Twenty-five years ago, at the start of the dot-com era, few could have imagined the internet underpinning our daily lives the way it does today. And a decade-plus ago, at the start of the mobile era, few could have imagined how inseparable we’d become from our smartphones. Phil Libin, CEO of All Turtles, believes that artificial intelligence will become similarly essential to our lives over the next several years. In this episode of Mastering Innovation on SiriusXM Channel 132, Business Radio Powered by The Wharton School, Libin describes his company’s approach to identifying problems that can be solved for the first time with this technology.
Despite decades of immersion in the world of West Coast venture capital, Libin is skeptical of what he calls the “Silicon Valley startup VC treadmill.” Sometimes it works spectacularly well, he says, but it’s important to remember that “it’s just one of many possible ways to make innovation.” With All Turtles, he’s set out to identify AI solutions with a disciplined approach that’s more typical of an established firm than a small company. In doing so, he’s challenging the popular idea that it takes a startup to make paradigm-changing innovations.
An excerpt of the interview is transcribed below. Listen to more episodes here.
Phil Libin: At All Turtles, we’re trying to build products that are high-impact in a way that’s more efficient than the typical Silicon Valley startup VC treadmill. The idea that conflates, in the technology industry, innovative products with small new companies, is, I think a silly idea that’s inefficient and bad for the world. So, we’re trying to do it better.
Saikat Chaudhuri: Give us some examples.
Libin: The way that people expect to make innovative stuff in the tech industry is to make a small company, make a startup. This has been unquestioned wisdom for the past couple of decades. It has been around forever. There’s this concept that if you want to make a product in the tech industry, there’s basically two types of organizations you can work for. You can go and work for a big company, most of which are perceived to be relatively professional. They know what they’re doing; they’ve got discipline; they’ve got process; they’ve got distribution. But they’re slow because the process is set up to protect their existing assets, so it’s hard to be entrepreneurial. Or you can go and be entrepreneurial, but then you have to make a startup, and so it’s a flaming clown car, but you get to do what you want.
Neither of those approaches is necessarily the right approach for making innovative stuff, and other industries don’t do it this way. We’re just stuck in this world, in tech.
What if we could make tech products the same way that Netflix or HBO makes original content? Netflix makes really original innovative and creative shows with really creative, entrepreneurial people, but they do it on time, on budget, repeatedly, all over the world. We’re trying to bring some of the discipline that we find in other industries into technology and say, “You can make innovative products without having to make tiny companies first.”
We’ve been on for about two years, so this is an experiment in changing the structure of how things are made. We’ve got a couple of good results, but it’s too early to tell for sure. We’ll know in a few years whether this is a pipe dream or not.
Chaudhuri: I love this. It really resonates with me, what you’re saying, because the Mack Institute decided that we don’t want to focus on just the startup and venture creation process. Colleagues at some other universities do that really well. We want to focus on these big companies and how they can be more innovative. What you’re trying to do is essentially take the best of both worlds, in some sense. Are you able to share an example of, perhaps, one venture or idea and how you’re applying that discipline while maintaining entrepreneurial freedom and not succumbing to the bureaucracy of decision-making, as we often find in the established players?
Libin: The first product that we created, released, and that’s out in the market is called Spot. It’s an AI for workplace harassment and discrimination reporting, found at talktospot.com. Like all of the things we work on, it meets four criteria. It solves a real problem, that’s the first step. We only want to build things that solve real problems. Way too many startups, in particular, aren’t trying to solve real problems, because they don’t necessarily know what problems are; they’re chasing opportunities. We say, “Okay. First thing is, what is the actual problem?” Point to actual living people or companies that face this problem today and quantify the impact of the problem. Obviously, harassment and discrimination are real problems for lots of companies and lots of people.
The second criterion is that we need to be able to have top founders, some of the best experts in the field, to work on this, which is hard to hire into a startup. The third criterion is that we only do things that have direct revenue models. So, nothing indirect, no advertising. The only acceptable business model for us is that people or companies pay us money to use our product because they want to.
“The only acceptable business model for us is that people or companies pay us money to use our product because they want to.” – Phil Libin
Chaudhuri: That’s the good old-fashioned way, right? That’s sort of how we’ve always done it, historically.
Libin: Yes. And I think a lot of what’s becoming broken about society started getting broken when we moved away from that direct revenue model. As soon as you have indirect revenue, all sorts of problems happen.
The last [criterion] is timing. We only want to build things that we can bring to market in 12 to 18 months. We’re not working on long-term science projects — not that there’s anything wrong with that; I’m very happy that other people and companies are working on long-term things. It’s just that the structure we have is set up for 12 to 18 months to market.
In a way, that approach would have been impossible had we tried it three years ago. The question is, why can we build something in 2019 in a mainstream way, but it would have been ridiculous to try it three years earlier? Usually the answer to that is, “Well, the technology’s gotten better.” Something in the tech platform, usually in what’s called the AI stack, has improved, and now we can do something that before would have been ridiculous. Then we develop it in a disciplined way; we recruit people; we have customers; we have revenue. So, the first one is Spot, that’s out and launched and doing great.
The second one, just came out a few weeks ago, is called Sift. It’s in the Apple App Store called Sift News Therapy. It’s a product that’s meant to combat the problem that so many of us feel around news anxiety, burnout, and depression. It’s actually a health app, a meditation app, but rather than disconnecting, it’s trying to connect you to contentious topics in a way that rewires your brain to be able to experience them more calmly. We have several other projects, and almost everything is in this space of making people or companies or culture healthier. We really believe in, basically, health apps. None of them are traditional health apps; they’re health of companies, health of meetings, mental health, things like that.
Chaudhuri: Yes. So, there are are both individuals and organizations at some level that you touch. That’s fascinating. Now, why the focus on AI-based solutions?
“It sounds stupid because everyone knows that it’s going to succeed so radically, that in a few years you’ll have to stop saying it because it will be part of everything.” – Phil Libin
Libin: AI is a particular kind of buzzword; it’s what I’d consider a “load-bearing” buzzword. It’s a buzzword that sounds kind of stupid, but for a very good reason. It sounds stupid because everyone knows that it’s going to succeed so radically that in a few years you’ll have to stop saying it because it will be part of everything. It will embed itself into the fabric of almost everything. This happens once or twice a decade. The first one of these that I remember at the beginning of my career, one of these-load bearing buzzwords, was “dot-com.” I started my first company in ’97, and back then everyone was running around saying dot-com. You can shake any tree in Boston or Silicon Valley and say dot-com three times. We said dot-com a lot. And everyone knew that it was a buzzword, that in a few years you wouldn’t say dot-com anymore, because everything would be an internet company, which was true.
Then, 10 years later, when I was starting Evernote, that same kind of buzzword was “mobile.” Everyone’s running around saying, “Mobile. Mobile. Mobile. What’s your mobile strategy?” No one says mobile anymore, because everything is mobile. The next one of those, 10 years afterwards, starting maybe, 2017, 2018, is “AI.” It’s the same kind of buzzword. It’s starting to grate on the ears, it’s starting to sound a little bit stupid, but only because we all know that within a few years AI will be everywhere, and so you won’t really say AI anymore. We try to find these kinds of giant platform shifts and focus on what kind of existing problems can be solved for the first time ever because of this new technological advance.
About Our Guest
Phil Libin is the co-founder and CEO of All Turtles, a company that builds AI products with offices in San Francisco, Tokyo, and Paris. Phil founded All Turtles in 2017 to create products that address real-world problems using AI. Prior to All Turtles, Phil was a Managing Director at General Catalyst, a leading venture capital firm, where he is currently a Senior Advisor.
Previously, Phil was the co-founder and CEO of Evernote, growing the product to hundreds of millions of users. He also co-founded and served as CEO of CoreStreet, which was acquired by ActivIdentity (now owned by HID Global) in 2009. Earlier, Phil was co-founder and CEO of Engine 5, a leading Boston-based Internet software development company acquired by Vignette Corporation (VIGN) in 2000, where he went on to serve as principal architect and chief technologist for applications. Phil studied computer science at Boston University, where he is currently on the Board of Overseers.