Juxta's John Ferrara on synthetic fingerprinting: GPS alternative that works indoors, underground, and in combat zones

have someone in person. Who is it? It's John. How you doing, John? What's happening? Welcome. Thank you.

Grab a seat.

Better in person.

Yes, even better in person. What brings you to LA?

Well, a couple of other meetings, but I figured I had to stop by.

Yeah.

And you know, little known fact, I was actually in journalism school.

Okay.

Before along with data science, I was also majoring in journalism in Northwestern.

Mhm.

And if this had been an option when I was still in school, I might have not dropped out and just tried to get a job here.

Who are your heroes? Who Who's the Who's the GOAT journalist in your opinion?

Oh, man. Well, I dropped out for a reason. didn't learn anyone's name.

Uh, that's funny. Uh, well, uh, reintroduce yourself. Tell everyone what you're working on, what you're building.

For sure. So, I'm the founder and CEO of JXA, which is a GPS alternative that can track anywhere on Earth, outdoors, indoors, underground without relying on hardware like satellites, beacons, or cameras. So,

that's a bunch of buzzwords to say without external

without external hardware, right?

You still need your hardware. We use uh IMUs that are already installed on devices like phones, robots, embedded devices, medical devices.

So, no magnetic fields involved.

No, that's kind of a um

different tech, right?

It's a it's different tech. And and the key differentiator in what we've built is something called synthetic fingerprinting, which is this idea of being able to simulate IMU measurements at scale for different people and objects. And the value behind that is essentially from this desk here, we could map out and simulate uh an environment anywhere in the world, a Amazon warehouse in Boston, and then have that up and running like in an hour.

Okay.

If you wanted to typically use like a magnetic um approach, you actually because it's hard to simulate magnetic movement because it's usually dependent on objects that are hard to predict being in the environment.

Sure.

You actually have to go to that warehouse, collect that data manually, and then train models on that. Mhm.

Um and so our idea is that the ultimate scalability with any positioning system is one that can be deployed fully remotely cuz I can't possibly go to, you know, every corner of the earth and install beacons, cameras, or you know, collect fingerprint data.

So what's the what's the textbook device? Android phone, iPhone.

Yeah. So I mean every every mobile phone today has an IMU inside of it. So like the same thing that's powering your Compass app, right? Accelerometers, gyroscopes are what we rely on.

Are they getting better? because I feel like the compass app came out I don't know a decade ago and it works as well as I can tell.

I mean they've gotten better like so the traditional problem you'll see in the like inertial measurement unit space if that's one space that you find yourself in is that drift is pretty

prominent. So um you know in theory if we can do something that's faster, better, cheaper and more accurate you know everyone's like why haven't this been been done before right so the idea is typically because IMUs drift a lot

so if you want to actually you know track someone over or an object over a long period of time typically they veer off course and then you end up losing

the example is the phone's here you know the phone's here to start then you raise it one foot the IMU knows you went up a foot you put it down a foot the IMU knows you went down so you know where you are probably accelerometer like a altitude sensor but we're talking about um usually speed and orientation vectors. So the simplest way to think about it is right if you know your starting position and you're stacking vectors related to speed and direction

um that's called dead reckoning. That's the approach where you're not actually that's just like very drift right and so what our what we do with our simulation and then our kind of um the models that we've built

is what mitigates drift by like 90 plus%.

Yeah.

And is able to track people and things over long periods of time. So the value ads are places like in defense and military, right? Cuz GPS gets jammed like instantly in any war zone. Um in indoor environments like warehouses, hospitals, uh any sort of like logistic center. So those are kind of where we target first. Obviously our long-term goals sort of be like a like I said a pretty widescale GPS replacement. Um you can imagine like that way down the road eventually someone like FEMA being able to use this to find survivors of natural disasters under piles of rubble where a GPS signal might not be able to penetrate through. Um but in the near term those commercial opportunities are pretty

Do you have any customers yet or do you have pilots running and how big are these firms that you're working with?

Yeah. So we work with like I said kind of those three industries mostly right now. Logistics, healthcare and defense. Um we have three customers. Our ACVS are in like the mid to low six figures. So um those deployments the first deployments are actually going on in March um which is exciting for us.

And do they have to change anything on the hardware side?

No. So we're we work kind of um out of the box by design. Again, the whole goal is

we want to eliminate as much especially compared to like RTLS solutions like beacons or cameras. Um we want to eliminate as much friction as possible. So we integrate directly onto device um and they don't require any setup.

Yeah. How did you get into this? Were you just annoyed by messy noisy data and you were like I got to fix that.

Going into GPS denied zone.

Yeah. You find yourself there all the time. Well, that's the same reason why if you're ever like driving in a city and it tells you you're going the wrong way on your maps app and it's really annoying. That's why um for reference those build even outdoors those tall buildings get in the way of the signals. But I got into it um largely because both of my parents were in the US military in the Navy. And so um for a cumulative set of years they were deployed between Afghanistan, Asia, um all over. And so I think I kind of was exposed early on to the idea of like wanting to know where your parents were when you're a kid.

And it's funny how you develop these like random hyperfixations maybe at that age that feel random and then over time um you know an elementary explanation from my mom or dad kind of converted into like a prolonged interest.

That's cool. What's the status of the company? How much have you raised?

Yeah, so we raised uh we out of YC, which we were in the summer 25 batch, we raised a little over $5 million.

Let's hit the gong.

Nice.

You ever hit the gong for this race?

Congratulations.

It hits different in person, too. No, it's better in person. But um raised 5 million led by CRV uh out of YC

Charles River Ventures. Correct. Exactly. Well, I Yeah, I did not I should have probably known what that stood for before I took the money. Uh I learned that after the fact.

You locked in.

You looked it up.

Yeah.

Uh so how big is the team

right now? It's seven. Um it's a pretty technically dense problem and so essentially all of our talent are like the top there's a very small community um of hyper talented engineers and researchers mostly in academia who focus on like no hardware tracking, simulated data, all that stuff. We've been fortunate enough to probably have what I feel comfortable in saying is probably the highest concentration of those people at a single company today. Maybe maybe Whimo or maybe maybe Tesla who does a lot of LAR and SLAM stuff. But um yeah, seven people pretty much all engineers

and and the IMU manufacturers, they don't want to uh get into this side of the software problem for their clients. Provide an API that is less noisy. Yeah, it's like a probably a business opportunity for them, but it's um it's a pretty drastically different space than like manufacturing sensors that are like the size of your pinky nail into into sort of building like simulations and and all these like drift loss functions that are like again pretty concentrated in the knowhow among the 10 or so people that can actually build that

across the world.

Yeah, makes a lot of sense. Uh what was the biggest thing you learned in YC? Well, YC was interesting for us because we are I think like 95% of our batch was B2B SAS or like agents I think even maybe um so which was which was Yeah. Yeah. You were there. Uh

we got agents for your agents.

Yeah, it was a lot of agent infrastructure companies. I think you were uh

you know we're actually in our office at Union Square's browser base's old office. Oh, no way.

So hopefully we're summoning some of the good juju from them. But um no YC was good for us because they basically shove all of the

uh non- aent companies into one group office hour. So like our our our like eight companies in our weekly meetings was like

uh like a missile company um like per I think Percy's defense um um Nox Metals if you're familiar right like uh us robotics like a robotics company um there another couple guys building like magnesium alternatives so it was a pretty eclectic group um nobody was really focused on the same thing but they were really good conversation so my favorite part about that was probably the group office hours

um if you can like finagle your way into the eight or nine deep tech companies.

Yeah.

What are you what are you trying to prove out uh before the

before? Well, I mean realistically like I think for us

it's like a technical question more than there's like a market question, right? Cuz again if we're coming to right now 90% of all human and object movement globally is in GPS denied areas which makes sense if you think about it. Like right now we're in one anytime you're indoors underground in a combat zone there's no location data being tracked. So you're missing tons of analytics. So I think the appeal is pretty strong in terms of the need and we can do it if it if our technology works as we say it does at a fraction of the price and a fraction of the time with the same accuracy and reliability. So essentially it just comes down to proving out the tech. Um and we have a lot of we also have a lot of pre-orders aside from the customers we already have that I think will propel us to all the revenue benchmarks that are important for us at this point. And it's essentially proving out um that we're able to again mainly reduce drift meaningfully so which so that we can track people for hours or eventually days on end without it being a without it being a significant problem.

How I got to ask how you're thinking about privacy security all those things. I'm sure the internet if you're successful you one of the most hated people on the internet just from the from the skits.

Well I've been seeing all the hearings online of all the Yeah. the hearings get pretty tense. Other modification is

Yeah, exactly. So, I'm like maybe next up, but um but uh uh no, I mean privacy is essentially something that like we're usually selling to vendors or companies who have end users on their side. Sure. So, by and large like our direct interface is not with the people who are being tracked with their location. That doesn't mean it's not important for us. Obviously, we have our methods about it being encrypted. Um working a lot with defense and in those environments has to be very secure. Uh but generally like for like we work with healthcare for instance you know tracking staff is usually a matter of well does it fall within the union rights like is it something that uh if it's on a hospital issued device usually they're allowed to track them without it being a problem. If it's on a you know personal device then it's obviously comes down to the direct discretion of the the employee

and even the personal devices like you know Apple can track you but Apple has spent probably billions of dollars talking about how they treat privacy seriously. you sort of know that if you open up the maps app, you're painting the GPS,

right?

And and and and for what it's worth, like we're using also the standard infrastructure that Apple is also using in their devices. So like the actual things that we're going off of in terms of hardware have already been green lit and approved by hopefully institutions that people trust. Um again though, it'll ultimately probably come down mostly to the interactions and the relationships between the customers that we sell to and their customers.

Um you know, my long-term vision though is right now we're selling commercial. It'd be great over the course of you know 5 10 years you can see a use case where you know there's a consumerf facing version of of JXO instead of Google maps or Apple Maps you get the added benefit of again a navigation tool that will never lose your direction when you're in a big city you can route it from your house to your exact appointment room at a hospital it'll give you routing on the road to the parking once you're in the hospital navigate you to the room so there's those easy you know added benefits but more than that like when I think about FEMA tracking you right are people is are people going to download a FEMA to be tracked by FEMA in the event of a natural disaster probably not. So I think about like in terms of how we can ultimately eventually interface with people directly is we might partner with companies or agencies like FEMA or whoever else who want to track location in sparing moments

and they basically as a user on like Jux Maps you can opt in

to FEMA tracking you if you live in a natural disaster area and one strikes and then that's sort of a consent process that we would deal with directly with the customer but for now that's probably much too ambitious and way down the road.

Yeah. Also, the phone would probably want something like this. Like, you know, Apple just like randomly rolled out like, "Hey, if you're if you don't have cell service, like you can send one text message via a very slow satellite, and everyone's like, "Awesome. If I'm ever lost hiking, I'll probably text someone." Uh, what does the data side of the business look like? Is this a big data problem where you need to actually source a ton of data to run a model on top?

So, so the big again the big breakthrough with us has been the the synthetic fingerprinting process. So, we're the first company that I know of that's built an IMU simulator at scale that's as reliable as ours is. So, essentially we can give be given an input of an we do a lot of asset tracking too. So, we do a lot of like like warehouses and and different like shipments. Um given the dimensions, weight, height, whatever of a object or person, we can then simulate it in our physics space engine, have them moving around a 3D space. We because we render we basically take in a map or a satellite image of whatever the space is.

We use computer vision to render it into this 3D model. Mhm.

Then we simulate um all these people things moving around the space and create these basically anchor points, benchmarks of what IM measurements are going to be at certain spots given how that person is moving and where they started and all that other stuff. And that allows us to to to benchmark that data. Um so we're able to that ability to synthetically generate it is what makes it so scalable. in terms of like again long-term really long-term visions like my anticipation is we'll be one of the largest holders of all labeled map and floor plan data in the world by a good margin right there's actually it's a pretty untapped uh but important space to have labeled data as we can imagine right yeah and and and because the the input the only input that we get from users is the floor plan or the satellite image of the space they want to track

so if we were to partner with let's say a major grocery store chain across the United States you know we'll probably be the company at the moment that has that much labeled data about aisles, shelves, um, checkout lines, everything like that. So, that's maybe not a that's more of like a monetization route down the road in terms of data that we need as our input. The synthetic process kind of allows us to circumvent a lot of that.

Interesting. Cool. That makes a ton of sense. Well,

great to get the update.

Congratulations. Hang out hang out for a little bit.

Yeah, we'll hang out after. We'll do. Thanks. We got to plant the bomb. We got to tell everyone thank you for watching the show. What do we have? There's one more video I want you to watch. I don't think you've seen it yet.

This one is for you.

Okay, I'm excited. Let's pull it up.

Sharing it with team

going on. We got through a lot of stuff today. Oh, I think I know what you're pulling up. I think I know

you guys got it.

Let's see. Bluewater Autonomy had a announcement. Okay, let's let's watch this. POV nightly real scroll, but you're performatively into physical media. So they print them out. I could get into this.

Commenting.

Commenting. Okay. Double tap a like. Okay.

This is This is This is how you're going to start consuming short form.

Yeah. People People always lament the fact that we don't print out the tweets anymore, but a lot of the tweets are videos and you can't print those out. But maybe you can. Maybe you actually can print out the videos. Uh

uh. Anything else you want to talk about before we get out of here?

No. fun show today.

You don't want to talk about the Hermes leather pool table in emerald green? Plant that bomb and we'll tell everyone about the leather pool table.

It's emerald green. This would look great in the Ultra Dome. It's only $264,000. What a steal for an Hermes pool table. Absolutely. And they put the H on the Q ball. I like that little detail. Let everyone know you got the Hermes down for generations. Leave us five stars on Apple podcast and Spotify. Sign up for our newsletter at tbpn.com. And we will see you tomorrow at 11 a.m. Pacific Shark. Goodbye.

Nice work, brothers. I'll see you on the next