Are Brain-Computer Interfaces Actually Ready for Humans?

Hello everybody, and welcome back to TWIST.

This is Alex.

Today is Monday, March 16th, 2026, and we have 3 amazing interviews with world-class, world-changing founders coming your way.

First up, we're gonna talk to Paradromics.

They are building brain-computer interfaces.

Yes, BCIs really are coming.

Then we're talking to JetZero, which brings me to a question.

What if you had a plane, right?

And it was just one big giant wing?

Then we have Nutrisense.

Yes, I know health tracking is all the rage because it's fair enough.

Knowing what's going on in your body is a pretty important thing if you want to health max.

And maxing is a big deal these days.

I mean, just look at me.

I'm looks maxing right now.

Nutrisense, I think, is on the cutting edge of consumer health, and so I wanted to learn more.

It's a great show.

I learned a lot.

I hope you enjoy it.

We're back on Wednesday.

Let's have some fun.

If you are an AI fan, you must have heard about the big hype in talking to your AI, talking to your agents, using your voice.

Well, That's well and good if you have that capability, if that's in your personal capacity.

But what if your body doesn't quite do what you want it to do?

You're going to need a little bit of help.

And that's why I'm really excited that companies like Paradromics are working on taking BCIs out of the lab and putting them into our brains, not only so we can do cool stuff, but also

so that everyone can participate in the future digital and AI revolution.

So please join me in welcoming to the show, it's Matt Engel from Paradromics.

Matt, how you doing?

Good.

Thanks for having me.

Dude, my absolute pleasure.

I love brain-computer interfaces because although I can type very quickly, I really do think though that my hands slow me down.

And as AI gets faster, I just want to be able to do more with my mind.

So you are the exact person to talk to, but I'm getting ahead of myself.

Tell me about your brain-computer interface and how it works, if you don't mind.

So we've built a brain-computer interface that is implantable.

And so one of the reasons why you want to implant a device in your brain is that it has access to a higher level of signal resolution if it's close to the brain.

And particularly if we have, our device has little tiny microwires that are smaller than human hairs that get close to individual neurons.

And that's the highest level of resolution that you can have when you're recording from the brain.

The more neurons you can record from, the higher the data rate of the device.

And we'll talk a little bit about that later, I think.

But the way our device works is that it records from large populations of neurons and transforms that neuronal information into something actionable by a computer.

So in the example of a person who can't speak any longer, take Stephen Hawking, the late physicist.

He was paralyzed.

He couldn't move his muscles any longer, but his mind was completely intact.

With a Paradromics device, he could have spoken again through a computer by attempting to speak and then having the little device in his brain read out what he's trying to do and inform

a computer.

That's how I understood it.

But I'm curious about how the mechanics actually function, because to me, if you put some wires next to my neurons, they're going to record what, electric impulses, essentially a series

of of, uh, data points.

Individual neurons, you can— they— you can think of them like tiny electric eels.

They undergo these binary signaling events called action potentials where for about 1 millisecond they change their voltage a lot, and then that results in them sending signals to up

to 10,000 other neurons.

And so if you go into the brain with tiny little electrodes and record from a lot of neurons, you can see when each neuron is firing if your electrode is within about a tenth of a millimeter

of the neuron you're recording from.

And you're basically recording that electric field around the neuron when it undergoes its signaling event.

So by electric field, you're more keeping tabs of an area of electric activity versus just what that one neuron is doing, because you said they're connected to 10,000 apiece.

So we don't record from individual— we don't go inside of the neurons, but we sit in the sort of surrounding space.

And it's almost like dropping microphones into a crowded cocktail party.

Got it.

So we record the voices of neurons that are in close proximity to our electrode.

Okay, so that helps a lot.

Now I get now how you can get the data from the brain, but the thing that I've always been a little bit uncertain on is, cool, we listen to the brain, we drop the mics into the cocktail

party, we have this background hubbub.

How do we turn that into what the person wants?

What's the translation mechanism from signal into, okay, this person wants to say X or do Y?

We put the device into the area of their motor cortex that tells their body what to do.

So there's enough information in this area of motor cortex for us to understand what they were trying to do with their muscles.

And so the way the training works is that someone is looking at a computer and it says, "Try to say this sentence," and they'll attempt to say a sentence, and then they'll get another

sentence and another sentence.

And eventually we have enough paired recordings of neural activity

together with the sentence we know they were attempting.

And so we can build a decoder that can detect what they were trying to say based on the neural activity.

That's really interesting.

Is this a place— and I hate to bring up AI in a non-AI conversation, really— but is this a place where modern AI tools and technologies have increased the pace at which we can decipher

the microphones at the cocktail party to better understand more quickly what people are trying to do with their brains to improve our ability to read the signals?

Absolutely.

So the most popular forms of AI right now are large language models.

And a language model is exactly what you want to do if you have a noisy signal of what someone's trying to say.

It allows you to use the system even faster, incur some errors, and then go back and fix the errors or even predict prospectively what someone's going to say.

And so language models, even before the era of OpenAI,

a lot of what they were being used for was speech recognition when we talk to our phone and we get speech-to-text.

So those are language models that are cleaning up what you're saying as you're saying it.

and we can use those same things in a thought-to-text or thought-to-speech model.

I know Kinexus, your BCI, is working on helping people talk.

So we're talking about translating signals about speech.

Does the same concept apply to other things, moving my body or expressing a thought or a concept, if we put the sensors on a different part of my brain that was more used for that activity?

Right now, in clinical trials across the country, there are people controlling robotic arms by thinking about moving their arm.

And in general, this idea of using AI, you know, to predict sequences of events, you know, in the case of language, it's really easy to understand because we all have examples.

We have examples and experience with large language models.

But you can also have a motor language.

You know, there are predictable sequences of motor events.

Like, for instance, if I want to reach over here and pick up my cup at a certain part in that sequence, it becomes very predictable what's about to happen.

And so we can use those kinds of predictive models for motor decoding as well.

That's really fascinating.

Even as we move kind of up the chain into higher cognitive areas, we can start to develop new frontier models to understand essentially the statistics of thought in different areas

of the brain.

So I'm a big science fiction dweeb.

And so this next question is just for me, not for the startup community, but can Can we use this sort of technology, maybe we need multiple sensing areas, but to essentially be able

to watch people's dreams?

Yeah, I mean, to the extent that you experience your dream, it is fundamentally just neural activity.

And so if you had electrodes in the right area, you could, in the same way that we can reconstruct if we're recording in a sheep's brain, this is actually how we test our data rate.

We put devices in sheep primary auditory cortex and play it sounds.

and then we build decoders to understand what sounds the sheep is hearing.

So when we're recording from a sheep's auditory cortex, we can know what the sheep is hearing by decoding the brain activity.

And that could be the case for someone who's sleeping as well, because the same substrate for

dreams is the substrate for sensory experience.

Matt, did you just let me know by accident that you can speak sheep?

Yes.

That's incredible.

I've raised sheep.

Sheep are dumber than dirt.

I would love to know what they're thinking.

November of 2025, Paradromics gets the FDA to grant it what's called Investigational Device Exemption, or IDE, to essentially start using the Kinexus BCI in people.

And you were shooting for a clinical study this quarter with 2 people with impaired speech and limited extremity movement.

Where are we in progressing from taking this technology out of the lab and out there into the real world testing?

We're talking to potential patients right now.

We haven't implanted the first patient yet, but that is, I would say at this point, imminent.

So in the world of biotechnology, imminent can mean something very different than in the world of startups or software.

So imminent is in like, is that weeks, months, years?

I don't know how to read that.

Yeah, let's say it's in the next 4 to 8 weeks we're expecting to implant.

How long does it take to get data back from those early implantations of your BCI to begin to iterate and improve and learn from?

Is it right away or does it take a while to accrue?

It's usually about 3 to 4 weeks after the surgery that we begin the rehabilitation, uh, component.

Building out your team is one of the most crucial things you have to get right in your startup, and finding the right developers is particularly important.

But now there's Lemon.io.

They're gonna save you time, money, and headaches by doing all the time-consuming legwork for you.

They've got an experienced lineup of pre-vetted developers working for competitive rates.

Just 1% of applicants are accepted into Lemon's elite program, and they're not just out there finding this great talent, they're also working with you to integrate these new members

into your team.

Plus, if it's not a good fit, hey, and sometimes things don't work out, Lemon will hook you up with a new developer ASAP.

I've seen startups go from just pretty good to amazing after filling out their teams with developers from lemon.io.

Go to lemon.io/twist and find your perfect developer or technical team in 48 hours or less.

Plus, TWIST listeners get 15% off their first 4 weeks.

That's lemon.io/twist, lemon.io/twist.

Politely, how invasive is the implantation process?

Invasive is a weird word.

It's been bandied around, but it's not actually a medical term of the art.

I'll say this because there are a lot of devices that are touted as more or less invasive, but then that doesn't correlate with actual safety outcomes in surgeries.

So I'll describe what the surgery is, which is an incision in the scalp,

a window in the skull.

There's a thin skin called the dura that covers the brain.

We open up a flap in the dura, and then we place a device that's smaller than a dime on the surface of the brain.

Then we replace the dura, replace the skull, replace the skin, and we have a wire that tunnels under the skin down to the chest, and we have a pacemaker-type device in the chest.

So everything is completely under the skin.

And wireless.

And the pacemaker, is that a pacemaker-like device?

Is that simply to provide electric power or to record signals, or is that actually doing something with my cardiovascular system?

It provides power wirelessly through an inductive link that's worn on a vest.

So the power comes wirelessly to the device, and then the data comes out wirelessly as an infrared signal through the skin.

That's awesome.

If someone wasn't— if the wearable vest is the only visible part of the system, you know, if someone's taking a shower and you wouldn't be able to tell they have a device.

But it's charging by proximity and data transfer via IR because you want to keep this stuff inside or underneath the skin without a lot of open ports.

Okay, look, I know you're doing this to help people speak and so forth, and that's all very lovely and high-minded, but when it comes time for tests outside of that, just for dweebs

who want to play with it, can I volunteer to be one of your early guinea pigs?

Because I want the in-body BCI.

It sounds awesome.

Yeah, not today, but I mean, the direction that we're moving is that every capability that we build in the clinic can become very interesting for people outside of the clinic.

So even the first participants to get a speech BCI, they can now prompt AI models directly with their thoughts.

When we start building devices for hearing, you know, a hearing prosthetic, you know, think of the applications that you can run in the loop for hearing prosthetic.

You do real-time translation.

Yeah, someone hears Mandarin and Mandarin in English to the brain.

It sounds a lot like what you're doing is by helping people who have a particular— I'm going to— I'm probably butchering the polite terminology here, but let's say a deficiency in their

hearing or movement or whatever.

By tackling those specific tasks, you're also creating at the same time the capability for people to do quite a lot more with their minds than they could otherwise.

So it seems to be kind of like You help people, but at the same time, you also open up this amazing aperture of enhancing human capabilities.

Our goal is, you know, we're giving these devices to people who have disabilities, unmet medical needs, but we want them to come out of the operating room not just with some restored

function, but like, if we can give them superpowers on the other side, that's great too.

And then the process, you know, every 3 years the technology is going to get better.

It's just like a semiconductor play in the same way that Intel got better every cycle.

NVIDIA gets better every cycle.

Paradromics is going to get better.

Every device that we put out is going to have twice as much data as the device before it.

We're going to have more health data.

You know, the devices get safer.

And so this risk-benefit calculation changes.

Right now, you know, the initial markets we're focused on are people with severe disabilities, but we want to build out capabilities that are so compelling that one day anyone would

consider getting them.

This brings us to Tempo, which, as I understand it, is the second kind of major Paradromics product.

And this is to track my moods, mental health.

Tell me about that.

One of the things about the Kinexus system, it was designed specifically for a user population that is— that they live in a wheelchair.

And so a lot of the design decisions revolve around the idea that it's going to be powered off of the battery on their wheelchair.

There's actually no implantable battery in the Kinexus.

Medical batteries are a mess, and we didn't want to have to spec a battery and decide how long is this going to last versus how big is it going to be.

At the time, we didn't like the trade-offs and we said, well, these people have a battery on their wheelchair all the time.

Let's just continuously power it inductively through the wireless.

But we want to start moving into patient populations that are mobile.

They're walking around, they're not in a wheelchair, and so we need to move to a battery-based form factor.

And so we've spent a lot of time reducing the power consumption of the entire system so that we can use a relatively small battery to power it for more than 12 hours a day.

And so we're moving into a world where we want to have these devices be smaller and smaller and more applicable to a larger population.

At the same time, we're pushing to make the data rate higher.

Right now, each individual cortical module has 421 micro wire electrodes.

The next generation will have 1024.

And so each, um, as we multiply the number of electrodes, it essentially reflects how many neurons you can record from.

And so that, that kind of, um, higher resolution— the analogy here is pixels, I presume, or we could think of as Moore's Law.

You know, in Moore's Law, it was how many transistors can we pack into a given area.

You know, in, uh, BCI, it's about how many neurons can you record from in the right areas.

So

There's some nuance there, like if you put 50,000 electrodes in visual cortex, but you're building an auditory prosthetic, that won't be very helpful.

You have to put them in the right areas, but it's really about getting a high density of electrodes in the areas that you really care about.

That's what determines your ability to transfer information in and out of the brain.

And this higher data rate allows the Tempo idea to kind of, quote, capture brain states in real time to inform mental health treatment.

Does that take a lot of data to do well?

Exactly.

So we're very excited about, as we move into sort of mobile patients, having maybe not just motor restoration, but we know that you can read out cognitive states, you can read out mood

using a BCI.

And so imagine what it would be like, a world of mental health where there were objective signals that you could optimize against.

Right now, you go into a neurologist or a psychiatrist's office, they have checklists, they have a big book called the DSM, and they're like, check your scores against that.

They're trying to fit you in a category, and then they have like a bag of possible prescriptions that they give you one and say like, come back in a few weeks.

Like, this is very dated, medieval way.

Yeah, it's medieval.

And, and my wife, just my wife is a psychiatrist, so I hear a lot about this.

So compare to, you know, what, um, the treatment of diabetes was like 25 years ago.

You know, before continuous glucose monitors, the treatment of diabetes revolved around blood sugar from a pinprick, and you had some, uh, scarcity of data to try to optimize your treatment

of diabetes.

Now with continuous glucose monitors, you have constant access to your blood glucose, and that's changed.

Even though CGM doesn't treat diabetes, it's changed the treatment of diabetes, and even people who don't have diabetes are interested in it.

Because measuring your blood sugar can be great for bio-optimizers.

Absolutely.

The same thing is going to happen in mental health.

We'll develop more sophisticated ways of reading out brain states, and that will change how people take medicine.

It'll change how traditional neuromodulators work in the brain.

Like if someone gets a deep brain stimulator, instead of using that in open loop, now they have real-time data to close the loop and help that treatment be better.

Data basically makes medicine better.

Oh, for sure.

I'm just thinking about how we have, you know, if you go to the hospital, they put the little heart rate monitor on you.

They want to keep track of what your heart's doing.

It's almost kind of odd that we don't have more and better brain signal monitoring in general, because I feel like it's another amazing dataset.

I'm just thinking through what you're telling me.

Absolutely.

I mean, everyone— So this does feel like the future.

Yeah.

During COVID everyone learned about pulse ox, and then you were like, oh, actually that makes a lot of sense.

Like $10 thing you can have at home and put on your finger.

Like, that's really helpful to know if you're getting enough oxygen.

And the brain is so much more complicated than those other systems.

And yet, you know, the way we interact with it is so crude.

Okay, so talk to me about the next, next couple of quarters.

So you're going to get the people signed up, you're going to get the system implanted, you're going to start learning.

The last time that Paradromics raised, according to the data I could find, was, I think, 2023.

And it seems like everyone just wants to fire their money into GPU clusters.

So is the company gonna have access to enough capital to fund itself through this next stage of work to get this out into the market as a real commercial product?

And I hope the answer is yes.

It's easier than ever to build a great new product and launching a startup even as a solopreneur is getting easier and easier.

But as my experienced founders already know, there's a lot more to starting a company than just putting up a website or even building a product.

If you're serious about going into business, you need a Delaware C-Corp.

That's going to give you a serious leg up on your competition and you can be taken seriously by investors.

That's where Northwest Registered Agent comes in.

Just a few clicks, they're going to give you the perfect start to your new enterprise.

A real identity, a domain, a custom website, a business email, all the public filings done, and even a phone number.

And you'll complete this process in under 10 minutes.

Plus, you're going to get all sorts of free tools and resources from step step guides, compliance reminders, and an online account that's going to keep everything organized even as

your business grows.

So get all the advantages of a Delaware C-Corp regardless of where in the U.S.

you're operating out of.

NorthwestRegisteredAgent.com/twist.

So last year we demonstrated the highest data rate BCI in the world by a factor of 20.

Now we're bringing that into the clinic in an area that, I mean, let's Neuralink stock is trading on the secondary market at $20 billion, a $20 billion valuation.

We have a number of companies that are raising now at the near or unicorn level valuation.

And, you know, we have the best BCI in the world.

So we're pretty confident about our upcoming fundraise, which will be on the heels of the data that we're collecting over the next couple of months.

There's a lot of interest in BCI.

So BCI is considered by a lot of people in the AI space to be kind of an adjacent companion technology to AI.

No, I think it makes a lot of sense because if we're building synthetic brains and we have physical brains, we probably want to have a way to have them talk and link to one another.

So to me, this is a step in a direction towards me being able to computer without having to open my eyes.

And as somebody who spends too much time in front of screens, bring it on.

Okay, before I let you go, Max, I'm Really loving this, but I do want to ask about like 5 years from now.

Where do we get in like a half decade if you had your optimist hat on?

I'm not going to call you back in and beat you over the head with predictions, but I'm just curious, like when you're at night just thinking about the future, where are we in half decade

with BCIs?

We're going to have a lot of trials, right, in what are called early feasibility studies, which means that people in all different application areas in mental illness, degrees of paralysis,

will have next-generation devices, which at that point will be even better than what you're seeing right now coming out of Paradromics and Neuralink.

It's likely to be Paradromics and Neuralink 5 years from now that are leading because we're the two companies that are building wireless high-channel count, what are called intracortical

systems, and those are the most scalable systems.

5 years from now, you'll see these people will be using BCIs to control exoskeletons, robotic prosthetics.

You'll be reading out cognitive states associated with

psychiatric conditions, reading out mood states associated with mood disorders.

To some extent, those things exist already in the clinic, but we'll have much better hardware.

You'll have in-market devices, or the first generation of devices for communication and computer control.

So you'll see people who are paralyzed and have lost the ability to communicate getting devices to speak, getting devices to control computer cursor, computer keyboard, the, you know,

I think you'll have, you know, in the order of maybe 1,000 people

with really exciting devices of different types.

I think what you'll have is just this menu of offerings for patients and the suggestion of very, very big markets, which have been anticipated right now in theory, but will then be

clinically proven out.

You know, there was that famous moment in the history of, I think it was insulin, when they began to inject people that were like in comas and they all kind of started to wake up because

suddenly their body started to work.

I wonder if we're gonna have a similar moment in which people who are currently disabled in speech or sight or mobility all kind of become brought back to full interactivity in a way.

And I wonder what that's going to look like about in terms of how we approach this societally, because we Once we have that capacity, I feel that we have a moral requirement to bring

it to our fellow humans because

how could we not?

Which I know sounds like a weak argument, but I really mean that from my heart.

So I'm hoping that even though it's fun to talk about the sci-fi element of this, that we do get some way to pay for all the healthcare we can do with this.

And is this going to be price exclusive?

Is this going to be like prohibitive?

You're definitely hitting on the right question, which is that to date, I would say regulators have been extremely engaged with the BCI community.

The FDA is really smart and really on top of it and has, I would say, not significantly slowed down the industry.

Like they've, they've worked hard to make sure that safe devices are coming through this process and try to find safe opportunities to accelerate.

CMS has less staff than FDA.

A does, and they're not up on all of the different, uh, possibilities, uh, that BCIs have to offer the patient population.

And so, like, probably the next factor will be political advocacy, um, education of CMS, because they are the— they're the payer for most of the patients that'll get first-generation

BCIs.

And private payers, insurance companies, usually follow the Center for Medicaid and Medicare Services in determining what they're going to cover.

And so like, it's sort of like, it sounds like bureaucracy, but like probably the most important decisions in understanding how quickly these devices will get to patients and how quickly

the field will thrive have to do with how easy it will be to get good reimbursement for BCI devices.

That's why I wonder if the Veterans Affairs is going to be a big potential early customer because they can make that decision for themselves.

There are a number of researchers in the BCI field that are already affiliated with the VA in some way.

And DARPA, Defense Advanced Research Projects Agency, was the initial funder for most of this research.

So it's absolutely the case that the military is looking at this as a way to help warfighters.

Well, I do love to start my week off with a bit of good news and some optimism and a nice ramp to the future.

So, Matt, an absolute pleasure.

It's paradromics.com.

You can check it out.

And Matt, just before we let you go, is there a job you're looking to hire for where you can't find the right candidate and you want to shout it out into the void in case someone here

watching is the perfect person for you?

Well, for anyone who's listening, I will say that if you live within 2 hours of Ann Arbor, Michigan, we're now enrolling in our clinical trial at Michigan.

And so you should definitely, you should definitely look at that if you have someone close to you that could benefit from this device.

We're always looking for awesome technologists.

And so, and of any ilk, if you are, if you're incredible at what you do and you really want to get into BCI, please, you know, shoot us an email.

We're excited to talk to you.

All right.

Appreciate it, Matt.

We'll have you back on in a couple of months to see how things are going.

Until then, peace.

All right.

Thanks a lot.

I have a question for you.

Have you ever taken a flight?

If you have, I presume it was commercial and therefore you were on a plane that we can probably name a Boeing 787, 777, or perhaps an Airbus A320, A330.

A350, or if you're having a lot of fun, maybe an A380.

But no matter what plane you were on, it was probably a tube with a couple of sticks coming out the side called wings.

That's one way to build an aircraft, but it's not the only way.

And one company, JetZero, wants to take on a very radical approach to airline design, airplane design, if you will, that could save lots of fuel and really change the entire game of

getting people and stuff through the air.

So please join me in welcoming Tom O'Leary, co-founder and CEO of JetZero.

Tom, welcome to the show.

Hello.

So glad that you're here because I get to ask the expert to explain what a BWB or a blended wing body or an all-wing aircraft is.

The people out there don't know, Tom, teach them.

Absolutely.

So simplest understanding of an all-wing is that it is lift across the entire wingspan.

So you have a fuselage or that's blended into the wing, also known as a blended wing body.

And these, these terms are basically a physics first, first principles physics way of solving for the most optimized airframe that delivers the lowest fuel burn, the lowest emissions,

the lowest noise.

And once the engineers go to work on it, it'll produce the best cabin with the best characteristics for the market.

I want to get to that in a minute, but I'm just perplexed because prepping for this, I was looking into different airframe designs and it seems that all-wing aircraft are awesome on

paper at least.

Yeah.

And I know it's not a new concept.

We've been kind of tinkering this as a species for a while.

So what has changed in, I don't know, material science or design tools that allows this idea to go from cool idea but not practical to we're going to build a demonstrator and get it

in the air by next year?

Yeah, there's really just 3 key enabling technologies that take you from a tube and wing to an all-wing.

It's the aero, uh, that's first principle physics, lower drag, higher lift.

And then it's the flight controls because you have to have stability and control of that particular shape.

And then it's the structures, the composites.

So actually, NASA's been working on this for 30 years.

They've put over $1 billion of, of research

into, into this whole, this form.

And so really, that's what we're taking, those 3 key enabling technologies And different from what we did at Tesla back in the day where we verticalized the supply chain, we're just

taking those 3 things, making a new airframe, and then using existing supply chain, existing parts to speed the introduction to the market because it's a great big task.

If you watch the video version of the pod on YouTube, you've probably noticed that Alex and I have radiant, beautiful skin.

I don't like to talk about it, but We all see it, right?

And that's not just a genetic accident.

It's because of our newest partner, Caldera Labs.

They've spent years developing and testing skincare products specifically for men, and they make your skin feel smoother and healthier, gives you a more youthful experience.

And who doesn't want that?

When you're getting up there in your late 30s like me, you get this puffiness around your eyes, maybe even dark circles like I have.

But Caldera's eye serum makes your skin more resilient and leaves me looking well-rested even when my sleep number— yeah, sometimes it dips a bit, right?

I only get 6 or 7 hours.

So it's time to elevate your skincare routine with Caldera.

Head to calderalab.com/twist and use the code TWIST at checkout for 20% off your first order.

That's calderalab.com/twist.

Calderalab.com/twist.

Now, on the lack of vertical integration, definitely different than Tesla.

My, my guess here, my understanding is that there are already very strong suppliers that can give you what you need to help accelerate the process, versus back in the earlier days of

electric vehicles in the US, there was less of an existing supply chain to tap into.

Yeah, absolutely.

And it's just a challenge, uh, but, but aerospace, there's a great supply chain.

We're working with companies, some, some of whom have invested in JetZero, companies like RTX, 3M, Northrop on the military side, they're an incredible partner and enabler.

They're actually, through their subsidiary Scale Composites, building our demonstrator, which is a full-scale demonstration of this, this whole capability.

Now, you guys have raised some money from the Air Force.

You've raised some money from United.

So clearly there's a dual-use affair going on here.

You can use these airplanes for all sorts of things.

And now for people out there who have only ever flown in tube and wing aircraft, Is this a plane, the one that you're building, that's going to be able to, to function at normal airports?

Because one thing I recall is when the A380 came out, people had to retrofit their airports and that made it, I think, less commercially viable.

So does this thing slot in to, you know, my beloved TF Green or JFK?

TF Green, love it.

I've flown in and out of there many times.

I went to high school in Rhode Island where TF Green is.

Yeah, that's Barrington High.

So TF Green was just over the bay from us.

And it's like a lot of airports.

It's a smaller airport that's grown up from the '90s when Southwest came there.

And so it's the question then becomes like, will this plane work within the existing infrastructure?

We've been working with 15, now 16 airlines that come to our hangar twice a year for a couple of days.

And we go through all of the parameters.

Gate access, baggage, seats, bins, everything that you would think of from a commercial standpoint.

The short answer is yes, if it's in a Group 5 gate.

Um, and, and it's basically set up to be the solution as opposed to the problem because it's a shorter plane, so it can actually pull up to any gate and, and fit in.

And that's something that I think is underappreciated about the blended wing because we think in one dimension width as opposed to two dimensions, length and width.

I want to get back to the technical stuff here, but I'm now very curious.

One thing that everyone has to suffer through is airline onboarding and offboarding.

Yeah.

Um, does the fact that the plane is a little bit shorter and a little bit fatter perhaps, um, does that allow us to get on and off faster, or will that not actually impact the element

of what I would call passenger sanity?

Your hunch is correct.

Uh, it will mean faster boarding because You've gone— yeah, exactly, for the wind.

Um, just think about a single aisle.

We know what the biggest single aisles take about 45 minutes to board when full.

And then you get on a widebody which has 2 aisles, or if they board from the middle, it's effectively 4 aisles.

And all of a sudden you're like, wow, I, I was expecting a logjam but it's not here.

Well, think for this plane, it's actually 6 aisles— 4 in the back, 2 in the front.

You board from the middle.

So we anticipate the full boarding time for this plane is going to be something more on the order of 10 to 15 minutes as opposed to 30 to even 45 minutes.

And that's just a whole revelation, right?

I mean, that's a significant pain point.

But more importantly, our design has a place for everyone to have their own carry-on bag at their own seat in a space so that there's no like, you're not sitting outside the gate wondering

like, I better get on board so that I can, you know, be assured that I have it.

You know, before you even walk on the jet bridge that there's a place for your luggage.

Anxiety diminishes.

And like, that's the future that we all want and that this plane and this shape can deliver.

And well, I think if that's the case, Tom, you're going to harm some incidental airline incomes because they do love to charge me for a checked bag because I worry that I will have

space.

So I cough up extra money.

So I wonder, I wonder if you're solving some of their kind of marginal revenue by accident by making a plane that does away with some of those pain points.

What we see in the market right now is the friction is more do they have enough premium seats?

When we talk to airlines now, their, their number one business problem is that they have frequent flyers who are not getting upgrades because people have bought all the premium seats

post-COVID.

And that's a different problem to have.

Like, they're making more money from premium than they are from coach because people are upgrading themselves.

We've finally come to the point where people have— they're, they're voting with their feet and they're buying the premium tickets.

They love that.

So It's really, that's where the market is today.

And you see like Southwest just went to assigned seats.

I mean, we're reading those tea leaves, right?

We're reading those tea leaves and you're seeing that people are demanding a more premium experience when they fly.

It shouldn't have to be heartache.

It should be a great experience.

And we can just forward that.

That'll be changed, but we'll adapt to the change.

It'll be— I have a I could talk our entire time about my views about the airline industry, but I'm going to bring us back to the progress of technology here.

Okay.

So you guys built a scale model, the Pathfinder, that's gone through design, building, and testing.

What were the learnings from that?

Did it go as planned?

And then I want to talk about the demonstrator after that.

But first, take me to the, the miniature.

Yeah.

So there are multiple— Pathfinder is a program, not a plane.

We've done 4%, 6%, 12%.

Right now we're doing rapid iteration in the middle.

The Goldilocks plane for us is the 6% because we don't have to certify it when it's done.

It's just under the drone limit.

So we can do rapid iteration.

We do it on truck tests.

We learn tons.

It's basically like the cheapest wind tunnel on the planet.

Do a change to your subscale model, put it on the back of the truck on a stick, and run it down the airport runway.

This is a great— I mean, it's fantastic.

I mean, we do wind tunnel tests too.

We have a glorious, shining stainless steel wind tunnel model, which great learnings from that.

But sometimes we get more learnings just from that rapid iteration cycle with our Pathfinders, either on the back of a truck or out in the desert, which has been a little bit problematic

this year because the dry lake beds are lakes, right?

You know, that's, that's a challenge.

Darn rainfall and improving reservoir access.

What are we going to do?

It's slowing down our— on these, you know, truck stick demonstrator Pathfinder and then, you know, running it around.

I'm almost surprised you're saying that because when I think about what JetZero is building, I'm thinking about, you know, the future.

We're taking an old concept and we're updating it for the modern world, new technologies, and you're not using digital twins.

So far as I can tell, you're not doing as many digital simulations as I expected.

You're doing real-world out there with, I presume, high-vis paint.

So why am I wrong in my expectations about the best way to design aerodynamics today?

Well, it's actually all of them.

So digital twin is core to our strategy, and we will be the first large jet manufacturer to have a system that's built on a digital twin.

And where the digital thread ties through from design to test to certification to manufacturing to service, that is absolutely core to what JetZero is doing.

But it's, it's all of the things, right?

And it's the right tool for the, for the right time and the right job.

So we're absolutely using quantum compute, computational fluid dynamics accelerates us in lots of ways, but Sometimes there's just nothing better than putting a scale model, dynamically

scaled, on the back of a truck and running it down the runway.

And you learn things in real world that, you know, it's wild.

And it's actually, there's nothing that gets the team more excited when you're just around the hangar and then all of a sudden here goes the scale model team is out on the runway and

everyone like runs to the window and sees the thing, that they're calling the tower and asking them to hold flight.

We were in a commercial airspace here, and they're super— down in Long Beach, California.

Down in Long Beach.

And the, the airport here is super helpful with us, and they'll integrate.

They'll be like, we just closed taxiway Kilo for you, go for it.

And it gets everybody— it's just excitement, right?

Like, I know when there's a new CFD run, people aren't like going around some computer to say like, oh my gosh.

But when that thing hits the runway and then plane spotters come out and they post it on Reddit and stuff and it's, it's a thing.

Yeah, it's great.

That's fun.

I will say that with a wind tunnel and, you know, operating knowledge of CFD, you could also just branch out and build an F1 team.

You're— you have all the ingredients, you know, just, just an idea.

You don't have that there for the future.

You know, it's amazing is my co-founder Mark Page, who's the aerodynamicist who ties all the way back to the beginning of this type of work.

Back in the '90s when Boeing and McDonnell Douglas merged.

His boss was like, you're an innovator's innovator.

You fly, my pretty one.

You need to leave the nest and go innovate.

And the first thing that he did was he started to work on IndyCars.

So open wheel racing.

Yeah, yeah, yeah.

It was back in, you know, '99, 2000 when aero started to become— now it's like a controversial, like one little tweak and everybody's in an uproar.

But back then it was, it was, you know, it was in a sense in its infancy compared to where it is today with CFD.

So that's, that was one of the places where he's learned a lot, uh, and it actually had a surprising amount of contributions to his understanding of the blended wing.

So it's, it's pretty, pretty interesting background.

And people say that racing is a waste of money.

No, you see, it's actually how we get better, better airplanes for the future.

Okay, let's move to the demonstrator.

This is the project you guys have had pipped for 2027 since you first, you know, raised the money from the Air Force.

So are we on target?

And 2027 is a big time zone.

So when should I expect to hear from you guys about the demonstrator taking off and flying away?

Yeah, second part of '27.

The beauty of this team is extremely extraordinarily capital efficient.

We've, we've made huge amounts of progress with relatively small amounts of money from an aerospace context or perspective.

The plane will be built this year.

It'll go integration and then into series of tests, right?

So you're testing, you're ground testing, and then you get into flight test.

Second half of '27 is the expectation.

And everything has been up until this point on time, under budget.

We've hit all our milestones.

We've hit them in time.

We have NASA and the Air Force.

It's an Air Force program, but a whole, a whole host of NASA folks are on loan, people from Air Force research labs, industry partners.

The FAA has, has a few people that, that are in that program.

They're vetting everything we do.

We don't clear a milestone without their say-so and their approval.

And we've, we've hit the time and the budget to a tee so far.

And that in and of itself is a pretty major win.

That's an enormous win.

Nothing in the world of megaprojects goes on time and on budget.

So points.

Now, you mentioned a whole host of people that are interested in what you're building, investing in it, helping out, lending people, whatever.

And that's because the way that I see it, if your plane, the Z4, will end up as efficient as hoped, you could reduce the per-passenger— the passenger-mile— no, I'm going to butcher

this.

Fuel burn per passenger mile.

Fuel burn per passenger mile.

That's a mouthful.

Yeah.

By up to 50%, which is a staggering amount of money.

Operating costs for airlines are often predicated on fuel costs and profitability too.

So do you think that you'll get up to that 50% savings in the first iteration, or is that more of a long-term goal for the technology and maybe the second version of— Yeah, in a first

iteration, well, it's all about what you're comparing to.

So that, that 50% is comparison to things have been in the market for a while and that are flying.

But, but, uh, United and Delta are still flying a 767, uh, cross-country.

And so again, right, that against that aircraft, those are pushing into their 30s now.

And so they're replacing those aircraft.

But as we compare to those, it's in that up to that one plane, up to 50%.

The way to think about this is it's 30% aerodynamic efficiency above a tube and wing.

That's, that's the L over D.

So that's still a pretty staggering piece because as you, as you point out, it's about a third of the cost for, for an airliner is fuel.

So if, if you can make a massive dent in that, you know, even at 33%, you're still cutting something like 10% of their cost structure.

That's massive, right?

I mean, you just think about the billions of dollars that they spend even just in fuel costs on— is that also why the military is interested?

Tankers getting freight around— the US military has bases everywhere, so logistics is a big part of the task.

Is the fuel savings the core thing, or is there something else that your plane can do that is particularly appealing to the DOD?

It's the flip side of the coin, right?

So lower fuel burn for them I mean, yeah, they want to lower cost and lower emissions and all those things.

But keep in mind, we're working at the behest of the Air Force and Air Mobility Command, right?

Dominance is fighters and bombers and air mobility is tankers and transports.

Their motto is you can't kick ass without the gas, right?

So I thought that was a tanker motto.

It is.

It's basically right.

But it's, but it, but, but this is the deal, right?

So that's the tanker motto.

And So their focus is, but what is that efficiency that has lower fuel burn, lower emissions?

What is that to them?

It's how do I take a huge amount of payload across huge distances?

You think about Operation Midnight Hammer, you know, that was, you know, a handful of B-2s supported by a couple, three dozen tankers.

Yeah.

So that's what it takes.

And so if you can lower the fuel consumption, then you're actually increasing the range.

It's just the flip side of that coin.

So same reason, different kind of a benefit.

Okay.

So the question that I had pinned down about this entire idea of greatly more fuel-efficient aircraft is, you know, can all-wing aircraft take over the U.S.

or global airline industry?

And one thing that I found you guys talking about was the underserved middle market.

So help me understand the economics of different plane sizes in the airline context and how the Z4 might fit into that and unlock something perhaps, as opposed to just replace the aging

767s of the world.

It's just a beachhead, right?

It's like if you've got this amazing technology and you're going into an existing market as opposed to opening up a new market, you're just a question of product market fit.

We all know what product market fit looks like.

It looks like a product that the customer can't not buy.

Ding, ding, ding, ding, ding, ding.

Who doesn't have one of these, right?

Like, only smart people.

You can't not buy a smartphone and operate in today's world.

So that's product market fit in a nutshell.

And then, so when you look at the, look at the airliner market, for example, there are no planes between 200 and 250 passengers.

So the single aisles top out 3-class.

So you got to normalize.

So you're doing an apples-to-apples comparison.

You got to normalize for what class is 3-class, you top out at 200 passengers in, in that biggest narrow body.

But then the bigger— the, the smallest wide bodies, you don't start coming in until 250 or even more, right?

And so we just looked at that and said,

how could there be a better product-market fit?

We can take everything that the airlines want— low fuel burn and all of the rest and we can bring it to a place where they have no plane.

And this just seemed like the, the most obvious product market fit beachhead for me, having worked in tech.

You know, I worked at eBay, Tesla, dealer.com, then got into aerospace, uh, over a decade ago, worked with beta technologies, worked on product market fit with all those companies.

When I saw the potential for product market fit here

It just blew me away.

There was no— I've never seen anything like it.

There's literally no product in the middle of a very mature market.

And the TAM is trillions.

But you just want to be able to have that beachhead, prove that thing, and then grow from there.

And it will eventually.

By 2050, all planes will be all-wing of a certain size and greater, because how will you compete?

And we're seeing that, right?

Like Airbus is already saying that publicly.

They're like, yep, this plane is inevitable.

And we're not doing it just yet, and we're probably going to do it with bigger planes, not smaller planes.

But they're clearly messaging that that is an inevitable future.

It's just a question of who can absorb the risk

and take it to market.

I just find it really, really amazing that Boeing is too afraid to make a new airframe.

And here you are with the financial equivalent of like $8 and two sticks doing that and building the future at the same time.

Kind of an indictment of Boeing?

I'm not trying to pit you against them, but that's the way that I think about this.

Okay.

So I think it's more— I think let me just— because I would not indict Boeing.

And we have many people here who came from Boeing and who have great affinity for that company.

We have admiration.

We have a phrase around here, which is respect your elders, right?

Where we're providing an alternative.

But ultimately, this is what startups are meant to do.

It's not an indictment of the company.

It's basically saying, let's remember, sometimes when a change is needed and a paradigm needs to be shifted, it's not the incumbent who can do that.

The risks are too high.

They're not set up for it.

We can provide that and they will ultimately benefit, I believe.

Yeah.

And there's a book written about this, something about a dilemma and people who make change, I think.

Yeah.

Maybe you've heard about it.

Innovator's Dilemma.

Yeah.

Okay.

I know you guys are doing manufacturing in the United States.

Long term, it's going to be based in Greensboro, North Carolina.

My question is, what does this thing cost?

Now, commercial operations are supposed to be early 2030s.

So clearly demonstrator to commercial is going to take a little time.

But what do you think it's going to cost and how will that be in kind of a seat-to-seat comparison with, I don't know, the 737 Max or the 787?

Yeah.

So cost is proprietary.

It's going to cost billions for us to put it together.

And these planes are, you know, coming into the upper mid-market where planes are selling pretty deep into the 9 figures, right?

It's not— these aren't $100 million planes.

They're more like $200 million planes out the door.

And list prices are a lot higher than that.

It's very common to discount in the space.

It's not common.

It is, you know, a de facto, steeply discounted.

And so prices are very closely protected.

Our feeling there is, is pretty straightforward.

We, we can project the costs of the program and of the development of the manufacturing facility.

And we're using existing supply chains, so we're not guessing, right?

We're looking at, hey, we've got to produce a clean slate airframe.

If we're leveraging existing supply chain for those parts, then those are known commodities.

And so we have a really good handle on the cost and the idea that we can create margin, particularly because we're lowering the cost.

So that, that's the way we look at the cost picture.

But The TAM is huge and our ability to sell into that TAM is going to be amazing.

It's just about people want to see that proof point.

That's why we're doing a full-scale demonstrator.

Absolutely.

And for anyone curious about the math on TAM here, when United invested in JetZero in 2025, they had something like a loose agreement to buy 100 with another 100 maybe down the road.

200 times $200 million adds up pretty quickly.

And that's what we're talking about here.

Not, you know, selling software seats for $10 a month for 20 people.

It's an entirely different category of TAM.

All right.

So last question for you, Tom, is just, is there any remaining technology risk that you can see, or is this just simply build a demonstrator, get it out there, test it, learn from it,

and iterate?

But there's nothing enormous you still have to leap over.

There's no chasms in the road.

There are no knowns, right?

There are no knowns that you have to burn down the risk in, in terms of your test on flight control, stability and control.

On composite structures.

It's like when we're building this full-scale test model, there are no knowns.

That's why we build a pressurized cockpit test, because we're going to put humans in the cockpit to fly it.

So we build one, we destroy it, and, you know, we do a destructive test, and then we move on and build the final article.

So those are the, those are the kind of things.

But as, as we were pointing out earlier, We're talking about decades.

We had two, not one, but two X-planes, X-48B and X-48C, that NASA did.

And the chief engineer of that program is here working at JetZero.

So these are, you know, there's a lot of known knowns.

It's just about execution.

I feel like we're all dancing around the old Rumsfeld quote about, you know, unknown unknowns and known unknowns.

Slightly unfair question, but I'm really glad to see the progress you're making.

I am desperate for some sort of innovation in the airline space.

I would love to have a better experience as a commercial flyer and as a United guy myself, I'm going to be flying on your planes.

So, Tom, will you come back next year when the demonstrator is ready to go and tell us how it's going?

Oh, you bet.

We'll show you.

Fantastic.

We'll bring a picture book.

Actually, last question about capital, because I'm really curious about the venture world right now.

AI is just hoovering up all the money.

It sounds like.

Now, I know you raised from B Capital and a bunch of like RTX Ventures and so forth, but you will need more money.

Do you think that the venture world is a fit for JetZero moving forward, or are you going to look for other sources of capital?

Because there's a lot of ways you could finance this.

I know it's been challenging up until now, and the momentum right now is really strong.

I think people are starting to see that hardware is a great buy.

Dual-use hardware is an even better buy.

And if you've got a really solid use case for both of those, then you can do what, do what we're doing now.

And you see that in our Series B, we've turned that corner early on.

There was a lot of, wow, this looks incredibly risky.

I don't, I'm not sure.

How can you?

It's very small.

It's, we've been in business for 5 years and we've gotten to the point where we've proved that there's traction with supply and there's traction with demand.

And the full-scale demonstrator gives people that proof point.

So what we're hearing from growth capital folks right now, which is the round that we're moving into, is this year is putting together our growth capital round.

You know, B is kind of the end of your venture-style rounds, and it's going really well because, you know, what happened in the market a couple of weeks ago, right, where it's like

everybody's wondering if AI is just gonna completely demolish software, and people are saying, aha, hardware, hardware, hardware, this is a, this is a solid place for capital right

now.

All right, Tom, you got to leave it there.

Come back next year.

I want to hear all about it.

And also, I cannot wait to take a flight in one.

I'm gonna guess 2032.

I think that's my guess for you guys right now.

So we'll see how that bears out.

Tom Atreid, we'll have you back.

Thank you.

All right, thanks so much.

We're going to take a little bit of a detour into a part of the startup world that we don't spend enough time on, which is wellness.

This is a topic that is near and dear to my heart because I used to be very, very unhealthy, but a little bit of rehab, a little bit of exercise, and some actual food has put me back

onto the right course.

But I have been a very unhealthy person, and I know what that feels like.

It makes you slow, it makes you feel stupid.

You don't take good care of yourself, you don't do good work.

And that's why I wanted to talk to a company called Nutrisense.

They are working on helping people track their own glucose, make better goals for themselves, handle their stress, and generally live better and more productive lives.

So please join me in welcoming to the program, it's Dan Zavorotny, the co-founder and CEO of Nutrisense.

Dan, hey, how you doing?

Appreciate it.

So Nutrisense, when I think about technology startups and founders, I do not think healthy people.

I think a lot of late-night Red Bulls and pizza, but I presume that you are here to change that for everyone out there.

So first of all, tell us what Nutrisense is and then target demographic to start.

Sure, sure, sure.

So when we first started, we were the first company to take continuous glucose monitors.

And these are devices that you put on your arm and historically use glucose to make sure that you understand how you respond to food, stress, sleep, and exercise, right?

This is in some ways like the Oura Ring or the Whoop of nutrition, right?

However, it's been used for type 1 diabetics who take— so we said, what if we get people who don't have type 1 diabetes but type 2, not insulin, prediabetes, weight loss, PCOS, Hashimoto's,

longevity, everything you can imagine.

And everyone said, you guys are crazy.

No one is going to put this in their arm.

It's going to penetrate skin.

Why would anyone want this?

So we launched anyways.

We said, why not?

And it turns out a lot of people want it.

And after that, the mass market flooded, and there's more and more people ending in space.

Uh, so that's how we started.

However, we said, you know, that's not enough.

You need to create infrastructure.

You need to create the ability to understand what this means.

And so then we added a layer of software on top, which helps you put in, you actually track your carbs, macros, your micro macronutrients, great.

Pull in all the data from your other wearables.

And then the third piece is how do you have another human coming in, a doctor or dietitian, and help you interpret this data?

Because it's very, very complicated.

And so that's what we've been doing for about 7 years now.

We worked with over, I think something like 120,000 to 130,000 people at this point.

Wow.

Wow.

That's more, that's a lot.

Now for folks out there who are not as familiar with how your glucose moves or your blood glucose levels in relation to food, can you just explain, um, that relationship and why it

matters?

Sure.

So glucose is one of the most important metrics for your health, basically.

Uh, when you eat something, your body takes, especially if it's a carbohydrate food like a cheeseburger or a burger, uh, your body takes that glucose, that carbohydrate, and it breaks

it down in your system.

And converts from carbohydrate to glucose.

That's what spikes glucose in your system, in your bloodstream.

And then that glucose is usually either absorbed into— A, into your liver, or B, into your muscles and converted into glycogen, right?

And the way that works is your pancreas releases insulin.

That's what pulls it out.

And we unfortunately in the US and most of the world have become incredibly pre-diabetic.

Right now, a third of Americans are pre-diabetic.

And the reason for this is we're eating so many carbohydrates and so much saturated fat that our bodies cannot produce enough insulin to pull it in the right places.

And so it ends up just floating around your blood instead of going into your liver and your muscles.

And that's a bad thing.

And so when that happens, you have all these cardiometabolic issues like heart disease, diabetes, uh, weight gain, fatigue, stress.

Yeah.

Yeah.

So essentially our blood is too sugary and it's not good for us and we need to take better care of that.

Now, this brings us to monitoring.

I have never had diabetes, prediabetes, anything like that.

So I'm not as familiar with glucose monitoring and what goes into it.

So can you tell us in the Nutrisense context how you help people monitor their glucose and how, if I can be impolite, intrusive that is?

So it's not that intrusive.

It's a little microneedle that you put inside of your— on top of your arm and it penetrates your skin.

A couple millimeters, but you feel nothing, right?

And it lasts about 14 days.

And what happens is that data gets sent directly to your phone.

And what you're doing is you're just talking to an app and you're saying what you're eating.

And I can show you guys, by the way, as well, if you'd like to see an app, it might be easier.

Yeah, well, I mean, we do love a demo here on Twist, Dan.

So if you want to show it off.

And also, as a big fan of using voice dictation to both write and also to interact with AI, This is right up my alley.

So here you have a glucose chart and it shows your glucose, exact glucose, so minute by minute that shows you what your glucose is responding to everything your day, right?

And so then if you scroll down here, if you across, you can actually see what is driving the glucose response.

So here this person worked out and they had a glucose spike after the workout.

And the reason for that is your muscles release glycogen in your body to help you work out.

And that's positive, right?

Okay.

Versus here, this person had a dinner, and they recorded, they had, you know, they had avocado, cucumber, broccolini, arugula, pretty healthy person, and some tuna.

And again, their response was 93 milligrams per deciliter, right?

So it's a lot of data.

And for someone like you or me, we probably go and biohack and understand it, but it's kind of complicated, right?

And so how do we simplify that, right?

The way we did it is twofold.

We basically added dietitians so you could grab a, grab a time with dietitian, press I want to talk to dietitian, and insurance will cover it.

Means that doesn't hit the deductible, it doesn't do anything, just a free call.

That data will be reviewed by the dietitian and she will, she or he will give you feedback on how to improve and what does this data even mean, right?

And so what we've seen is pretty fascinating.

Like we've seen people that stay around for 12 months or longer with us, they're losing something like 20 to 30 pounds.

Right.

And this is, that's a lot.

It's a lot.

This is no prescriptions of drugs, pharmaceuticals.

This is purely just understanding your glucose, understanding how you respond to things.

So for example, we've seen certain factors, like there are people who respond very well to potatoes, but you give 'em rice and they're spiking their glucose in diabetic range for 2,

3 hours.

That's very interesting.

So essentially, even if I'm eating foods that don't always spike other people's glucose, I could be accidentally pushing myself towards prediabetes without even knowing it.

Exactly right.

And it's those repetitive spikes over and over and over.

It's kind of like if you think about, you know, if I did, you'll give you a little light tap on your shoulder, you won't feel any pain.

But if I give you a light tap on your shoulder 30,000 times, sooner or later you're gonna have a bruise.

And it's exactly what's happening to glucose and diabetes, right?

When you have glucose spikes once in a while, it's not a big deal.

But when you do consistently for 30 years, that's the issue.

And so if you identify these issues ahead of time, you could prevent it.

Dan, one thing I'm really curious about is the inclusion of dietitians and humans in this loop, because here on TWIST, we've been talking so much about AI and automation.

And to me, what you have is a large dataset of people's glucose levels and how they react to food, and then individuals.

And it seems to me like a place where I would kind of expect you to be pitching an AI solution versus a licensed dietitian.

So talk to me about that choice and why humans matter in this particular context.

Context?

I wish overnight everyone said AI takes over the world, but when it comes to health, there's just too many differentiations between all of us, right?

We have our historical disease states, we have how we're feeling.

Like, I think one of your colleagues before said, like, you broke up with your husband or your wife, or your boss is yelling at you.

It's too hard for AI to understand that context.

But when you talk to a human, they're able to track all that context and put it into the systems.

They remember it over time.

So what we did is is we added the contextualization through the individual.

However, we took that contextualization and we added AI to it.

And so now you can have do two things.

One, you can actually make the interactions between dietitian calls way more impactful with AI.

And two, you can make the dietitian way more impactful as well.

So historically it's taken an hour for dietitian to come and talk to you so they know everything about you.

Yeah.

Now they press 2 buttons, we create a report, they review the report, and now they talk to you about what you did 3 weeks ago and they remember it.

So it feels more personalized.

That's not me taking notes or me remembering what we talked about a month ago, but going directly into you and you specifically.

So you feel heard and understood.

So essentially, instead of having AI jump in and do an entire process start to finish, we're instead using AI to, to allow and extend humans to do what they do best, which is understand

fellow humans.

And therefore not only can the dietitian probably see more people and therefore do more business, if you will, but also me, the user, gets personalized information without massive waits,

which in the healthcare world, yeah, I mean, that, that almost sounds like a magic trick.

I mean, try to get a dermatologist appointment, you know?

Exactly.

It's fascinating.

We call Nora, right?

Okay.

And so what Nora does is she takes all the conversational data the human has with dietitian and she puts it in here.

She takes all the glucose data, all the nutrition, your workouts, everything you can imagine, puts it in here.