Thomas Copeman and Dr. Nathan Wilson are something of an odd couple. As I sit down to join them in a Brooklyn restaurant, the two business partners do nothing to play against type. Copeman, CEO, in a flannel and with a beard flecked with grey, has the easy confidence of a serial entrepreneur (he conducted successful turns at Lululemon Australia and BodyGlide). Wilson, CTO, younger and soft-voiced, seems like he’d be more at home with one of the more affable cast members of “The Big Bang Theory” (Wilson did graduate work in neuroscience and computer science, and recently published a paper in the foremost scientific journal, Nature). I caught up with Copeman and Wilson to learn more about their company Nara, which brings serious algorithmic firepower to the challenge of personalizing the web. Nara launched a thorough redesign of its website yesterday, and expanded its restaurant recommendation engine from eight cities to 25.
FAST COMPANY: What is Nara?
NATHAN WILSON: Nara is a next-generation personalization platform that we believe is the first one that can do Pandora on all things. Nara tries to build a neural network that approximates how your own brain thinks and connects things together.
For now you do restaurant recommendations. But you talk about being the next way we interact with the web. Are you the next Urbanspoon, or are you the next Google?
THOMAS COPEMAN: We sit in between a Google and an Urbanspoon. We’re starting out with restaurants, and then we’re extending into other key consumer categories. We’re merging into hotels, and looking now to power wine recommendations.
NW: What we’re trying to do is nail the restaurant category, then nail hotels, and so on, and make an expert system on each one, and then link those together. Then we can see that people who stay at this hotel go to that restaurant. It’s silos that are interconnected, like the brain. The brain is a collection of silos: there’s a seeing brain, a hearing brain, a smelling brain, yet it’s all connected.
Pandora has enough trouble being the Pandora for music. How can you take on multiple categories?
NW: Music is a hard problem, so hats off to them. They have to come up with a new song every few minutes. What we’re doing is saying here’s a restaurant for tonight, and one for next week–it’s an easier problem in that regard.
TC: We want to turn searching into finding. We want the Internet to work for you, and not the other way around. With Nara, you sign up, you tell it what you like, and it gets to know you. Then it goes out on your behalf and brings the web to you. That’s where it becomes a personal Internet platform.
You use the phrase “neural networks” when discussing Nara. What’s a neural network?
NW: The essence of a neural network is it’s loosely coupled nodes with lot of connections. “Loosely coupled” meaning each node is autonomous, yet highly connected.
Nara maps the Internet like a neural network–nodes with lots of connections. What are the nodes, or “neurons,” for Nara?
NW: The neurons are the things we’re trying to personalize: the restaurants, wine bottles, and things. The beauty of network architecture is that if you build a rich network for restaurants, and a rich network for hotels, and then link these together sparsely, all these maps can be linked together–and you can look at the pattern found by activating parts of the network, and see what else lights up. We’re trying to find patterns just like the brain does.
Only loosely joining the restaurant network with the hotel network is enough to light up the network as a whole?
NW: We think so, because if you have a really good map of restaurants, it’s almost like you’re in a set “vibe,” as Tom would say. Once you can make it over to the right vibe of hotel, you’re kind of set.
I knew someone who would walk into a café, use a kind of Spidey Sense to check the vibe, and then would walk out after a moment if he wasn’t feeling it.
NW: Nara walks into every door for you and checks the vibe.
TC: I’ll walk into an environment and say, “I can’t hang here, I have to leave.” I don’t like to go to pubs. I don’t drink beer. But we had somebody out at Nara who had to go out on St. Patrick’s Day. I used Nara to find a pub, and it worked for me. I was in a place where it worked for me, even though I didn’t really want to be in that genre.
Nathan, how did Tom lure you from your academic track?
NW: I was all geared up to be faculty, had paid my dues, when Tom derailed my life, in a way. I’m a believer that great breakthroughs happen with focused challenges. In academic research in AI, everyone’s working hard, but there wasn’t that focused challenge for AI at the right level of difficulty. Having to land on the moon is a challenge–we can conceptualize that–versus just, “We’re going to generally study propulsion.”
How did people react when you left academia?
NW: People say, “I hear you left science, that’s so sad.” I think, “Wait, why do we have to ‘leave science’ to connect to the outside world?” Fundamentally I’m a researcher, and I saw Nara as a multiplier, and opportunity to do better research than in a university. Having the focus and the drive, that’s going to lead to a faster breakthrough.
Nara seems to have pretty big ambitions.
TC: It’s wildly ambitious. This is my third company. I want to make a significant mark, a contribution to the web going forward. I want to be pioneering the next evolution of the web.
NW: We do swing for the fences.