Pod Insights Eye On A.I. #291 Naveen Jain: How AI Predicts Cancer, Diabetes & Chronic Illness About this episode This episode is sponsored by AGNTCY. Unlock agents at scale with an open Internet of Agents.
Visit https://agntcy.org/ and add your support.
Can AI and RNA testing make illness optional?
In this episode, Eye on AI host Craig S. Smith sits down with Naveen Jain, founder and CEO of Viome, to explore how RNA sequencing and artificial intelligence are transforming our understanding of chronic disease. Jain shares how a personal tragedy led him to launch Viome, a company on a mission to digitize the human body, predict illness before symptoms appear, and revolutionize healthcare.
Together they discuss how Viome uses metatranscriptomics to analyze microbiome and human gene expression, what makes RNA a better indicator of health than DNA, and how large molecular AI models are paving the way for early detection of cancer, diabetes, Alzheimer's, and more. Jain also reveals his bold entrepreneurial framework—"Why this, why now, why me?"—and his belief that asking better questions is the real key to innovation.
If you're interested in the intersection of AI, biotechnology, and human longevity, this is an episode you won't want to miss.
Stay Updated:
Craig Smith on X:https://x.com/craigss
Eye on A. I. on X: https://x.com/EyeOn_AI
Transcript Speaker names marked ? are inferred from speaking time and the show’s known hosts — they are not voice-verified.
0:00 Speaker 00 One of the interesting thing about chronic diseases is they don't happen overnight. You don't say, hey Craig, you know what? I was hanging out with a bunch of guys last night drinking. I think I might have caught diabetes. You don't catch diabetes. You don't catch these diseases. They develop over a long period of time. It takes eight, 10, 15 years for you to develop a disease. That means we should be able to understand when you're moving towards a disease. Imagine a world where illness is optional. And then you open it up. Inside that, there are three things. There's a prepaid envelope. You give us a spit of your saliva, touch of your stool, four drops of finger prick blood. Send it to us. We analyze 100 million biomarkers.
0:45 Craig S. Smith? Build the future of multi-agent software with Agnity. That's A G N T C Y. Now an open source Linux Foundation project. Agnity is building the Internet of Agents. A collaborative layer where AI agents can discover, connect and work across any framework. All the pieces engineers need to deploy multi-agent systems now belong to everyone who builds on Agnity. Including robust identity and access management that ensures every agent is authenticated and trusted before interacting.
1:26 Craig S. Smith? Agnity also provides open standardized tools for agent discovery, seamless protocols for agent-to-agent communication, and modular components for scalable workflows. Collaborate with developers from Cisco, Dell Technologies, Google Cloud, Oracle, Red Hat, and more than 75 other supporting companies to build next-generation AI infrastructure together. Agnity is dropping code, specs and services, no strings attached. Visit agntcy.org to contribute. That's A G N T C Y dot O R G.
2:11 Speaker 00 So first of all, I'm Naveen Jain. I'm founder and CEO of Viome.
2:17 Speaker 00 My background has nothing to do with science. I am not the science person. I'm not the computer science person. I'm not a doctor. However, I'm as, you know, I would say a startup junkie. This is my seventh, this is my seventh company. And having done, and I think I'm a whole lot older than probably you, Craig.
2:40 Craig S. Smith? No, no, no, no, impossible. I'm almost seventy. I turned seventy.
2:44 Speaker 00 So am I, so am I.
2:46 Craig S. Smith? That's a great age.
2:47 Speaker 00 So I am actually turning sixty-six in September.
2:50 Craig S. Smith? Oh, well, I'm ahead of you. I, I turned seventy in October.
2:54 Speaker 00 Okay, but my biological age actually is now thirty-three.
2:58 Craig S. Smith? Oh, really?
2:59 Speaker 00 Yes.
3:00 Craig S. Smith? That's impressive. I don't know what mine is.
3:02 Speaker 00 Um, I will, I will send you a test and you'll find out.
3:05 Craig S. Smith? Okay.
3:06 Speaker 00 Uh, interestingly is that, so basically I generally, when I start companies and I'll tell you my framework. So as we get going, I'll tell you the frameworks I used to start companies and why I started Viome. So you wanted my background? My background primarily is industrial engineering. Um, and my, and I have an MBA in uh, human relations.
3:32 Craig S. Smith? Okay, that's interesting.
3:34 Speaker 00 Yes.
3:36 Craig S. Smith? And so how did you start Viome? What was the, the inspiration? And how did you pull together the uh? And where are you based?
3:46 Speaker 00 So we are based in Seattle, Washington.
3:48 Craig S. Smith? Okay.
3:50 Speaker 00 So I started Viome eight years ago. And Craig, this is, you know, every adversity is always an opportunity as I would, you know, as one says. So here I was working on Moon Express. Uh, we became the first company to get a permission to leave Earth orbit. Don't ask me what, what permission you need and who grants that permission. We got the law changed by President Obama where we actually got, it's called Space Resource Act, which is anything we bring back to mine the moon, from mining the moon, we get to own it rather than become a property of the country. So here, our goal was to actually mine the moon for helium three for fusion reactors.
4:32 Speaker 00 And don't ask me, why would you want to go and mine the thing and when is the fusion reactors coming? That's some other day. And then we became one of the six, six companies to get two point six billion dollar NASA contract. So you would think that I am here on top of the moon. And then my dad was diagnosed with stage four pancreatic cancer.
4:51 Craig S. Smith? Oh, wow.
4:53 Speaker 00 And here I am thinking that I didn't even know he was sick, let alone a stage four cancer. He was given three months to live and unfortunately, that's all he got. It's got me thinking that, you know, why is it that we know so much more about everything, even in outer space? And we know so little about the human body? And what if, what if we actually can find out what is changing in the human biology? At the onset and during the progression of these chronic diseases? So it's not that he died from pancreatic cancer. As he was getting old, I had just taken it for granted, of course, he's gaining weight, he's getting old.
5:38 Speaker 00 Of course, he's got diabetes, that's what people get when they get old. Of course, he's got a high blood pressure, he's got to take his blood pressure medicine. Of course, he's got a heart disease, he's got to take his statin, right? I, I just take it for granted, that's how life is, that's how it should be. And then I started to think about it, that, you know, we as homo sapiens haven't really changed in the last fifty, hundred years.
5:59 Craig S. Smith? That's right.
5:59 Speaker 00 Yet, yet, more and more and younger and younger age people are getting sicker and sicker. What is going on in this body? And we said, look, this is the time where we should be able to crack this problem. And anytime I start a company, I ask myself three questions. Why this? Why now? Why me? And then I'll apply that to how I started, why I'm through there. But I want to just give you a framework so everyone who's listening to it. Want to start any project, any company, at least I can give them some wisdom of having done this now for thirty years. And seven companies, how you do that? So why this is really simple.
6:38 Speaker 00 You ask yourself, God forbid, I'm actually successful in solving the problem that I set out to solve. Would it help a billion people live a better life? And it's not because I'm philanthropic, far from it. I'm a very capitalistic person, but I do know. If I can build any product, any service that can help a billion people live better lives. I can create a hundred billion dollar company.
7:02 Craig S. Smith? That's right.
7:03 Speaker 00 But I never wake up in the morning and say, what should I do to create a hundred billion dollar company? So I know very clearly, making money is simply a byproduct of doing things that improve people's lives. And if you stay focused every day and say, what can I do to improve people's lives? You will have a great company at the end of the day. Right? And this is really everyone who's listening to it, I will tell you a basic trick. I now for the last thirty years, I have no alarm clock. I wake up at four A M. every single day. And when I wake up, I jump out of the bed. So here is a challenge for you. Then you wake up in the morning and if you're not jumping out of the bed. Whatever you're doing, you should quit that day because that's not your calling.
7:46 Speaker 00 When you find something that you really, really love doing. There's no way you'll be lying in bed and not getting up and jumping out of the bed, wanting to do the things you want to do. Right? And Craig, I guarantee you, you're one of those people that jump out of the bed thinking, oh, my God, what an amazing day.
8:03 Craig S. Smith? Yeah, it's true.
8:06 Speaker 00 So that's the first thing you do. The second, the second thing is, why now? And here you look at the stuff and see what had changed in the last two to three years. But more importantly, what do you expect to change in the next three to five years? That will allow you to solve this problem at scale in three to five years. And this problem could not have been solved five years ago. And the reason you do that, Craig, is very simple is you don't want to be using the technology to solve tomorrow's problem that is five years old. Because by that time you're ready to scale, somebody will come along and build a new stuff and you get disrupted even before you get a chance to scale. So you want to intercept the set of technologies that are coming up.
8:52 Speaker 00 So you can actually solve, scale the problem. And to do that, you take the massive problem and you break it down into what are the pieces that need to be solved? Yeah. And I'll come back and explain that to you in a sense, how to do that. So, you know, and so for example, if you want to live on Venus, you don't say, well, I don't know, that's not what I happen. You simply says, okay, what problems have to be solved to live on Venus? Well, you have to lead the Earth orbit, got it? You have to go from Earth orbit to Venus orbit, got it? You have to land on Venus, got it? Now you have to create a habitat on Venus, got it? Four problems to be solved, let's go one at a time and figure out what has already been solved, what we know about the second stage and what we need to change there.
9:36 Speaker 00 But that allows you to start to think about these problems in a modular way. And then you know exactly which problems are solved, which problems are solvable and what are the unknowns? And you say, okay, let me tackle that and say, can we solve that? Because then you know the problem gets solved, right? And that's really the way of looking at the things. The last part of this thing, this is the probably the most important part, which is why me? And why me comes down to is what questions are you asking that are different from what everyone else in the industry is asking? Because the questions you ask are the problems you solve. So for example, if you say, I want to solve world hunger.
10:18 Speaker 00 And every expert will tell you, if you're going to have a fifty billion people on planet earth, you have to grow the food more economically, you could increase the yield of the crop. There's so much food that's wasted during the transportation, you've got to reduce the wastage. You have to grow the food very close to where people live. But no one will ever ask the questions, why do we eat food? Because when you ask the question, why do we eat food, you said, oh, you need energy, you need nutrition. What are the different ways can you get energy? And she said, well, plants get energy from photosynthesis. There are bacteria that are growing up in radioactive nuclear waste.
10:59 Speaker 00 They have figured out how to protect their DNA against radiation and use radiation as a source of energy. What if you can take a genetic material from these bacteria? In vivo do a CRISPR, and then suddenly you're radiation resistant and say, honey, do you want to go get some radiation instead of getting some pizza, right? But point I'm making is, it's not that suddenly by asking a slightly different question, you opened up the problem to the solution that you would have never, ever thought about. And that's the beauty of asking a different question. So now I'm going to answer your question. Why did I start Wyoming and how did we go apply these principles? So first thing we said was, what if we can actually understand what is happening in the human body?
11:45 Speaker 00 Before you develop a chronic diseases. And the reason we chose chronic diseases is because we, I believe that despite the pandemic that we saw, infectious diseases is something since the invention of antibiotics and antiviral things. We've got mostly control. And yes, we're going to have every hundred years or so, we're going to have some pandemic. But this is not a problem that is facing humanity. Ninety-seven percent of all of our healthcare dollars are spent on diabetes, obesity, the heart disease, the cancer, the Alzheimer, the Parkinson's, all of these chronic diseases. And the beauty of these chronic diseases, if I may call that word beauty, is, or at least one of the interesting thing about chronic diseases is, they don't happen overnight.
12:29 Speaker 00 You don't say, hey Craig, you know what? I was hanging out with a bunch of guys last night drinking, I think I might have caught diabetes, right? You don't catch diabetes, you don't catch these diseases. They develop over a long period of time. It takes eight, ten, fifteen years for you to develop a disease. That means we should be able to understand when you're moving towards a disease. So my first question was, if we can solve this problem, would it help a billion people live a better life? The answer is all eight billion of us. If we can solve this problem, check that billion people, you got it. Next question, why now? And we say, look, to solve this problem, the three things must happen. You have to be able to digitize the human body.
13:15 Speaker 00 You have to be able to process massive amount of data that comes out of digitization. And then you have to be able to use AI to make sense of what is going on. Well, cost of sequencing is coming down. And even then, we started about eight years ago. We saw the costs were coming down for a whole hu- human biome- human genome sequencing to be thousand dollars. And we said, look, you know, in three to five years, it's going to plummet down to a hundred dollars because the costs are plummeting. And today, we sit here and we can do a complete sequencing, unbiased sequencing, give or take five, six dollars.
13:52 Speaker 00 Now, cost of processing, when we started the company, it was forty seven dollars. And we said, look, God, that's high, but you know what? Cost of processors are getting faster and faster. Cost of history is getting cheaper and cheaper. In three to five years, that cost will be down to ten dollars. Today, we pay dollar twenty five for it, right? Everyone knew the AI was going to be there and AI is here, right? So we say the time to start the company is now, and we were right. The last part was the most important part, Fred. I, as not being a scientist or a doctor, the biggest advantage you have is you get to ask really dumb questions. So everyone in the industry were absolutely convinced the way to solve the human diseases is to know,
14:38 Speaker 00 personalize for you, and what's the most personalized thing? Your DNA, right? Every crime show tells you that every person has a unique DNA. I need to know your DNA. And my first reaction was, interesting. Does your DNA change when you develop a chronic disease? So you do my DNA test and I gained 200 pounds. Has my DNA changed? No, now I become diabetic. DNA didn't change. Now I have a heart disease. My DNA hasn't changed. And by the way, and then I die. Hundred years after I die, you look at my DNA. It's still the same DNA. We still look at the DNA of mummies, right? If DNA can't even tell you you're dead or alive, how will it ever tell you you're healthier or sicker? So that my first thing was, wait a sec.
15:22 Speaker 00 This is not how we solve. And I went back to Khan Academy. So what is it that happens to the DNA? Well, DNA gets transcribed to RNA and RNA is what actually determines what is going to happen in your body. And if you're not very smart, what do you say? What the fuck? Let's just measure RNA.
15:45 Speaker 00 But nobody ever told me. In this principle, by the way, you never focus on how you're gonna do something. See, in my framework, you focus on what needs to be done and you never worry about how. And most entrepreneurs fail because they actually are expert in their field. And if they believe they don't know how to solve it, they say, eh, this can't be done. When you are in, you are some naive like me, we say, sure, RNA? Sure, we'll figure out how to do the RNA testing. That was number one. The second part was,
16:20 Speaker 00 I had to go, you know, I'm an information junkie. So when I want to start a company, I would probably read not one book, not two books, not five books. I read dozens of books on the subject. And then I go start reading science papers. So I am now reading thousands of science paper. And I'm reading that every chronic disease from Parkinson's to Alzheimer to cancer to diabetes to metabolic diseases, all connected to your microbiome. And I have no idea what this microbiome thing is. People are talking about the gut microbiome causing that. By the way, today the new research comes out that how oral microbiome is linked to Alzheimer and dementia. It's like, okay.
17:03 Speaker 00 So here I am reading all these things and I'm thinking wow, it's really interesting that how gut microbiome is interesting. Well, it turns out ninety nine percent or ninety nine point seven percent of the genes in our body don't come from the mom and dad. They come from these microbes that live in our gut, on our mouth, all over us, one hundred trillion microbes. Now, I'm thinking if everyone now believes this is the big problem and there are tens of companies doing microbiome testing, then why is this problem not solved? So I went back to my first principle, what questions are these companies asking? And it turns out every company, until YO started, even now, to date, every company, when you do a microbiome test,
17:48 Speaker 00 they do the same DNA test of microbiome. They will tell you what organisms are in a crack's gut, what organisms are in the bean's gut. And not knowing what organisms mean, I'm thinking organisms mean tiny, tiny human beings. Trust me, don't ask me why, I thought they're tiny human beings inside my gut. And what if they are like human beings? You take the same person, put them in a good environment, good behavior, put them in the bad environment, bad behavior. So what we need to focus on is not who they are, we should focus on what they are doing. Are they doing good things? Are they doing bad things? What are they expressing? And how it's going to interact with the immune system? And that is the core of the problem we need to solve.
18:32 Speaker 00 So great, simple, we're gonna go do the RNA testing of the human body. We're gonna do all the gene, gene expression RNA of all the microbes in our body. And then we're gonna put the AI together to do that. It's starting to feel I got the problem right. So then, so that's basically, we decided that's what the YO is going to be. All I had to do was figure out how to do this now. So I went around the country, I thought this is the easiest problem to solve. I go to Stanford, MIT, bunch of these labs, they've got, I figured it out, I license the tech and I'm good to go. Turns out, no one has done it. So I am now going to Los Alamos, sorry, Lawrence Berkeley, Lawrence Livermore. I was at NASA Kennedy Spirit Center, NASA Houston.
19:16 Speaker 00 I'm looking through all the labs, Stanford, MIT, Duke, nothing. I was at Los Alamos National Lab now. They were working on bio defense project. You know what Los Alamos National Lab is famous for, right, Craig? Nuclear bomb.
19:32 Craig S. Smith? Yeah, certainly.
19:34 Speaker 00 What could they be working on next? Biological stuff. So they're working on a biological defense project. And remember the project now, what is the problem they have to solve? If there is a biological terror in our great country, how would we ever protect our citizens? We couldn't care less what was in the biological bomb. What we care about is what is it it's producing? How it interacts with the human body? So we can create anti-growth for it, right? So they solve the same problem for a different use. So we licensed that technology from Los Alamos National Lab. Exclusive license to start Vio. Now, fast forward eight years.
20:16 Speaker 00 Eight years later, we have now analyzed, Craig, one million samples. One million samples, collected over one hundred quadrillion biological data point. When you do a test on Vio. And I don't want this to be an infomercial for Vio. I just want you to know, because this is something so interesting, someone who knew nothing about it. You get a kit that looks just like this. Vio full body intelligence. Our mission is imagine a world where illness is optional. And then you open it up. Inside that, there are three things. There's a prepaid envelope.
20:58 Speaker 00 You give us a spit of your saliva, touch of your stool, four drops of finger prick blood. Send it to us. We analyze one hundred million biomarkers.
21:13 Speaker 00 And then we tell you everything what's happening in your body. Your biological age, your cognitive health, your heart health, your immune health, your gut health, your oral health. And then Craig says, nah, that's for, that's for some, you know, people who are new. Tell me some nerdy stuff. Great. Here's your LPS production. Here's your urinocxin production. Here's your lapstick production. We give you everything you want to know. But then we say, look Craig, don't eat avocado right now because your uric acid production is too high and you already, uric acid is very high in avocado, you can end up getting a gout. Don't eat avocado. And by the way, despite Popeye telling you spinach is good for everyone, you should
21:57 Speaker 00 not eat spinach because your oxalates are not being degraded, you're going to get a kidney stone. Or don't eat broccoli right now because your sulfide production is very high, you're going to end up getting a lot of inflammation. But if you want to eat red meat, eat as much as you want because choline and carnitine in your red meat, your microbes are not converting them into TMA, which is trimethylamine, which gets absorbed in your blood and your liver uses enzyme called F M O three that converts TMA into TMAO, trimethylamine oxide that causes the heart disease. Fortunately for you, your oral micro speech and jawalas is not highly active. And by the way, don't take vitamin B three because that increases the production of
22:41 Speaker 00 F M O three, but you can eat red meat safely. So we can tell you every food you should eat it and here's why, don't eat it, here's why. And then you say, Craig says, well, you know what? But I, I, I'm telling you, I have constipation. What are you going to do about that? We say, oh, interestingly, there could be five reasons you have constipation. Let me tell you exactly why you have a constipation. It could be high production of methane gas that slows down the motility. It could be the high production of, low production of sulfide that's actually causing you to actually have this inflammation that causing constipation. It could be low G A V, right? It could be.
23:24 Speaker 00 Here are the five reasons you could have. In your case, it is the methane gas that is causing the problem. In that case, please don't eat artichoke. Because that has high inulin, is going to actually increase the methane gas production. And by the way, make sure you're taking alpha leucatic acid, you're taking these nutrients. And then we walk you through everything your body needs for everything we found. Take eighteen milligram of lycopin every day. Take seventy nine milligram of amylase because your proteins are not being fermented properly. Take twenty two milligram of berberin sulfate. Take eighty milligram of elderberry. And we walk you through every vitamin, mineral, herbs, digestive enzyme, amino acid, food extract.
24:09 Speaker 00 And guess what? We custom make it just for you every month. There is no pre-made stuff. We custom make it just for you. So this stuff is mine.
24:22 Speaker 00 And it says expires on, doesn't say manufactured on. I'm sorry, it says manufactured on, not expires on, my bad, my bad. It was made for me two weeks ago. Right, every month I get the new one made. Then we made the personalized biotics for your gut. We make the personalized oral lozenges for your oral microbiome. And we make the personalized toothpaste for your morning and for your evening, just for your oral microbiome. Now you say, all that is good. So how do I know it's going to work for me? So we did blinded placebo-controlled studies. Where we showed people in ninety days, not some in one year.
25:07 Speaker 00 People who had IBS, fifteen percent of population suffer from IBS. Sixty four percent of people who had constipation became healthy compared to ten percent on placebo. People who are pre-diabetic, their A1C came down by point four two. fasting insulin down by twenty five percent. Depression PHQ nine is scored down by fifty percent. Anxiety GAD score by down forty seven percent. Right? Just using food as a medicine. Now, from all the data that we collect, Craig, we are now able to diagnose stage one oral cancer, stage one throat cancer. And we launched that product today. Ninety five percent specificity, ninety percent sensitivity.
25:53 Speaker 00 And if I were to ask you, you know how you detect oral cancer today? Correct?
25:58 Craig S. Smith? The.
26:00 Speaker 00 Your dentist rubs the finger on your gum and says, I don't feel your cancer. That's your cancer detection today. Honestly, God, in the twenty first century, when we can land on Mars, we are detecting the cancer by feeling the cancer. Right? So now we have a saliva test that can molecularly tell you if you have a biomarker for stage one cancer. We now have same test IBD, stage one IBD that's Crohn or colitis. We are validating it for Colin Politz right now, which is seven years before you develop a colon cancer. And to bring you back home, in next four months, we're going to launch a stage one pancreatic cancer. For the first time in the human history,
26:46 Speaker 00 we'll be able to do a pancreatic cancer stage one test. Ninety five percent specificity, eighty four percent sensitivity.
26:53 Craig S. Smith? Yeah, okay, let's, let's look at some of these on, on, as, as I've mentioned, I, I had molecular you on the podcast, they track blood biomarkers. And, and on the subject of pancreatic cancer and it's, and we all at this point probably know somebody who's died of pancreatic cancer. And the problem is, is as you said with your father is he can't, it's asymptomatic until stage. Um. They had a case of a woman that they had been tracking for some time. And while they have no uh specis specificity.
27:35 Speaker 00 City, yeah.
27:37 Craig S. Smith? Uh. They could see that something was going awry with her biomarkers. She went in for a full body uh CT scan. They found lesion on her pancreas. Turned out she's a stage one and she was operated on and and uh and is healthy today. Uh. What specifically are you looking at in the case of pancreatic cancer? Is it the the the the bio? I mean, the uh the gut biome? Uh. That's one question that I wanted to talk to you about the role of senescent cells that accumulate as you age. Yeah.
28:13 Speaker 00 Very nice. So first of all, remember, we are unlike most companies, we just don't look uh look on one sample. So most company would do just blood or they would do justice tool. Uh. So we do your saliva. That means we're looking at all the oral microbial activity and all the human gene expression in your saliva. Same thing in stool. We're looking at all the gut microbial activity and all the epithelial cells that are shedding. In the blood, we're looking at you all the stuff, same thing, mitochondria, all your inflammatory marker, all of the human gene expression, right? So in this, in the case of pancreatic cancer, we found the three forty ages that actually are the indicative of the pancreatic cancer and
28:56 Speaker 00 the polyamine production in your oral microbiome. So if you were to infect this Google, you can see the paper or the polyamine production in your mouth is indicative of pancreatic cancer. So we look at all of these biomarkers together to actually detect. In fact, in the case of oral cancer and throat cancer for stage one, we found hundred and fifty different things. And some of them were human gene expression and some of them were the microbial gene expression that were indicative of stage one cancer.
29:28 Craig S. Smith? And so that the oral markers, that's your cells, is that right?
29:38 Speaker 00 To both your cells and the microbial cells.
29:41 Craig S. Smith? And I see. Yeah. And what? Can you talk about what it is that you're seeing in in those two panels, I guess, of saliva and microbiome?
30:01 Speaker 00 Yeah. So basically, again, I can, you know, there's a paper I can send you that all the list of the things we do, there are about 150 of them. Many of them are basically the inflammatory markers in our immune system that actually go up when we have a early stage of cancer. So we look, so there are a lot of inflammatory markers, some of them are hallmarks of cancer that we have known about the thing. It's the expression of the gene that starts to change. So unlike your DNA, and the reason many of the cancer companies like, you know, let's look at Grail. So Grail is an unbelievably great stuff what they did for a pan cancer test. But if you look at the Grail, what they're looking for is cell-free DNA, right?
30:45 Speaker 00 Problem with the cell-free DNA is your tumor has to be big enough to shed enough DNA for you to be able to detect it. So they're really, really good at a stage three and a stage four, right? But they suck at stage one. So average sensitivity for Grail test for stage one is 27.1%. That's the average number, right? Ours is 90%. And the reason is the RNA changes at the onset of a disease. Your gene expression starts to change at the onset of a disease. And as a matter of fact, it's changing throughout and you can see the predictive behavior. You can see you're moving towards a disease. You're going to be diabetic, you're going to have a heart disease.
31:30 Speaker 00 So we are developing what we call multi chronic disease test, early detection test. So fifty different chronic diseases, including many types of cancer. With all of the saliva, blood and stool, we are able to actually predict and early diagnose fifty different chronic diseases. So Hashimoto disease and different types of cancers like colon polyps, which is seven years before you develop the colon cancer, right? So we are actually going through this whole panel called, it's called MDED, multi chronic disease early detection test.
32:03 Craig S. Smith? That's interesting. Yeah, and in, I've seen those studies about
32:12 Craig S. Smith? proteins in the saliva
32:16 Craig S. Smith? acting as
32:19 Craig S. Smith? biomarkers for pancreatic cancer in particular. Yeah, I think there are four or five proteins. The, the other
32:30 Craig S. Smith? people I've spoken to are looking not at
32:37 Craig S. Smith? the biome, but at specific
32:44 Craig S. Smith? blood markers.
32:47 Craig S. Smith? And
32:49 Craig S. Smith? and, and how that, one of the questions that, that they're, they're focused on is how that relates to senescent cells, because as you get older, these senescent cells, which play a positive role accumulate and then start
33:09 Speaker 00 Become negative.
33:11 Craig S. Smith? Yeah, expressing basically toxic compounds. Do you, do you work at all on, on senescent cells?
33:20 Speaker 00 So we see all of the, and so we can see the cellular senescence. Remember, you have to look, unlike most people, we don't look for anything specific. That's, so now we look for everything, right? So we are analyzing every single transcript, even if we have no idea what this transcript does, just to be very clear. Because even despite my telling you how much we know about microbiome, the fact is, just like the universe out there, almost seven, ninety percent of is dark matter.
33:54 Speaker 00 We know ten percent of at most of what is going on and ninety percent is dark matter. And in fact, so much so, when we were looking at our data, we only took about three, four hundred people data RNA. And we put that in the UCLA, one of the professor, we were doing ideas, study with them and we published the data. Stanford professor looked at our data and they found a new type of life. So if you saw the paper called Oblisk, so they found a new type of life, completely new a life that never seen. It is a a virus like thing, except it has no cell. So it has a no protective cell, but it only is RNA that codes for protein.
34:40 Speaker 00 So it is an RNA coding for protein with no cell that means it can't survive outside the body. And they found that looking at our data. So my point is, when we are looking at RNA, we are finding these RNA viruses. We are finding these phages which are RNA phages that impact bacteria. So it is all of these things and some of the times we can, they are annotated, so we know who they are. And most of the time, we have no idea who they are and they get, you know, some annotated years later and we go back and put them in the data. Or a lot of the times, AI can say, look, these are the set of transcripts that are implicated in the people who have these diseases.
35:24 Speaker 00 And then we go out and do say, okay, what could this transcript possibly be doing? And then we go to the basic science and say, oh, my God, look at these transcripts, they're all encoding for IL thirteen.
35:38 Craig S. Smith? Yeah, yeah.
35:41 Craig S. Smith? And and how long have you you you were talking about these studies with
35:47 Speaker 00 Yep, yeah.
35:48 Craig S. Smith? placebo controls.
35:49 Speaker 00 Yeah.
35:49 Craig S. Smith? How large are those studies and how long have you been conducting them?
35:54 Speaker 00 So they are generally, so all these studies all have the power analysis. So based on a primary endpoint, sometime you need hundred and fifty people, sometime you need two hundred people, sometime you only need hundred people. So we do the power analysis based on the primary endpoint. So we don't go out randomly, say, let's get take twenty people and do the study. We say, all right, if our A one C is our endpoint, how many, if you want to do the variation of no more than point one, then how many people is studies do we need to be? And we say, all right, we need to be hundred and seventy three people, correct? Let's go to hundred and seventy three people, right? So we do completely based on statistical significance and the clinical significance using the power analysis.
36:37 Speaker 00 So these studies generally are between, I would say, one fifty to two fifty in range.
36:43 Craig S. Smith? Hmm. And and how long do you track these people?
36:47 Speaker 00 So we generally like to do, because we think the significant changes happen in about ninety days to one eighty days. And that's so most of our studies are somewhere between ninety days to hundred eighty days.
36:58 Craig S. Smith? Yeah. And then do you follow these people over time beyond that?
37:06 Speaker 00 Not in a blinded placebo-controlled study, but in our data, yes. Remember, we have been around for eight years. People come back and redo the test every six month, every year. So we have lot of longitudinal data about people, about how their body is changing, when they're developing their new symptoms, what's going on. So we have lot of longitudinal data. And having a million samples really allows you to do much, a lot larger scale with hundred quadrillion. In fact, I would argue that if you have lot of population level data, even if you don't have longitudinal data, and if you have lots of people at different stages, you could actually proxy it for longitudinal data.
37:46 Craig S. Smith? Yeah. And you know, as I said, I had molecular U on.
37:50 Speaker 00 Yeah, yeah.
37:52 Craig S. Smith? I had Eric Topol on.
37:55 Speaker 00 Yep.
37:55 Craig S. Smith? Topol, well, first of all, it does, Viome, is it complementary to what molecular U is doing? Is it, is, I mean, how much overlap is there? And what do you think of molecular U's process?
38:15 Speaker 00 But first of all, all I can tell you is that, you know, I hate commenting on other people's, what they do or don't do. What I can tell you is that we are the only company in the world today that does a complete unbiased RNA sequencing of all three samples, period, right? That means if you have a hammer, everything looks like a nail to you, right? So if you do a blood test and you do blood chemistry standard panel, there are companies out there by they, they will send you to Quest, they give you 24 tubes of blood, take the report that comes out of it, put the chat GPT and send it to you. And those companies are making shitload of money selling whatever the same commodity stuff that have been around for a long time.
38:59 Speaker 00 And you know the company I'm talking about, right?
39:03 Speaker 00 So these companies, they think just because they know your LDL and HDL and stuff, somehow they can now base everything on that. Well, guess what? You know,
39:16 Speaker 00 just because you know what is happening, doesn't, you don't know why it is happening. The beauty of the thing is not just to know you have high LDL, you need to understand why do you have high LDL and what is it that you can do about it, right? And that's the only way to do that is to looking at the transcripts to see, look, this is what's being expressed. This is what's being caused. And by the way, these pathways can be intervened by these different types of substrates.
39:46 Craig S. Smith? Right. And now the other thing, Eric Topol,
39:51 Craig S. Smith? uh, who's written about, uh,
39:54 Speaker 00 The pioneers and he, and he talks about a lot of our gut microbiome. He's a big believer in the gut microbiome.
39:58 Craig S. Smith? That's right. But he's very critical of companies that then sell supplements.
40:05 Speaker 00 And then he talks about us. He is absolutely. So what he is, uh, he talks about things that things that he thinks people still need. He talks about you should take creatine. He does talk about you should take all this stuff. And the reason is, you know, to large extent, the new, the supplements are basically augmentation of your new food. It is not a replacement for food. And this is what Eric talks about. The nutrition is the number one key for human health. Let's talk about what food is. Food is nothing but a stack of substrates inside that food. Strawberry is not just a strawberry. It has a folicin. It has a whole bunch of things.
40:49 Speaker 00 Now, you could argue that, let's assume my histamine level is very high in my body already.
40:57 Speaker 00 I can't take a strawberry because strawberry does also include the histamine increasing compound. So if I say, don't eat strawberry, but I do need antioxidant effect of folicin. There is nothing wrong taking that folicin as a supplement. And that's what we do. We go look at the stuff and say, you do need antioxidant stuff. And you cannot take a strawberry. So we're going to give you folicin. We're going to give you a blueberry extract. We're going to give you the amla extract. So these are all natural food extracts. They are the natural compounds that come from food and plants. These are all, by the way, grass. These are not some synthetic compounds.
41:39 Speaker 00 These are all coming from something that we eat every day.
41:45 Craig S. Smith? Yeah. Uh, and, uh, and, and then, uh, let's say, uh, someone does the test and they follow the, the supplement regime and whatever else is recommended, presumably some.
42:05 Craig S. Smith? Weight loss or or something like that.
42:07 Speaker 00 So we have shown everything. We have shown, in fact, even for employers, we did, we actually did the thing where employer gave it to their employees. Average weight loss was about eight point nine pounds, and this was in six months. Average quality of life improved substantially, absenteeism came down, productivity went up. And most people told them they have higher energy than they ever had before. So the energy level went up. Their digestive symptoms completely went down and anxiety and depression was one of the biggest thing that came out of it. People says my anxiety levels, my stress levels came down significantly.
42:48 Craig S. Smith? Yeah, but then you test again in six months to a year.
42:54 Speaker 00 Yeah.
42:55 Craig S. Smith? And and and then do you adjust the the supplements?
43:00 Speaker 00 Everything, your food gets adjusted. So, you know, even even between the test. So if we know that, you know, from your Oura ring or from your Apple, Apple watch. We noticed that you're not, for the last two weeks, Craig hasn't been sleeping well. We can say, all right, Craig, we're going to adjust your food and we can also now increase the magnesium. And ashwagandha in your supplement, so you can sleep better. We noticed that you've been working out a lot, so we added creat, we added creatine in your supplement. And by the way, we change your food recommendation for you to have more protein.
43:37 Craig S. Smith? Yeah, yeah. Yeah, I mean, that's fascinating. And and and when when did you productize this? I mean, when was it launched?
43:47 Speaker 00 So it was launched seven years ago. So I started the company eight years ago. It took us a year. We built all fully automated lab. By the way, we run our own CMS clear certified lab. So it's all under run our own lab. We run, you know, unlike many other people, we actually do the complete science. We have a whole AI team that actually builds the whole molecular models for it.
44:08 Craig S. Smith? And how has, I mean, uh, seven years is, is a long time in the world of science and artificial intelligence today. How is it, uh, changing with, uh, generative AI and, uh, you know, these models?
44:27 Speaker 00 You know, honestly, the, you know, if you think about it, you know, a lot of people may not understand that generative AI has been around for fifteen, twenty years. It was Google who actually wrote the first paper, attention is all you need, right? So this transformer architecture and they've been applied to language and they're called large language models.
44:46 Craig S. Smith? Yeah.
44:47 Speaker 00 But they were applied to the language. Now imagine if you understood the language of biology. Now, if you do the language of biology, you can apply not the existing L L Ms. But you can create a transformer architecture that uses the molecular large models. So you can create large biology model, large molecular model. And now you're using A C G T or in our case, A C G U as the way to actually now look at that language. And that actually we have done, it is fundamentally does as opposed to building the one machine learning algorithm for one disease. You can now build actually a large model.
45:32 Speaker 00 And by prompting with a very small set of positive sample, it can create whole bunch of synthetic data and come up with a classifier for a disease. That means you can create an early diagnostic biomarker for almost any disease using these generative models. And that couldn't have been possible seven years ago.
45:51 Craig S. Smith? That's right. Uh, yeah, and, uh, uh, I'm just curious what you think of Insilico's approach where he's, he's tackling, I think. I B F or, or one of, uh. Inflammatory diseases, yeah.
46:11 Speaker 00 So we, you know, again, I hate commenting, but let me tell you what is that they are really doing is they are going after not the biology, they are primarily for drug discovery, not the target discovery. So think of us as a finding the biomarkers, that's a target. Now you find a target for this is my pathway target. Now which molecule, what chemistry? So they have built the whole chemistry and database that allows them to find the right set of molecules that will actually work on that target. And I think, you know, the interesting thing is, it is a way, one way of
46:54 Speaker 00 at AI accelerating, finding the set of molecules that will work. And then you have to go through a whole bunch of toxicity and all this stuff because human biology unfortunately is extremely complex. Just because a target, a synthetic target works on a particular pathway. It becomes hard to know what else would it do? And unless you can create a complete digital twin of the human body.
47:21 Craig S. Smith? Yeah.
47:22 Speaker 00 And what you and I both know is that
47:26 Speaker 00 even the drugs we take today, how they are metabolized, changes based on it's an impact based on your microbiome, right? So for example, if you're taking levdopa, levdopa can be, which is for Parkinson's disease. Microbes can actually use tyrosin to completely metabolize levdopa and you get no drug in your body. Right?
47:50 Craig S. Smith? Yeah.
47:50 Speaker 00 In fact, even the cancer vaccines like immunotherapy, its efficacy depends on your gut microbiome. In fact, they did the research that shows the people who had melanoma and the people, seventy percent of the people where immunotherapy did not work, they did a fecal transplant of healthy person and suddenly the drug started working.
48:10 Craig S. Smith? Yeah, yeah.
48:11 Speaker 00 Right. So it's really interesting that how, you know, your microbiome and your immune system, how they all work together as ecosystem rather than, so we are really a super organism, rather than being a human beings that has a bunch of these microbiome.
48:28 Craig S. Smith? Is there anything I haven't asked that you think listeners should know?
48:32 Speaker 00 Well, Craig, my point, the only thing I would tell you is that every one of the listener is need to start developing this idea of intellectual curiosity.
48:43 Craig S. Smith? Yeah.
48:43 Speaker 00 The best advice I can give people is that you don't need to be an expert to disrupt an industry. As a matter of fact, your biggest asset is not knowing things. As long as you're intellectually curious about the things. And to be intellectually curious, you have to be able to say, what if? You have to be able to say, what if everything that I believe in, what if it is wrong? And that needs to be the way you learn, that needs to be the way you teach your children. So when your child, Craig, says, hey, Dad, look at that beautiful blue sky. Most parents will look up and say, wow, that's indeed extremely beautiful. Rather than using that as a critical moment to say,
49:28 Speaker 00 son, do you know the sky doesn't exist? The sky is a fragment of our imagination, because simply the blue light that's scattered that we call a barrier called sky. And by the way, talk about the color blue. You know, there is no color that exists in the out there. It is simply the photons and electromagnetic waves that hit our retina and our mind creates the color. And it's not about teaching them science. What you're teaching them is that something they are absolutely certain about. They absolutely believe in, what if that is wrong? And that allows them to learn to challenge the prevalent theories. And that's where the disruption happens. And it always comes from the people who know nothing about that field.
50:13 Craig S. Smith? Yeah. Yeah, that's right. That reminds me, I interviewed Albert Hofmann years ago, who was the synthesizer of LSD. He lived, he was a hundred years old. He lived on a, on a mountain in Switzerland. And yeah, he was looking out the window and saying how. Amazing it is that we see all this uh shape and color and it's it's not out there. It's coming from within. That is just our our.
50:47 Speaker 00 Exactly.
50:48 Craig S. Smith? Yeah, yeah.
50:50 Speaker 00 And I would say my second advice will be, you know, dream so big that people think you're crazy. That means when you tell someone what you want to do, they got to be able to say you're crazy. And that means you're on the right track, right? And and just know there are no failures in life. Everything is an experiment. In experiment, you have an outcome A and outcome B. When outcome A happens, you do C and D. When outcome B happens, you do E and F. So there are no failures, there are only outcomes. And if you just look at the life the same way, there is no no failure ever.
51:23 Craig S. Smith? Build the future of multi-agent software with Agency. That's A G N T C Y. Now an open source Linux Foundation project. Agency is building the Internet of Agents. A collaborative layer where AI agents can discover, connect and work across any framework. All the pieces engineers need to deploy multi-agent systems now belong to everyone who builds on Agency, including robust identity and access management that ensures every agent is authenticated and trusted before interacting.
52:04 Craig S. Smith? Agency also provides open standardized tools for agent discovery, seamless protocols for agent-to-agent communication, and modular components for scalable workflows. Collaborate with developers from Cisco, Dell Technologies, Google Cloud, Oracle, Red Hat, and more than seventy-five other supporting companies to build next-generation AI infrastructure together. Agency is dropping code, specs and services, no strings attached. Visit agntcy.org to contribute. That's A G N T C Y dot O R G.
Previous #290 Joel Hron: How Thomson Reuters is Approaching The Next Era of AI Next #292 Peeyush Ranjan: How Nurix Is Redefining Voice AI for the Enterprise Page updated November 1, 2026 · transcript and summary are machine-generated
