Science Corp's Max Hodak on restoring vision with a retinal chip — and the biohybrid brain interface beyond it

recently in the New England Journal of Medicine. These patients go from being unable to recognize faces, they can they can walk around because they've got a little bit of residual peripheral vision. The trial was in an age- related macular degeneration, but they definitely can't read.

they are like really profoundly blind and disabled and the the best patients in this trial could read could go from reading none of an eye chart to reading the entire eye chart. I mean there's videos of these patients like filling and crossword puzzles. It's really very cool.

Um and then separately from that we also have like the another key program in the company is a different approach to building brain computer interfaces where instead of placing wires like not like not in the retina but like in in cortex instead of placing uh wires or cables physically into the brain or genetically modifying the brain um so that like there's some things you can do like with optogenetics or sonogenetics.

Um, instead what we do is we have we grow up these heavily engineered biological cells that we hide from the immune system and then we just sit this on the surface of the brain and and so we don't place anything into the brain itself but what these cells do is they grow in and they wire up and they form new biological connections and they can form billions of them.

And so I think like the way to understand this or like I think actually fairly direct reference for how to think John's face right now this is like so we call it we call it a biohybrid neural interface. Have you seen James Cameron's Avatar movies? Yes. Yes. Yes. Do you know the ponytails that the aliens? Yes, of course.

I know exactly plug into like their tree memory store or like their horses. So, I think the question is like if you wanted to build a ultra high bandwidth neural interface like how would nature do this? I think what nature would do is it would grow a a new cranial nerve that has like a USB port at the end. Yeah.

And that uh I mean that that is super cool research, but that is definitely a research project. And so the way the company's architected is that is kind of paid for by the retinal prostthesis which is a like very practical near-term medical device.

We hope to have we've submitted for marketing approval in Europe for that. We're going through the review process now. Um the FDA actually is being much slower.

It it it's possible it won't reach American patients for a little bit longer and that's kind of crazy that Europe is going to get it first but that's where we are. Um no that's [clears throat] rare kind of narrative violation but exciting narrative violation for sure.

Um, but so yeah, hopefully that'll be on market next summer making money and that is big enough to pay for the rest. That makes sense. So why go uh with a uh a a retinal prosthesis instead of uh cutting a hole in the skull and um putting electrodes directly on the brain and trying to deliver the signal that way.

There's a lot of uh BCI firms obviously you co-founded one uh that that are are seem to be approaching the problem that way. Um is that just uh like like do you do you have a particular uh view on that?

Is that just farther out and you're and this is a way to get to market faster or is there something fundamentally like higher bandwidth about your approach? Like what are the different technical trade-offs? So BCI is not a product. BCI is a field. Okay.

And there are many different types of BCIs for many different types of applications. Like obviously you can't go into the retina to decode like a video game controller out of the brain. Um simultaneously you can't stimulate like frontal cortex to do like to do like sensory feedback.

And so it really depends on the type of thing that you're trying to do. And when and so in vision which we I mean we think that a visual prostthesis is a is a brain computer interface. We also think that coar implants are brain computer interfaces. Sure. Sure.

Um you don't need to be drilling in through the skull to get to cortex. Um so if you want to restore vision, you kind of have a choice of you've got the retina, the the output of the retina is the optic nerve that goes to a deep brain structure called the the phalamus and then the phalamus connects up to cortex.

And so within the retina, so let's just take a look at like the options that you have here. So in the retina, normally light shines in from the front, it hits the rods and cones. The rods and cones are the light sensitive cells. There's about 150 million of those.

These connect to about 100 million intermediate cells called bipolar cells and those compress down to 1. 5 million like optic nerve cells that connects to about 1. 5 million cells in the phalamus and that connects up to like 200 to 500 million cells in cortex. Mh.

And so no one I mean people have been trying to stimulate vision into the brain for many decades. And until the the clinical trial that we just finished, no one had ever gotten form vision, like structured vision that the brain could intuitively assemble into a hole.

Like you could get patterns that if you like looked at it carefully, it's like, oh, there's like a line here, there's a line here, it's like connected. That must be an A. Like here's a line. Like that's an N. But in the Prima trial, I mean, they could read off words at a time. And that had never really happened before.

And a big difference is that we're stimulating that first layer of cells, the bipolar cells in the retina. And we know that if you just go one layer deeper to the from the 100 million bipolar cells, the 1. 5 million optic nerve cells, if you stimulate those optic nerve cells, you don't get this.

You just get these diffused flashes of light that you can't really attend to and the brain does not intuitively assemble together because the brain has already compressed the signal. And you have to then figure out like what is that transform? How did the brain compress this?

Or how did even just the retina compress this? And the and so the phalamus has the same issues as trying to stimulate the optic nerve except it's under 8 centimeters of brain tissue under the skull.

And then once you're up in cortex, you're dealing with hundreds of millions of cells that are distributed over like large areas that you just can't stimulate selectively.

And so like people if you do this like you absolutely will get flashes of light, but converting that to like form vision that you can intuitively read is a totally different problem. Mhm.

And so and then even if that worked like even if that worked by perfectly one is like an outpatient surgery going through the soft tissue of the retina the other is like a four or five hour brain surgery drilling through the skull. I think like one of those is you're going to kind of win sales. Um yeah.

Wait, so uh you said outpatient. Uh walk me through uh comparing the level of intensity of the surgery to something like LASIK. So it's a little more than LASIK. Yeah. Um, but it's like not a ton more than LASIK. What about getting your wisdom teeth out? I was just coughing [laughter] stuff. I mean, you could do this.

So, in the trial, many of them ended up being done under general anesthesia. But to be honest, in these cases, general anesthesia is really like as much like a commitment mechanism for the surgeon and the patient than it is like there's any medical reason to do it.

I mean, you just like you can't change your mind halfway through. Okay. Commitment. Wow. you. But from a like experience perspective, so you could do this with you can make an injection next to the eye. The eye goes dark and numb for a couple hours and the uh and then you can go in through the soft tissue of the eye.

You leave the chip. There's a little injector. The surgeon presses a lever. It leaves the chip under the eye. They come out. They're done. Um and so it's a really very simple. Yeah. And then and then sorry just to be ultra clear like uh the chip is in the eye. Then how am I communicating with the chip? Is it wirelessly?

Is there is there a device that's on the other side of of my head or you said glasses maybe are interfacing with that? Yeah. So, if you look at this chip under a microscope, it it has all these little hex cells on it. And every one of these hex cells like the science prima chip. Um it's essentially a solar panel.

And the it works in conjunction with there's glasses that are worn by the patient that has a camera looking out at the world, although you could really get the video feed from anywhere. And then there's a laser that projects the image onto the back of the eye in the in infrared. Okay?

And because you can't see infrared, you can't like if you have residual peripheral vision or any if you're not totally blind, this doesn't interfere with that, you can still have that. But the infrared laser where it strikes the implant, it works like an overhead projector.

Like if wherever there's white that is projected, that is like that's energy. And wherever there isn't energy, it's dark. And wherever the laser is absorbed by the implant, it it stimulates the cells directly above that pixel. And so this is pretty cool.

Um, because the implant is powered by the information that's projected onto it, like like as a solar panel, this means that there's no implanted battery, there's no cable, like this tiny little fully wireless 2mm chip is the whole thing.

And also because the eye moves relative to the projector, like the projector is shining in from the front of the eye and then the eye moves and the image changes. Um, this means that the brain can easily fuse it together with their their existing vision. And so this there's like some pretty cool stuff here.

Like if you show one of these patients a solid green bar all the way across their visual field, they'll see a contiguous bar even though the implant only fills like a small area of the total area of of blindness.

And they'll say it turn it's like it's green and then it turns white because we can only get black and white right now and then it turns green again. But the brain fuses all of this together.

And so even though the implant only has 400 electrodes, as the eye moves around, you don't you don't experience the image that falls on the eye falls on the retina like a camera, the thing you experience is the brain activity of the world model in the brain.

And so as the eye is moving around, it's updating the world model. And that's the thing that you see. And so even though it's like 400 electrodes, you can't think of it like 400 pixels on a screen.

You think of it as just getting the information into the world model and the eyes moving around and the brain's cross referencing all that. So, it actually does significantly better than you'd think from being 400 electrodes. That's fascinating.

Um, how do you [snorts] how do you think about the analogizing around the artificial intelligence community of what you know of the brain? So, a lot of people in computer science or AI might say, uh, with with LLMs, we've built this piece of the brain. With the hard drive, we built the long-term memory.

Uh has has what has your research h have you mapped any of that onto the current state of artificial intelligence has it proven um uh insightful to help to for you to understand what's going on in AI?

This is funny because I mean at the very beginning of of Neurolink and OpenAI we were in the same building in San Francisco and we'd have these discussions about like oh who's going to learn more? Is AI going to learn from neuroscience or is neuroscience going to learn from AI? Yeah.

And I think it has been revealed that like like I asked I caught it with one of those guys a while ago and I was like oh like in retrospect what do you think AI learned from neuroscience. He thought about it for a second. He's like the concept of a neuron. Yeah that's [laughter] it. Like that's basically it.

It's like the very it's literally just in the name neuroscience [laughter] is neuron and that's it and then nothing else because yeah like the whole brain structure concept. Yeah. Anyway, continue and but going the other way I mean AI has been so useful. I mean there is a really interesting convergence going on.

This is kind of called like neuro AAI where neuroscience is learning a lot from artificial intelligence in ways that I don't think any of us really would have anticipated. Okay. Um and have you come across the platonic representation hypothesis?

Um so there's an empirical finding that different neural networks trained with different architectures with different objectives and different concrete data sets but for the same type of thing like images or or language or audio they produce these like these similar internal representations and what I mean by like rep like there sometimes you'll hear people say like oh these models are like stochastic parrots or they're just like ne like there's glorified autocompletes like these people are safe to ignore like the mathematical objects that you see appear inside these models are super interesting and look a lot like the representations that you see in the brain.

And so that is hinting at like there's some like deeper fact about the universe that we're figuring out here that basically if you have a lot of compute power and you kind of run it in these ways then you see these like these data these like mathematical objects kind of emerge.

And what evolution did and figured out in the brain is like looks a lot like stuff that you kind of see in in these transformers and these other AI models. And there's definitely some interesting unification happening there.

I mean, it's not it's still like it it it's a little more than speaking totally metaphorically, but it is still like I'd say like um instructive rather than literal. Um but that's getting like every month there's like some new cool thing that comes out around this.

Um, I will say that I don't know and I my view is that the transformer is like a reasonably good model of like what cortex is doing, but there's a lot more of the brain. Um, and so there's other parts that aren't fully cap that aren't that we're going to need something else, but it's not like the transformer is wrong.

I think it's like probably part of the story. Yeah, that makes a lot of sense.

Uh, I wonder what you think about uh just the idea that uh like it like it takes like millions of years to train a a model to like drive a car and it takes you know a 16-year-old like a couple weekends to do it or uh or the amount of energy that I will consume in one day of like reasoning generating my own reasoning tokens is like way less than what it takes to run a data center.

uh there seems to be some sort of like uh exchange ratio of uh that's like we're off by a couple orders of magnitude and maybe that's an algorithmic question. I don't really know. Do you have any thoughts on that?

Yeah, I mean evolution has done I mean it it has really it has minimiz it has been very good at minimizing energy and optimizing some other stuff. Um but I mean that I think is like the advantage of of the biological brains.

It's like every now and then I I see pitches from companies that they're saying, "Well, AI is really energy intensive. So, what we should do is we should grow up cultures of biological neurons and train them to like to do intelligence tasks.

" Cuz you can kind of do this like if you you can grow up neurons on electrode arrays and you can condition them to learn things by stimulating them in different ways. Like when I was in college, I grew up a counter for gamebot. Like there's actually not that hard to do.

Um, and uh, and I, and this is a thing that I think nerd snipes many people in this field at some point, but I I don't think that that's like the way to go. I think that there's like just really structural advantages in silicon like if you compare and contrast these two approaches. Yeah.

Like in the in the deep learning models like you can see all the weights, you can introspect them, you can like [clears throat] stop the model, you can change one, you can replay it like samples out, you can copy to disk, you can like send it over to the network.

Whereas with the the biological living like these organoid brains, you like you can't see the weights. You can't copy it to disk.

It's like the what it learns is the time integrated experience it's always had and like at some point four or five months in it will randomly get infected and die and you'll have to start over. Yeah. And so I think there's just like structural it's like the thing that the biology does is it's energy efficient.

But my response to this is like generate more power. Yeah. I like the idea. There's no such thing as a low energy wealthy society. Generate more power.

That's Yeah, I like the idea of like the solution to AI is just like have kids or something and you're like you you've reversed it all the way down to like if I want artificial intelligence I can do it biologically just have kids. Are you and the team getting much leverage out of models today?

Is it accelerating your progress or is it really just about having like deep domain expertise and being more obsessed with the problem than anyone else and hiring the smartest people in the world?

the I interestingly I think companies like this are really limited by infrastructure and so like I one of the things that I got from my prior boss was I totally received the gospel of vertical integration and the and we're really not it's not like we're held back by like genius scientific insights for the most part we're limited by like oh like well we need to get a new like a new material in our like like microfab deposition tool but this requires hooking up some gas plumbing which requires getting some specialist vendor to come out and like weld it to the machine or you're out of animal housing and like building that is like an architectural design process and then permitting and then construction like you're really limited by infrastructure more than your genius scientific insights and so even if you had this like I think when I think about like being in the takeoff era and what is kind of the impact of these of progress in AI like we're still limited by how that can impact the real world in really meaningful ways for these atoms things but I'd say there's two places that AI has had a bigger impact the first is ironically like comprehending regulatory standards and generating regulatory documents.

I was about to say you Yeah. If if permitting is the bottleneck, can you have like a permitting agent that just goes and like spams the permits until you get exactly what you need. Yeah. I mean, they're not quite like they you end up doing a lot of editing.

But the Well, you definitely just don't want to spam the permits. This makes the regulators mad. But the uh but I mean the the filing in Europe for to ask for approval for a retinal prostthesis, which we submitted last summer. Yeah. It was uh I forget exactly how many. It was like tens of thousands of pages.

There's a 65 GB PDF. Yeah. And that depends on like hundreds and hundreds of standards that we're just expected to know all the details of. And so being able to talk to these things is super useful like chat with this this these data sets is super useful rather than like having Yeah.

like having a big team of regulatory experts who are all kind of it's do this through meetings. Um and then the other place that we've had big success with with AI uh internally is in our protein engineering program.

M so there's um conventionally many of the problems that we're interested in you'd have to solve with like better electronics or better like micro like better physical devices somehow and we're now at a point where often when we find a problem we ask like can we make a protein to solve this.

Um so a couple months ago we published a paper on a new type of optogenetic protein which is these are these are proteins that can make a a a neuron light sensitive that is not normally light sensitive. So we could control it optically with light. And these typically are require very bright laser light in order to work.

And we were able to use AI models to find one that is so sensitive that it is responsive to like not just daylight but like indoor office lighting. And so that allows us to substantially reduce the power consumption so that we can because often the brain implants were often limited by thermals.

It's like how much energy you can consume depends on how much is limited by how much you can heat the brain. So if you have more sensitive options, [clears throat] then you can have your LEDs be dimmer and have more of them. Then you can think about going from like thousands to hundreds of thousands.

Um but then also that might turn actually into our next generation retinol product. Um we might that will have to go through clinical trials. That'll be a a process. But it's possible that in five or seven years you won't even need the chip at all.

You'll just get an injection and then we can just make the bipolar cells themselves light sensitive. Um and then that won't even need the glasses potentially. And that that really comes out of the the big breakthrough on on being able to solve these problems with proteins uh is an AI enabled thing.

Chat says uh we should have you ask how do how to explain this to preschoolers.

[laughter] Um I some labs various groups have talked about the revenue opportunity and just automating science and it's usually very general where they're just like we're going to come up with a bunch of ideas and then hypothetically they give the ideas to various people to execute on and they get some type of like royalty on it.

How much how much opportunity do you think there are for for a more general uh foundation model company to just come up with a bunch of ideas and then actually capture real value from it? It feels like uh in some ways the pharmaceutical industry maybe doesn't have a shortage of ideas.

They have a shortage of infrastructure and funding in order to test enough ideas to actually get viable uh you know solutions. Yeah, I think that a real bottleneck here is just the translation to human subjects research and then to the market it's really difficult.

Um, we have like the uh like a thing I sometimes joke about is like we're in a golden age of mouse oncology. Like if you're a mouse with a cancer like we've got some great things for you. [laughter] But the love for all the mice out there. That's great to hear for the mice. That's fantastic time.

We owe it to them for all that they've done for cancer. But like that's the trade-off. And so if you can um and the and I I get it. It's it's not you can't just say like, "Oh, well the FDA is the is the problem. " Like the problem like the problem is that human subjects research is like life or death.

It's like it's a it's like no joke. And like I've had the experience of like a patient goes into a s a new surgery for the first time or they [clears throat] you're going to inject them with something and you're going to wait and like that is a very stressful experience.

Like you want them like and that's like that's stressful experience for people like me. I'm not even the one getting it, right? Yeah.

And so the there's trade-offs there that I think are very deep in like our like the way that we in our civilization like value human life which I like I think is right and that um there and so you have this knob of like how much risk do you think of taking in human subjects research and how fast do you get new things and we know that the toolbox of science is very very powerful because when you look at what's possible in the animal models you have these amazing things that are possible but then getting that into humans is like and that and it's not just to say like oh well we um like like oh we should just deregulate all of this like there's it's more complicated than that um although I do think that there's a little bit of that um but at the same time I mean certainly this there's an intelligence effect like I think that there's it's going to be inevitable that these things I mean the fact that you can fold all the proteins at least in static forms is a big deal like I would not I would absolutely not bet against improvements in AI uh leading to improvements in in medicine and healthcare.

And I think there actually just to go one step further, I think then the thing that we're going to have to reckon with is healthcare.

So like if if you thought like like 20 years ago TVs and phones and computers were way way way more expensive um and now they're much cheaper, but we spend more on them total because this is like a technological growth industry.

Like normally if Nvidia sells 20% more GPUs next year and their revenue goes up and their earnings go up and the stock price goes up and like everybody's stoked about this, but as time goes on and there's more things to spend money on in healthcare that produce better outcomes for longer lives, um there's like spending should increase, but because we pay for this these like kind of insurance schemes which are kind of pseudo fixed buckets of money.

Like if if there were real breakthroughs in healthcare that allowed people to live much longer and have much better outcomes, but they cost money and you could spend like 10 times as much on healthare. Like that would be a catastrophe.

Like you do not want to spend like right now our system would not handle spending 10 times as much on healthcare. But that is kind of directly at odds with it being a technological growth industry. Yeah. What about the BCI industry? I think it's interesting.

We've seen like this boom in quantum computing and and there's public companies that do all this stuff, but it feels like well so specifically I feel like the interesting question with BCIS is like I want to understand your timeline.

It feels like there's these therapeutic use cases where somebody's blind and they they uh they're willing to take some level of risk in order to um see or at least see something.

Uh, and then there's like the utility timeline and entertainment timeline where I just have, you know, instead of wearing glasses, I just have a screen that's just embedded in my retina and I don't need to wear glasses and it's always on and I can, you know, turn it off.

And I I feel like that's where like it feels like that's where you're going is sort of a general purpose technology on a long enough time horizon, but maybe I'm off.

Yeah, I mean I think I mean it' certainly be cool if the uh like when your eyes are open you've got the world of bit world of atoms and when your eyes are closed you've got the world of bits. Um the uh [snorts] I mean the the near term is all these are medical devices.

Um especially on the cortical side, I think these are very serious brain surgeries that um like healthy 40-year-olds are not going to be like if your hands work, your a keyboard is a really great brain computer interface.

And there's a lot of research that goes into like the design of the Xbox controller or the design of like the VR inputs because if you can convey the signals from your brain to the muscles, like you can just capture that and that works really well. But at the same time, I think many people eventually become patients.

Like these bodies are great until they're not. and as they start to fail, we should have better options. And so I I think that BCI I think that there's a kind of a meme that it is an AI adjacent story, but I actually think it's like a longevity adjacent story.

Um I see like the like the only organ I really care about is the brain as far as I'm concerned. Kind of the rest of the body exists to like keep the brain alive and healthy and move it around and cause it to do things and I'm going to be fairly disappointed if I'm murdered by my pancreas or my heart. Oh yeah.

And so I think the you can get to this point where you say like well instead of you've got all these hard problems like yeah we've made progress on cancer and cardiovascular disease and metabolic disease but again like when you think about health insurance like you can't insure something with a 100% loss rate and we still that is what we were talking about in medicine and I think you can get to this point where the the brain is the thing that is really special makes you you and like when I look at a person I see an agent and a robot and biology makes great robots but the thing we really care about is the agent and if you can kind of deal with the agent directly and the rest become swappable parts, then instead of needing to cure cancer, we might just be able to avoid this entirely.

And that I kind of see as as one of the central elements of at least how I think about the promise of BCIS. And so I'd say um like one of our one of a thing that I think could really be possible is like by 2035, it it won't be widespread.

I think it'll take longer than that for sure, but by 2020 2035, can you offer patient number one the choice of like dying of pancreatic cancer or being inserted into the matrix? Yeah. What's the strongest steel you can offer for the anti-transhumanism arguments?

Because I I mean, I've been steeped in this exact Silicon Valley lore for basically my entire life. I've heard it articulated a bunch of different ways uh in media, video games, The Matrix. Uh and yet I' I've also seen a lot of push back. There's a lot of Yeah, there's a lot of push back against the transhumanism stuff.

Uh what what what's been the what's been the strongest uh argument that's uh that that that you've you've had trouble debunking, if any. Yeah. I mean, I I don't really like the label transhumanism. It just has like all these like connotations.

I also don't like the word the term like the I think it is important that these things like people don't want to be different. They want to be themselves.

I think it's important to talk to realize that things we're talking about should make you kind of just as much you as you've ever been and just like with the best quality of life you've ever had.

And I think like the a concern with AI is that we could it could be massively um like it lead to really undermining our agency. It could be like lead to a massive loss of agency for humanity in the emergence of something else. Whereas I think that these BCI technologies are really about increasing human agency.

It's really it's a very like um uh yeah it kind of exists to facilitate you and it should not and it should be it's not other than human it's like it it uh is a very prohuman technology. Yeah.

I mean as much as any kind of healthare right like we we try like you we do heart transplants we get we're working our visual hearts you have dialysis you have we treat cancer we like think that these things are worth trying to improve on and it's only transhumanists in the way that like any healthcare is that you don't accept that just like the state-of-the-art in the middle ages if you were like playing in the wrong forest and got a scrape on a branch you could suddenly die of a bacterial infection 2 days later like we developed antibiotics like this is only transhumanist in the way that antibiotics are it's this continuation of that story That's a great Yeah, that's a great argument.

I love it. Um, what uh Yeah, Jordy, you have anything else? I was gonna ask I was gonna ask about the merge. Yeah, I I'd be interested to get your depth personal definition of the merge. Sam Alman wrote in 2016 that it would could be a scenario where people become best friends with the chatbot.

Maybe this the merge is something like a AGI where we just keep moving the goalposts. Yeah. Yeah. or do do you think it's inevitable or do you think there's like some sort of fork or or where do you think the unexplored discussion surface era is around the concept of the merge?

I mean the way that we interact with AI as I mean there's many different ways that this could go.

I think like the the failure mode of like Terminator seems less likely but I definitely agree with Sam that like an underrated failure mode is people just start like delegating their decision-making to it because the models just make better decisions.

Like there's I was reading an article that one of the US military combatant commanders is now like running personnel decisions by chat GPT and if you're just like realizing like hey like these things make as good decisions as I make then you they kind of come like premerged like you don't need a device for that necessarily because we are just kind of like like causing it to happen in the world um like whatever it wanted and there's like some dark versions of that like you can imagine like an extinction scenario is that we just really become dep.

We're like, man, these things have great judgment. They really know what to do. We should like ask it for advice and listen to it and like persuades a bunch of humans into suicide. Like you should I think you got to keep an eye on those rates.

They won't be zero and I think it's unreasonable to expect them to be zero because you're talking about hundreds of millions of people, but like do those trends like what do those look like over time? Yes. Yes. Keep an eye on the rates. That's that's the great summation of what's going on.

It you the baseline's not zero, but if it starts climbing up, you got to watch it. Exactly. Just like self-driving cars, like humans are not perfect. Don't hold them to like like to these unachievable standards. Yes. But like keep an eye on the rates. Keep an eye on the rates. I completely agree. That's a great take.

Thank you so much for coming by the show. Uh congratulations on the progress. Jord, you have anything else? Yeah, this was incredible. I wish we had more time. Yeah, but uh come back on whenever you want. Yeah, I really appreciate this. This is Thanks for having me on. Good to see you guys. This was super fun.

Yeah, we'll talk to you soon. Cheers. Have a good one. Bye. Bye. Let me tell you about Profound. Get your brand mentioned in chat GPT. Reach millions of consumers who are using AI to discover new products and brands and get a demo at Profound. Um, our next guest is Andrew Dudham from Hims and Hers. Two companies.

No, just one company. Hims and Hers. Welcome to the show, Andrew. How are you doing? I'm great, guys. How you doing? Thanks so much for taking the time to jump on the show. Uh for those who who might not be familiar, can you uh just kick us off with a little bit of an introduction on the state of the business right now?

It's a public company. People know Tellah Health. Yeah. I thought you were going to say for anybody that hasn't heard of of HIMS because I'd like to find you guys have some great hair. So I feel like maybe we've got some customers on the call. Yes. So Hims and Hers, we've been around for about eight years.

It's actually our 8th birthday this month. We went public in 2021. Um, and the vision is to help people feel great through the power of better health. Yeah.

So, we connect you on your phone, you pick up your your iPhone, you click an app, and you immediately get connected to one of a thousand or 2,000 doctors who are registered in your state and then can immediately engage in everything from cardiovascular disease risk, at home testing for testosterone, menopause, weight loss, prescription treatments, um, all the way down to stigmatized things like sexual health and mental health and depression, anxiety.

So um uh you know we treat um you know millions of patients on the platform 10 to 15,000 patients per day.

Um so while we are in many ways a new and innovative tellahalth company as people what people call us we're actually probably one of the largest health care systems in the US today just given the pure volume of patients that we treat and it's all digital.

So no matter what zip code you're in, you have access to worldclass consistent care, standardized care, which I think is one of the most important parts of all this digital this digital healthcare revolution, which is no matter where you are, you get great care as if you're 10 minutes away from Stanford and go see a dermatologist, you know, right outside your door.

Um, so this year we're we're on track uh, you know, two and something billion in revenue, profitable, and uh, growing super fast, which is exciting. We're launching a whole bunch of categories. Hit it. Gong hit. gong for you. Um let's talk about labs products. So labs is an important one.

Uh we launched today two packets packages uh where you for a few hundred bucks can get over 120 biomarkers. This is the most advanced diagnostics frankly that are out there in the market.

And I spend a a ton of time figuring out with with Concier's doctors and others what is kind of at the leading edge where you can get twice a year annual testing and