← The Digital Executive
The Digital Executive
Building Emotionally Intelligent AI: Ritu Mehrotra on Mental Health, Innovation, and Stella’s Global Impact | Ep 1102
00:14:45
Loading summary
Transcript13 lines
- [00:00]
A
Foreign welcome to Coruscant Technologies, home of the Digital Executive podcast. Welcome to the Digital Executive. Today's guest is Ritu Mehrotra. Ritu Mehrotra is a tech leader, mental health advocate and global business strategist with two decades of leadership across companies like Booking.com, zomato Mahindra and more. She is the founder of United We Care, an AI powered mental health platform and the strategic mind behind Shunya Labs, its deep tech spin out building world class AI infrastructure for voice and reasoning. Under her leadership, United We Care grew from a social impact startup to a global force in AI wellness and in doing so incubated breakthrough technologies now commercialized through Shunya Labs which with a career spanning consumer, tech, wellness and B2B innovation, Ritu brings a rare blend of empathy, scale and system level thinking to the future of human AI interaction. Well, good afternoon Ritu, welcome to the show.
- [01:06]
B
Thanks a lot Brian for having me on the show.
- [01:08]
A
Absolutely my friend, I appreciate it. And hailing out of the great Silicon Valley today. I know you're near Palo Alto there doing a ton of podcasts out of that space there, but I'm in Kansas City so we're just a couple hours apart from ritu. If I could, I'm going to jump into your first question to get the conversation started. From Booking.com to Samado to Mahindra with role spanning APAC, leadership growth and corporate development, how do these diverse experiences shape your vision? When founding United We Care.
- [01:39]
B
So Brian, I've always been fascinated by how technology can solve large scale human problems. And each of my roles taught me a very different lens if you will. So at booking I learned the power of customer centric data driven scaling across cultures. You know I was heading 30 plus countries within APAC and it was multibillion dollar business so you know, a very large scale operation. And Zomato I saw how speed and agility could shape entire consumer habits, shape or reshape rather right because we were building the new habit culture. And Mahindra I understood the discipline of building sustainable long term businesses and United BCare was born at the intersection of all of these. So agility, scalability and most importantly deep responsibility, all of these were applied to a mission because which was deeply personal and the idea was that can we make mental health support accessible, affordable, stigma free for anyone anywhere. All of these actually came in beautifully together to kind of solve a large global personal mission.
- [02:55]
A
Thank you, I appreciate that. And I know that you know starts somewhere. You had a passion, you talked about Being fascinated how technology really being able to provide human solutions to the world. You took advantage of that, you followed your passion and you had a ton of experience at working some of these big companies like booking.com and Mahindra. And of course that led you to where you are today. Building a phenomenal platform to help people across the globe in this space to make the world a better place, as I like to say. So I appreciate that. Ritu Shunya Labs was launched as a deep tech spin out focused on advanced AI voice infrastructure. Could you walk us through the the origin story, what gaps in the markets you saw and how the team engineered solutions like asgr, stat gap and the clinical knowledge graph.
- [03:40]
B
So Ryan, I'm a cancer survivor and you know, after I was recovering cancer, you know, I went through my own deep mental health issues and what I realized was that obviously on one side there is humongous clinical shortage, but technology was really not as sophisticated. Right? When I want to interact with technology, technology has to be sophisticated enough to be able to understand my deep issues, what I'm communicating. So shunulab literally started with frustration. We realized that you know, everyone was chasing flashy AI demos, but the core infrastructure like voice recognition and voice recognition, especially for multilingual nuanced clinical use cases, for example, if you know I am talking about my deep mental health issues, you know I would like to talk in my own language sometimes mix of languages which was completely broken, that was one. The second thing was that technology was expensive and we are a startup, we couldn't afford that technology which is not as sophisticated and will be put to good use in the whole infrastructure space. So our platform needed speech to text that could work in multiple languages, understand, you know, multitude of dialects, handle code switching and maintain clinical grade accuracy. And as I mentioned, the existing models weren't as good enough. So we built our own ASR which is automated speech recognition that runs on cpu. Because our AI therapist coach, we affectionately call Stella Stella actually runs at the power of your phone. So it's a real time lip sync sync and you can actually make real time conversations at the power of your cell phone bandwidth of your cell phone. And we beat global benchmarks. So if you look at any of the automated speech recognition benchmarks, which is either European languages via Europearl or Libre speech or any of these kinds of benchmarks, we broke all benchmarks and it can work in low resource settings so that we can make it affordable and to make millions of people. And then came our real time speech to graph AI understanding context and our clinical knowledge graph that connects symptoms, emotions and interventions to give you or a user, you know, a clinical pathway. That was the whole idea. So we were building the missing rails to innovation so that it could reach millions of people really at the accuracy levels, at the precision levels and at the convenience levels it needs.
- [06:15]
A
That's awesome. Again, another story. Sometimes stories can start out tragic or sad, and your cancer story obviously had an influence on what you were trying to accomplish here. Your journey was wrought with physical and mental health challenges and you saw there was a gap in healthcare. And I sought myself in healthcare doing healthcare for 20 years. In the technology realm, Shunya Labs is built out of this frustration, finding a way we can leverage technology and better understand the patient and all the patient's nuances. And I love the story behind Stella and, and the fact that you have an accurate real time conversations, you know, using lower performers hardware while breaking top benchmarks, that was just amazing. And I certainly highlighted those out of your conversation there. Ritu, I'm going to move to the third question here. You said we didn't set out to beat the benchmarks. We set out to invent what didn't exist. You've said that about shooting labs. How do you maintain that innovative mindset while simultaneously steering towards commercial and scalable execution?
- [07:12]
B
As I mentioned at the time we were building United VCare, we just couldn't find technology that was sophisticated enough to be able to solve this problem. So we realized very quickly, Ryan, that innovation is not a department for us, it's a habit. The trick is to create a culture where the question isn't what's the market doing? But what should the market be able to do if constraints disappeared? And that's how you invent. If you look at our peers or, you know, people who are in the same industry, we are still at a stage where, you know, we're getting confused with, you know, if penicillin is allergic for a patient or is not allergic for a patient. And these are being confused by companies that have budget of multi billion dollars. We solve this, these kinds of problems, we are funded, but we did not spend billions and billions of dollars. But we used and utilized a differentiated approach. And that approach was while all the other people were feeding billions and trillions of data sets into the reasoning models, we said that, hey, can we actually infuse the data? Which is a bit noisy because, you know, it's a typical story of somebody taking a degree from KG to 12th grade. But if you continue to fill the same amount of information and expecting Your system to give you different results is not going to happen. So we had to work on the bringing the system from 12th grade to maybe a PhD degree. And for that you need differentiated, noisy, surprising data. And that's how we were actually able to accomplish and achieve what we did in a very small data set. And that was the reason a the kind of accuracy levels we broke or you know, our ability to be able to run on CPU purely because the data model is very, very small. So innovation without execution is just a good idea. And because we had a real use case for it to go on multiple cell phones where actually talk to the AI therapist or the coach, we used rapid prototyping, iterate with real customers and set clear commercial milestones and it's a constant balance, you know, moonshot thinking with feet firmly on the ground. And that's how we went from idea to breaking seven world records and securing 14 plus patents while also building a revenue positive business all these years.
- [09:35]
A
That's amazing. Love the story. Got a story for each question I have for you today, which is really cool. It was a struggle at first, you know, when you were building United We Care, trying to find that right and best solution initially was the challenge. But you sought out again. You and your team developed that innovation in your DNA, in your culture. And an example of this is how you're able to build a better platform on a fraction of the cost of the big players out there, the big investors. I really appreciate your leadership. You displayed more than just an idea. You, you broke world records and did a lot of great things while making a company profitable. So thank you Ritu. Last question of the day. United We Care's virtual coach, Stella has seen massive engagement, over 10 million conversations and high intent detection efficacy. How do you measure emotional depth or clinical effectiveness in AI and what's next for Stella's evolution?
- [10:27]
B
AI in mental health can't be just smart, it has to be emotionally intelligent. And if you look at who's philosophy around mental health it very clearly states the charter which says look, listen link. It doesn't say look, listen solved. You know if you look at the platform and we also know, we knew very quickly that it is not pure play AI or technology that can solve the problem. Idea was that if technology can take care of pre clinical subclinical needs then you know the clinician's time could be very well utilized for things they do the best. So we measure Stella's effectiveness across three pillars. Intent detection accuracy, which is where we beat intent detection on you know, multiple you know, leaderboards like Stanford Natural Inference, where VR number one and there is a complete benchmark analysis that, you know, you have to look at. Second is emotional resonance and clinical outcomes. So the number of people we suggest a particular clinician. And by the way, that also has to keep changing because clinicians specialize in one or the other therapies and an average cycle of a user may need multitude of these differentiated therapies and not just one. And that's why when you continue to see one clinician, it doesn't hit, you know, diminishing returns as a user because you now need to be able to look at something else. And that's where Stella actually comes into play, where we can actually knowing where you are in the journey, whether you want to look at programs. We have over a hundred programs around, starting from clinical depression to anxiety or social anxiety and so on and so forth. Also, can you pick I am fine what it might mean while, you know, you're saying that you're struggling and she can provide intervention that would lead to measurable improvements in the wellbeing scores over a period of time. So one is, how often are you staying on the paths that has been recommended and how far do you go? And second is, are your scores also improving over time? Idea is, can you stay in control of your emotions more often than not? Right. Because mental health is very, very complex. That's all on the user side, but also on the clinician side. Stella is far more deep because when we are giving recommendation to a clinician, we have drug, drug interaction, drug disease interaction, which is built on not just our automated speech recognition, but clinical knowledge craft. That's where, you know, we are also kind of giving the right kind of path for the user. And it's not just a scribing tool, but clinician is in the driving seat where you can actually accept, reject, you know, all the kind of recommendations that Stella could be giving you. And this flywheel actually gives us and helps us stay and detect accuracy and very, very high engagement rates across multiple user sets. And next, you know, we're expanding Stella's emotional vocabulary. It's going to be a journey, Ryan from that sense and enabling her not just to understand what said, but how it's said. Because a same sentence could be said differently in different parts of the world and could mean something very different and across cultures, across languages. So the experience a human feels and not just kind of helpful. So that's the journey we're kind of embarking on as we are sitting today.
- [13:51]
A
Thank you. I appreciate that. Sounds like Stella is very intuitive and I know you measure Stella on can. You mentioned a few things here, intent detection accuracy, emotional resonance and clinical outcomes. Of course, providing interventions that will put you on a quicker path to improve health really stuck out for me and I think that's important. But you know, having a platform that's top in class for speech recognition, it's got that deep clinical knowledge which clinician wouldn't use your platform. And I really inspired by what you've built because I've been in healthcare for 20 years and I love the healthcare space. I love what people like you to make the world a better place. Ritu, it was such a pleasure having you on today and I look forward to speaking with you real soon.
- [14:32]
B
Thanks a lot Brian. Thanks a lot for having me on the show as well and I look forward to seeing you again soon.
- [14:37]
A
Bye for now.