Exploring AI's Impact on Healthcare and Fintech (Ghanendra Shrivastava)

Full transcript

Welcome back to the podcast, guys. Today I'm joined by Ganandra, the director of AI and engineering at Tribe Beam. Welcome. >> Thank you. Thank you. Could you tell us more about yourself? >> I am currently director of AI and engineering at Beam. Other than that, uh I am CTO of our own product known as Kokuro. Doctor. I have around uh 10 years experience in AI industry and I really love it. So I'm like as the years are passing I'm just falling in love with AI. I graduated from IIT RII that is one of the best Indian institutes and uh then I went to Harvard for a diploma course on cloud deployment and yeah now I'm here and trying to hustle trying to work on multiple projects >> and how would you say that the Harvard education helped you in your journey so far? really helped me I would say because uh the startup company I'm working on cocoro.docctor doctor that is incubated by Harvard Innovation Labs. So that was my first point where you know like my company was incubated. I was working on the project.

There was supremely friendly mentors and a lot of facilities. We got a office in Boston. So Harvard has played a really really good journey. I have my professors there, professor Jeremy V, Mike Grandinity, all of them like really really supportive. So Harvard has definitely placed a very good role for my journey. >> And what made you start your own startup? >> So when I was studying there, there was a cardiology based project where we wanted to help uh patients with having cardiology diseases. It started as a research project. We were trying to build the best LLM model possible for cardiologist. And when we worked on it, we worked together. Then it came out that we can actually commercialize this. we can actually make it something product heavy. So it started as research project now it's an actual product with around 10,000 users and around 30 plus doctors and it's been working you know like in a stealth mode yet but uh definitely we are looking forward to it. >> What is it specifically? You mentioned that you are a big lover in AI.

How AI comes into place in this kind of field? So basically what we do is we generate AI prescriptions. We take patients medical data. So their eco reports, cardiac reports, ECG reports, angography reports, old prescription. So user just upload whatever they have and our model takes in everything. It is very fine-tuned specifically for cardiology and it creates a prescription. So a doctor whatever a doctor creates it creates a end to-end prescription of symptom analysis what are the lab test happening and overall thing but as we are very responsible and we don't want AI to directly give something to patients and it's a very serious issue on cardiac so it's a doctor platform so we generate this prescription for a doctor and that doctor validates it and it finally reaches user so it's basically for doctors to fast forward everything and scale it up for more and more patients. >> Interesting. I would also like to touch upon beam since that is also part of your current profession. So what are the biggest trends that you are seeing in consumer fintech right now?

>> So I'm in beam as director of AI and engineering from last 1 to 1.5 year. I started as a consulting and supporting them and I really know the founder he really want me in the team. So here what we see as the biggest challenge that a US consumer needs credit and they are very good in credit building as well but there is nothing which supports the bottom 80%. All the loan schemes everything which is coming it's for the top 20%. So we specially focus on the bottom 80% and we try to help them with AI based modeling and AI based credit selection. So here what we are doing in AI specifically first we have built an AI agentic system to help all the users get financial budget planning overall understanding and also giving them instant cash which is qualified from our model. So we have built a proprietary model which sees that this person can repay and he has maybe he's not having a regular job but he still can repay and then we give it to instant cash of $100 $500 and uh right now we have around uh 10 million plus users and uh pretty good revenues and pretty good scale and this is a 8-year-old company. Uh so yeah that's a wonderful journey. >> And what made you go into fintech field?

So when I started my career, I started with City Bank. So it started with fintech and I completed CFA also like chartered financial accountant when I was on City Bank. So I was always close to finance. I only like two domains finance and healthcare and I want to contribute as much as possible in these two core domains. >> Awesome. Where do you see AI creating the most leverage for you? Is it in the personalization or in fraud detection or maybe customer support? >> If we talk about AI as it is right now, I think it can easily replace and contribute a lot in customer support area. But as you see that open AI and anthropic everyone models have been improving and becoming more and more intelligent. So they will start hitting on consulting services, they will start hitting on marketing campaigns, they will start hitting on different verticals. But I personally think that finance can have a very good AI oriented things because the amount of data in financial industry is just way too much right like if you just take the patterns, the shares, the equities and everything. So it's humongous data and humans cannot process that much data properly.

So I would say finance is the thing where you will see the use of AI increasing to a very high levels in future. >> Yeah, that makes sense. Looking back with hindsight, what's one big thing you do differently if you were starting your cocoor doctor? >> That's a good question. I would say like we wasted quite a bit time in overgineering of the product. So if you will see our product it's pretty much as huge website a lot of features but uh we did the PMF iterations later. So product market fit was the thing which we should have done initially and now we have figured out the product market fit which make us lose like 3 to four months. So that would I would say is the main thing but overall I think that AGI direction we should be building much more. So right now what we are trying to do we are putting whole of our focus on building a model which is aligned towards AGI.

Obviously in our current funding status we cannot build AGI but we can move in the direction of artificial general intelligence and we are looking at the health care models in a specific way where the healthcare reaches the artificial general intelligence and it can actually create impact in society. it can actually help doctors to serve more and more patients and health care I think is a very crucial domain where if if AI cracks it is not just about money it's about actual impact that it can create maybe in Africa someone needs a doctor and kokuro doctor can connect them with an Indian doctor so that it actually get executed with AI so that is a kind of vision but definitely we are looking forward to it >> I'm curious how do you actually gather these doctors to try your product. >> We have an onboarding team around five to 10 people. It's majorly based on cold calling. So we try to connect with doctors and either from references or cold calling.

And if we see that there is some kind of alignment, we try to get them on board, show them a product demo and if it makes sense and if it picks up then we make a conversation to the onboarding. I'm interested about the compliance and the data management. Since you are in healthcare, I suppose that it's very strict when it comes to collecting data and storing them. How do you manage that? So firstly, we took the all the regulatory compliances like HIPPA certification and in India there is an ABA certification. So these are the things which we first procured. Secondly, we want to make our AI models not to be directly patient facing. So that's why we built all of our structure to be doctor facing so that patient directly doesn't take crucial healthcare advices from the AI directly. So that's why there is a doctor campaigns where patient only faces doctor but doctor internally uses only AI or 90% AI to prepare their prescriptions based on the data.

So patient just come connect with doctor upload their data and doctor process just in one click generate the prescription validates it and give it to the user. So end user have no risk and doctor's productivity increases forex. So that's how we are trying to build in this and it's more like a medical copilot for doctor. I think you must have heard for GitHub copilots for coding and all those. >> You should totally use that in your marketing art. I think it that would really work. So before we wrap this up, I'd like to know some advice that you would give to other AI directors. It doesn't need to be in fintech. I would say like uh these days AI has become a lot of buzzword like people are just capitalizing and just seeing the surface of it. But what I as a data scientist from 10 years see that AI is much deeper and it's the core lies in the data science and prediction. So I would say like already at a very good position leading CTO director still you have to go to the basics where you see how you can actually predict. So how I see AI is to predict anything with a very very very high accuracy.

That's AI. If you talk about LLM, they are just predicting the next letter with very very high frequency. If you see computer vision or uh image models, they are just predicting the next image with very very high frequency. So I would recommend the directors to go deeper from the surface actually understand what is happening and if you just take one big open-source model fine-tune it with your own artificial neural network knowledge you will have wonderful things and it might even beat something like chart GPT or enthropic in very less cost and very less uh overall design. >> Awesome Gandra thanks for coming on. I will add links in the description so people can check you out. Awesome. >> And thanks thanks everyone for watching and we'll see you in the next one. >> Thanks Nick.