Rapid Growth of Robotics | Andra Keay

To work around them and not aim to be too smart. Mabu from Catalia Health has just rolled out their first commercial deployments. Again, this is a startup that I have seen coming from research through several prototypes through success. And in their first funding rounds, they had a lot, a lot, a lot of trouble because it was early days and Corey was showing investors Mabu. This robot with a screen. And Mabu is a health companion, okay? She doesn't talk about baseball or the weather. She is a she specifically. She has a she voice. She talks about your medication and side effects that you might be feeling. She does not report your conversation back directly. It's not like a phone call to your doctor. But tracking results. So if you fail to take your medication for 3 days, that's tracked. That kind of thing can go back. So the idea is that she's your health coach. She talks to you every day and tries to find out, are you feeling okay? Do you really need to call the doctor? Is there something— why didn't you take it today? Why don't you take it now? Why don't you go for a walk? A very, very specific set of scripted conversations. that evolve around each person that are medically guided and approved. Okay? But the first round of funding Corey tried to get, and he had to go through a second iteration with Mabu, his degree is in embodied robotics from the Social Robotics Lab. His advisor, Cynthia Breazeal, who's founded Jibo, who I'm sure you've heard of because she is like the expert of social robotics. So you come from MIT, you've done your PhD in social robotics, you're all about embodied robotics, you're going out there and you're saying, we know through clinical research that people have 70% more chance of staying on an expensive cancer treatment if they have a robot talking to them. And the investor says, yes, but how difficult is it to produce that? Look, you've got a tablet, why don't you build an app? I'll invest in your app. And Cory's like, because it's 3 times as good when it's a robot. We've proven this. The app doesn't work. And they're going, yeah, but it's a lot easier to build an app. I can see that. I can invest in the app. He redid his prototypes, and he redid his pitch. And he came back and he pitched Mabu, the embodied health companion, with a partner app, and the world was just a little bit more ready to adopt robotics. And he closed his investments, Khosla Ventures, and a few other investors. In fact, I invested in them as well. I've been a believer for a long time in that one. So for me, the lesson was, and it was something I'd argued with him in the early days, the investors are saying build an app. I know that it's not about an app. But can you hear what they're saying and incorporate that feedback into what you're doing? And it just made so much more sense then. And he buries the robot part as well. Because when you're selling this to healthcare facilities, they're all about healthcare. They're not about robotics. So this is a healthcare management plan, a subscription service. And in fact, it's now prescription. So patients who are undertaking an expensive cancer or some other treatment regime have this robot prescribed to them for a 3-month period or a 6-month period. So I think that is the business way forward for robotics. And just this year, one of the spinoffs— I mentioned Willow Garage. A lot of the people went to Bosch. Bosch have been doing a lot for robot operating system here in the Bay Area, and also for home robots, and self-driving vehicles. So they've been really working hard on a household robot. So they've launched Kuri from Mayfield Robotics this year. It was the hit at CES. I think some people have actually got Kuris in their homes. For me, this is very interesting because 3 years ago, I was a complete Jibo fan. I thought that that was the ideal combination of the simplest possible hardware that you could have with the most effective interactive nature. Just tilting and following ahead. I mean, that is— if you pick one key behavior to do, to make something seem alive that you're having a conversation with. It is that ability to track a person. Okay? So for me, I went, great, they're producing the first household robot because they're doing it really, really simply. And then of course you can make it more and more complex with software. And then Jibo didn't come to market. And it didn't come to market. And Amazon launched Alexa. And Alexa has gone gangbusters. And I'm like, okay, Jibo, I'm sorry, you missed the window opportunity. They've been trying to build Jibo perfect because the reputation was on— they needed to get Jibo out the door and into homes before Alexa dominated. Because right now you can add more physical hardware features to Alexa and you've already built up a huge consumer base. And you're building up your knowledge of behaviors. And so you can start to target that. And we've been arguing over these strategies for when you build a household robot. What do you do? Do you aim to be the general platform right from the get-go? Or do you target and pick one thing, like Roomba? They do one thing. They vacuum. Now, they missed doing all of those other more sophisticated things. But think how quickly things are changing. I mean, Jibo lost this battle. in just a few years. Mayfield was nowhere near ready to launch. In fact, didn't exist when Jibo was announced. And now they've produced the hardware and they've launched. So we're seeing some companies grow rapidly, very, very quickly. And we're seeing the problems if you don't launch, or if you launch and there's a lot of problems with what you do launch. So you have to be very, very smart at working out What you're going to do. You know, are you going to do one thing really, really well, or are you going to just be the first in and then work on fixing your problems? It's a hard call, but I've seen a good probably dozen home robots come by my come by me in the last two years, and Onky's Cosmo. the toy robot. Mayfield's Kuri and Amazon Alexa are the only ones I've seen in people's actual houses. The rest, they've had huge crowdfund campaigns, but they're still not out there. This is the Cozmo from Anki, and they've just expanded their— you know, there's SDK for this. So one of the things that they've done Which for me is a great industry-forward thing, although it's limited to the OnKey hardware, is they've got an SDK for this where they've short-coded a lot of the sophisticated behaviors so that anybody who can build apps for Android or Apple can dive straight into the Cozmo code, start controlling it, and start creating games and apps. You can have Cozmo as a home companion. You can have Cozmo playing with your pets. You can, you know, have it patrolling your house and giving you feedback if somebody comes to the door, potentially. Because this little guy has all the hardware that you need for face tracking, facial recognition, for natural language processing via cloud most likely, but for also autonomous navigation, a lot of sensing and so forth. And now they've released a children's coding package with it. So if you have one of these, this can also be how you start your Kids coding. You start with a language like a block language. You lead into Python, and maybe other scripting languages. So sorry, Pepper. I don't really believe you're the future, but a great avatar. And in fact, this is the primary role for these sorts of humanoid robots, is kiosk robots, attracting you into the shop or the airport and waving at you. Very difficult to hold a conversation with them, but they live up to our desire for the humanoid robot. So robotics is going great as long as your expectations are clear about what the state of the hardware and software is, how rapidly it's going to move forward in some areas, but how incredibly limited it still is in others. So there's huge opportunity. opportunities when you line up the right combinations of things.