natural human intelligence with these other concepts?” That’s stuck with me, and I’ve been working on that ever since. So really, I’ve spent the past decade in this space.
“We thought that the ‘A’ shouldn’t stand for artificial. We thought of it more as augmented intelligence.”
What do you think makes you different from other people who were interested in AI as children? What made you believe it was actually possible?
I gave up trying to be gratuitously different from other people a while ago. Even in my Evernote days people would ask, “How’s Evernote different from these other note taking apps?” I don’t know. I didn’t know what the other ones were, I just knew what we did, and we were just trying to be really good. It’s hard for me to speak about other people’s motivation.
I was never interested in the anthropomorphic aspects of AI. I remember learning about the Turing test and reading Turing’s original papers in college and thinking, “Yeah this is interesting but it sort of seems like a wasted effort.”
Somehow the AI industry got into the direction in a really early stage where they believe the ultimate expression of AI is making a machine that can pretend to be a human or fool a human into thinking that it’s human. I always thought that was a weird goal. Super hard to achieve, but more importantly why do it? Like, who cares? Why do I want something to pretend to be a human? There are enough humans in the world.
I was much more interested in the question, “How we can make stuff that’s better than humans but also very different?” It doesn’t pretend to be human, it’s just better. I’ve focused my work on making very practical applications of this technology rather than trying to make things that pass Turing Tests because I think it’s much more fun, and much more useful too.
“Somehow the AI industry got into the direction where they believe the ultimate expression of AI is making a machine that can pretend to be a human or fool a human into thinking that it’s human. I’ve always thought that was a weird goal.”
Identifying terms like “Artificial Intelligence” and “Bots” can be a distraction to people can’t they?
Yeah, I spend a lot of time thinking about the personality of bots and design of these AI systems. I think anthropomorphizing them is not a good idea for the vast majority of products. People make a fundamental mistake when they try to make them look and feel and sound human.
There’s an obvious example from Star Wars. There are two droids in Star Wars. There’s C-3PO and R2-D2. R2D2 is the good one. It doesn’t pretend to be human. No one is ever confused about whether or not it’s a human. It’s amazingly good at its job. It doesn’t speak English, it beeps and boops, and people just understand it. It’s something that has a personality, but it’s specifically task oriented around a particular thing. It’s approachable but it’s also very clearly not human.
Then there’s the bot that does pretend to be human, C-3PO, and he’s the comic relief that everyone loves to hate. I laugh because even in Star Wars they got this right! So I try to not repeat that lesson.
“I think anthropomorphizing them is not a good idea for the vast majority of products. People make a fundamental mistake when they try to make them look and feel and sound human.”
Why did you name your new company “All Turtles?”
It’s an unusual name, which I like. It’s a very cognitively sticky, and it’s a name I’ve wanted to use since Evernote days.
The original reference is, “Turtles all the way down.” It’s an apocryphal story that was first told to me via Carl Sagan. The general story is that some important scientist is giving a lecture to a big auditorium about the structure of the earth and the solar system. After the lecture, everyone claps, but then somebody gets up and says, “What you said is all wrong. Everyone knows that the earth really rests on the back of a giant elephant.” The scientist then asks, “What is the elephant standing on?” And the person says, “The elephant is standing on the back of a giant turtle.” The scientist asks, “Then what’s the turtle standing on?” The person says, “Oh you’re very clever, but you’re not going to catch me that way. Everyone knows it’s turtles all the way down.”
The name also comes from the idea that whatever you build, you’re standing on the shoulders of the people who came before you, and you’re supporting the people that come after you. I like that analogy. Ever since I first heard it as a kid, it stuck with me. Also, since elephants stand on turtles, it’s the obvious sequel for me after Evernote.
“It comes from the idea that whatever you build, you’re standing on the shoulders of the people who came before you, and you’re supporting the people that come after you.”
When and how did you realize the current system of turning ideas into products is broken?
Well, I won’t say it’s broken. I just think it’s not enough. It’s inefficient. It’s a good start, but we shouldn’t be content with where we are. We should try for a lot more. I’ve spent 20 years in the startup ecosystem as the founder of companies and as an investor, and I’ve had a growing realization that there’s a massive inefficiency.
The Silicon Valley style of how we fund innovation works great, sometimes. But actually it rarely works because it leaves a lot of value completely unexplored. I’ve been thinking about this for a couple of years, and in the past 6–9 months, I’ve decided it’s time to do something about it. Plus, the fact that the AI Revolution is currently happening and is changing the way all products are made, this feels like a good opportunity to not only make a new generation of products, but to also change the way we make products in general.
“It’s a good start, but we shouldn’t be content with where we are. We should try for a lot more.”
It’s just like 2007–08 when the mobile revolution was happening. When we started Evernote, it was all about mobile. All sorts of new products were becoming mobile focused, but it also meant that we could reinvent the very nature of how we built things. All sorts of things that we now take for granted, like A/B testing, agile development, etc were very much fringe processes then. The mobile revolution entrenched those as best practices. Now that’s just how we build things. So every time there’s a big platform revolution, it’s also a good time to screw with the underlying structure.
The two things I’m most excited about are practical AI products and trying to make a more efficient innovation structure. Chocolate and peanut butter; let’s combine them.
“Every time there’s a big platform revolution, it’s also a good time to screw with the underlying structure.”
What are the key things that sparked the “AI Revolution?” Why is it happening now, rather than three years ago or two years in the future?
That’s the most important question. What I’ve tried to do throughout my career is get that timing right. I try to see the time when something important goes from being almost impossible to becoming just within reach of mere mortals.
I began building cell phone apps before the iPhone, but the difficult part of building mobile apps in 2000 & 2001 was very much a technology problem. Then, in 2007–2008, it immediately flipped to where it was still super hard to do, but you just knew that over the next few years it would rewrite the whole industry.
“I try to see the time when something important goes from being almost impossible to becoming just within reach of mere mortals.”
The successful mainstream apps that came out after the iPhone weren’t successful because they were deep, theoretical technology. They were successful because they were made by the people who were most in love with the problem. The axis shifted from being about technology to being about design. Right when that happens, right when some fundamental new platform goes from being academically interesting to mainstream interesting is the time to have it as a theme and to build it. I think that’s what’s happening with AI right now.
“They were successful because they were made by the people who were most in love with the problem.”
A lot of the individual building blocks have been in development for twenty years. Text-to-speech, speech recognition, NLP, machine learning. All of this has been around forever, but in the past two years, it has gotten good enough. Now there’s a basket of technologies, and there are APIs on top of them, that mere mortal developers can access. You don’t need a team the size of Google to pull these things together and assemble them into an actual product. What you need is deep understanding of the problem you’re trying to solve.
This is my fourth time around of saying, “Ok, I think we’re calling this timing as the ideal timing to work on this.” Hopefully we get it right.
What stops you from investing in a product, even if it’s a great idea?
The hypothesis of All Turtles is that we can expand that filter. Right now, if you’re interested in investing, you have to select people you think would be great at running companies. Especially with traditional VCs, each of these companies has to have a shot at being a multi-billion dollar outcome. That’s a very narrow filter.
I’d rather invest in the people who are going to be exceptionally brilliant at making the product rather than making a company. Product should come first. If you want to get into the VC tracks, you have to be amazing at making a product AND amazing at making a company. We’re relaxing that second requirement and doubling down on the first requirement.
Many more types of people should be able to create because we’re not selecting the same ten Stanford teams that everyone else wants to invest in because they’ve been reading term sheets since kindergarten. We’ll be able to get brilliant product people from all over the world to do this.
“I’d rather invest in the people who are going to be exceptionally brilliant at making the product rather than making a company.”
There are lots of things that would make us not want to work on something. In fact, getting into All Turtles should be really hard. It should be as hard as having your show made at HBO or at Netflix. We only want the best people, but when we say the best people we only mean the best people at making products, not necessarily the best people at building everything else. Many of these products will be companies, some won’t.
What won’t All Turtles invest in? Anything boring. Doesn’t matter how good of a financial idea it is. We only want to work on things that we have a high potential to fall in love with and that can impact the world in a good way. But there are many products, even enterprise products, that if done right, people can really fall in love with.
“We only want to work on things that we have a high potential to fall in love with and that can impact the world in a good way.”
I’ve done a bunch of thinking on what makes a great product founder, and the best answer I can give is that a great product founder is someone who’s in love with the problem rather than the solution. They know more about the problem than anyone else on the planet and have all sorts of practical ideas about how that problem could be solved, but yet they aren’t tied up on any particular technology or solution. This is very much a creative enterprise because it’s about people who want to create. You can’t take the art out of it.
Both of my parents are classical musicians, and I’m the only business person in the whole family. Going back many generations, everyone is either an academic, artist, or musician, so I grew up with a lot of people who were shockingly good at what they did. They could sit down at a piano or pick up a violin and do amazing things, so I’ve been around breathtaking excellence my whole life.
“This is very much a creative enterprise because it’s about people who want to create. You can’t take the art out of it.”
Brilliance is rare. I’m not saying everyone in the world gets to be an artist. If you literally take the one-in-a-million best product visionaries, that means there are 7000 of them on the planet right now. There are 7,000 amazing, Mozart level geniuses out there! Right now, how many of those actually have a chance to make their product and vision real inside something like the Silicon Valley VC startup system? Maybe 100? What about the other 6900?
What if we can create a structure that lets half of them create? We can literally 100x the number of amazing products in the world and the number of exceptional people that actually get a chance to make them. That’s a big impact.
“We can literally 100x the number of amazing products in the world and the number of exceptional people that actually get a chance to make them. That’s a big impact.”
This has happened in other industries. Think about writers. There are 7,000 people on the planet right now who are true geniuses at writing. How many of those people can actually write and have an audience? Today, a pretty high percentage. Twenty years ago, it was a very low percentage. The ones who happened to be in New York or Paris and had access to a literary agent could do it.
Now, because of the internet and blogs, etc, most of those brilliant writers are probably writing something and getting the world to read it. That’s new. That’s only been true for the past decade. I just want to do the same thing for products. Of the 7000 amazing product people in the world, I want 50% of them to actually make a product, instead of the 0.5% that currently get to do it.
Follow Phil on Twitter, and join the Bots group on Facebook to meet 24,000 Bot developers, entrepreneurs and serious enthusiasts.
Click the ❤ below to recommend Phil’s words of wisdom to other Medium readers exploring AI and product creation.
(This interview is part of the “7Q Series,” a weekly interview series brought to you by Chatbots Magazine.)
