Itamar Arel jokes that he’s a “recovering academic” who’s advancing his AI research in the place where he can make the most impact: Silicon Valley.
Just four years after leaving university life, Arel founded AI-focused startup Apprente and sold it to McDonald’s Corporation. Today, as head of McD Tech Labs, he’s helping the corporate giant imagine the future of fast food with its own AI research lab in Mountain View, Calif., just down the road from the Stanford AI Lab where Arel once worked as a visiting professor.
Arel sat down with us to discuss his journey so far, his advice for founders of data technology startups, and what it’s like to work with VCs.
Russ: How did you first get interested in tech?
Itamar: As a kid I was always fascinated by technology. I love getting machines to do things that are intelligent, making them operate autonomously in some setting. That’s always done it for me, and it continues to be the case.
I grew up in Israel and got a PhD in computer engineering in the early 2000s before AI was on anyone’s radar. That’s where I wanted to focus my research. I spent 11 years as a professor at the University of Tennessee in Knoxville researching things like deep learning, reinforcement learning and hierarchical architectures — topics that are sexy and fashionable now but at the time got very little attention. It was a very small community of people who were funded primarily by government research grants.
I took a sabbatical from 2013 and 2015 to research at the Stanford AI Lab and got to work on a significant DARPA project. At that time it became really clear that most of the exciting applied research in AI was shifting outside of academia to tech companies like Google, Facebook and Twitter. It was clear where things were headed, so in 2016 I moved my family to Silicon Valley.
Russ: What problem did you set out to solve with Apprente?
Itamar: We started out as a very small team of mostly PhD’s like myself. Our vision from day one was to build voice-based conversational AI agents for enterprise. This is in contrast to consumer-facing products like Alexa that have been taking off for years. We felt — and still feel — there’s a lot of catching up on the enterprise side.
It’s obvious to me that 10 years from now, every airport, train station and hospital is going to have stations where you walk over and say, “Hey, how do I get the Concourse C?” and they explain it to you. It’s going to be everywhere. But for some historical and technical reasons, the enterprise side of these voice-based customer service agents has not evolved nearly as fast as the consumer side.
Russ: You pivoted to focus on drive-thrus early in your journey. How did that come about?
Itamar: Initially we were focused on call centers. We pivoted very early on, within the first few months, when we came across an opportunity for automating drive-thru ordering processes in the quick service restaurant (QSR) space, which includes McDonald’s. It’s a huge market. It turns out that across that whole industry, over 70% of the revenue comes from drive-thru, which is mind blowing.
We worked with our investors to arrange meetings with the top 10 QSRs, and everybody said the same thing. This was an “aha moment.”
Russ: Looking back, that proved to be a key decision that led to your success.
Itamar: Absolutely. That’s a big lesson we learned. A startup has one chance to get it right, so focus is critical. It’s critical to go after one thing initially and build on that success. We thought this was a big enough opportunity to go after because of the feedback from the market, and it was focused enough that we could really dig in.
Russ: What challenges are you addressing with your technology?
Itamar: The good news for us is that ordering at a drive-thru involves operating in a much more restricted domain than the consumer space. People going through a drive-thru only talk about one thing. It’s much less open-ended than Siri or Alexa, where you should be able to ask anything about thousands of domains from tomorrow’s weather in San Francisco to booking a restaurant and looking something up online.
The challenging aspect is that, unlike Alexa, we can’t teach people how to speak to our system. I joke that what Amazon did better than anything was training us to speak Alexa in a way that we know it’s going to understand us.
We’re not going to change the way people have been ordering at a drive-thrus for 40 years. We’ve got to pick up on the million different ways people have for ordering and modifying their orders and account for the noisy environment and other factors. We felt that was a big and interesting challenge to address.
Russ: Was this your first startup experience?
Itamar: I was involved in several companies throughout my academic career. The first one was with my PhD advisor, making use of the first machine learning algorithms for packet switching solutions. We raised a lot of money and had an exciting ride, but ultimately the company failed.
Russ: What did you learn from that first failure?
Itamar: We made the mistake that I think a lot of first-time founders make: We fell too in love with the technology. We drank our own Kool-Aid to the point where we failed to pay attention to the market. We needed to really ask ourselves, “Are we building something that is really answering a need?” We believed we were, but the feedback loop with the customers was not there.
The lesson we learned was that, even if you have the secret sauce, it’s critical to make sure you’re answering a need out there in the market. You’ve got to keep a finger on the pulse and always seek that market feedback.
Russ: What does that look like, keeping a finger on the pulse of the market?
Itamar: When you present to VCs, you don’t want to just present your vision, your technology and what you want to do. You want to be able to say, “We spoke to XYZ company and their feedback really helped us refine the technology, so this is where we’re going.” If you can show investors that you’re testing the market and already engaged in those kinds of conversations, it can go a long way.
It’s not just about technology or changing the world. Founders need to demonstrate they’re addressing a real market and that they’re already connected to it.
Russ: What’s the hardest part of running a technically focused startup?
Itamar: I think that one of the most challenging aspects of building an AI company today is finding good people. There’s just so many opportunities out there, so many startups that have interesting stories and are backed by good investors.
After leading several companies, it’s very clear to me that the first 10 or 15 people set the cultural tone of the company. This core group of people need to be super talented and buy into the dream, and they need to be culturally aligned, too.
We interviewed more than 80 people to make our first 10 hires at Apprente. There were instances where we interviewed people who blew us away with their technical skill but we could tell from the first or second conversation that it wasn’t going to be a cultural fit.
My rule of thumb for hiring: If it’s a “maybe,” it’s a “no.”
Russ: What’s your secret for recruiting and keeping top engineers?
Itamar: Good engineers are driven by the vision of the company. They ask themselves, “Is this an exciting story that I can be part of?” It’s usually not about becoming a millionaire or building the next Google. It’s driven by the excitement around what you do, wanting to work in a small enough environment where you get to shape things.
We had people take big pay cuts from the big 10 tech companies because they felt stuck in myopic roles at huge organizations. They wanted to be in a smaller setting where they could influence things and get exposed to different topics.
Founders should look for these signs during interviews. If people join for the right reason, it’s more likely that they’re going to stick around and be successful. If they don’t want to work for a 100k-person company now, that’s unlikely to change.
Russ: What have you learned about fundraising?
Itamar: Especially in the early stages, investors are betting on the people. The technology needs to be there. The pitch needs to be there making the case for the market you’re going after. But at the end of the day, especially early on, an investment in a startup is an investment in the people.
Investors need to be convinced that the team will work well together. Even if they lack experience, investors need to know the team will be able to grow and acquire the experience needed to be successful.
Very few founders can say, “Our vision from day one was the same as the one we went to IPO with or got acquired with.” That very rarely happens. You usually have some idea and you end up 30 degrees to the right and left. Investors want to know that when that pivot does come, the team has more than the usual small chance of success because of their talent and the way they work together.
Russ: How do you view the role of CEO?
Itamar: In my experience the CEO role comes down to three things: First, You’ve got to raise money to keep fueling the ship. Second, you have to provide a vision for where you’re taking the company. Third, you have to hire the right people. For the initial phases, those are the key areas. Of course, when you’re just a few people, you’re also the kitchen cleanup person, the admin and all that, too.
Russ: What’s it like working with Morado Ventures?
Itamar: They are awesome people, super professional and helpful.
Part of what Henry (Sohn) does as a venture partner is embed part-time in companies to help them get their first business deals, their first wins, and help them maneuver the business landscape. I can’t say enough good things about how helpful that was.
For our first year and a half we were engaged with many of the top players in this space, and Henry was with us every step of the way. He helped us through every stage with McDonald’s, from the initial conversation, to proof of concept, to a pilot and then negotiating an agreement with IP and all the elements involved with that. Morado really was a wealth of knowledge helping guide us through all of that.
Morado also allowed us to incubate out of their offices, which freed us up to focus on building the business and the team. I can’t thank them enough for that. I know we’re not the only ones to do that.
Beyond all of that — and the funding, of course — Morado is awesome at opening doors. A good investor as a partner is one that doesn’t just give you money and disappears but really supports you throughout the journey. That’s exactly what we got from Morado.
They also fully supported both of our funding rounds, so they’re there with you all the way.