National Design Studio launches Ramparts, a 15MB browser-native PII redaction model that runs entirely on-device

their country.

Pure disrespect.

Anyway, uh there's a little golf cart. We got to talk about this at some point, but uh there's a new there's a new car. It's like a Twizzy. You're going to love it. It's

close. It's no Twizzy for you.

Let's bring in our first guest.

Anyway, let's bring in our first guest

from the National Design Studio.

From the National Design Studio. Welcome to the show, gentlemen. How are you doing?

Thank you so much for coming on the show. Uh, please start with an introduction of yourselves, the the the company, and then the announcement today.

Uh, my name is Edward Corstein. I run engineering at National Design Studio. It's technically not a company. It's a government organization.

Oh yeah, that's right. Sorry.

And I'm Tiger. I'm one of the engineers at National Design Studio.

Okay. And today the launch, take us through it.

We're launching Rampart. It's a local first privacy model that puts people back in control of the data that they share with AI.

Um we were we were just like, you know, kind of building a chatbot for fun.

Yeah.

And we were upset that uh none of the frontier models will actually fit in a browser. So you cannot do PII removal in the browser. Uh which is, you know, pretty damn important for our use case. uh you just have to like you know trust that the server is actually removing the information uh and not lying to you. So we're like okay well what if what if it was just all on device like personal data never had to leave your device uh it was just you know secure by default.

Okay so open source the weights are on uh hugging face runs in the browser under 15 megs. uh a technical user could go right now download the model from hugging face vibe code their own Chrome plugin and have it be running uh however they want but how do you see this actually rolling out? Do you want the government to inter to to implement this in various places? Do you want companies to are are is it sort of like open up the primordial soup of ideas and see where it goes or do you have do you have like a rollout strategy that you are advocating for? Um well the reason why we open sourced it is because we do want companies to use it and we want people to use it and we want people to make it better. So we want uh vibecoded chrome plugins.

Sure.

We want we want you know vibe coded tragedy extensions like what whatever value is derived from the product. You know this is just like a a total side quest for us. We just want to build software that's that's helpful for the American people. We've already launched a series of products then like Trump X has got 15 million users. It's saved over $500 million in in drug costs and

you know we rethought the UX there. So we're you know basically across everything we're working on we're just trying to find the first principles best approach for users.

Yeah.

And this just came as the derivative of that. Uh we're not MBAL you know researchers or engineers. We're just like you know we should just do this.

You created PII super intelligence. That's what people are that's what people are saying online.

Basically

it's like tiny intelligence. It's like the It's by far the smallest model. Like the other ones are like at least 50 megabytes. This is 15.

Yeah. So, did you like did you to what degree did you build on the shoulders of giants? Is this some uh pruned and distilled and fine-tuned open-source model? Is this something where it was easier to just start from scratch but use architectures that are more uh prevalent and wellestablished? Like how did you actually go about training this model? We tried 72 different base models.

Whoa.

Um and you know put them through a training set. We ended up on miniLM. So we definitely are standing on the shoulders of giants here.

Yeah.

Yeah. Yeah. We we we started taking a look at the open AI privacy filter that just got released recently.

Yeah.

Um we're trying to figure out is there a way we can you know just quantize it? Can we maybe remove some of the parameter? Like what what can we do here to try to make use of like the the state-of-the-art model and um you know we tried a lot of things. we just could not get it to fit into, you know, we want this to work on like legacy devices, um,

on an old Android phone, for example, or, you know, an older iOS device. Um, and it it just would not get small enough and still make any intelligent sense to try to actually run it. So, yeah, we we ended up uh essentially uh it's technically a fine tune, but we were we we we trained many and basically made it do exactly what we wanted to do.

Can you help me understand use cases a little bit more? because I feel like most of the time when I'm transmitting a document to a prescription website, RX or uh a financial institution, uh the PII is like the the potentially the only important part. They're often sending me a blank form and asking me to put my PII in there. What is the inverse scenario where I want to redact my information but I still need to transfer something because in most cases that would just be the template or something in my in my estimation.

Yeah, the the flags are set at compile time. So you can decide like for our use case it's really important that we have this data or that we don't have this data.

And so we hand all the customization back to whoever wants to use the library. uh the model just says oh you know this is a phone number this is a name uh this is a surname etc etc and then ultimately it's it's you know whatever you want to do with the the model you can just do it um fundamentally what we were looking at was there are a lot of cases where people will ask a question pertaining to a document of like okay for example with the template how do I fill out said template um because you know the government is pretty bad with forms there's like way too many forms nobody knows all they mean like you got to pay people to do government forums.

Yeah.

Um so like that that was a use case we had in mind. PI is like not not super helpful for that.

Um and it's also kind of like the breaking point. It's where uh you know the product will lose trust. So we're like okay two birds one stone. Let's build this thing.

That makes sense.

Uh how do you guys how do you guys think about side quests at the National Design Studio in general? Like I imagine every single day there's opportunities that come up and you guys are in a unique situation where

Yeah. sort of design. they have a mandate, but at the same time, there's so many different places that

the government uh you know, you know, interacts with people's lives. I'm I'm very curious.

It's pretty hard to pick what to work on because there's a lot of exciting things. There's like everything is huge scale. Everything could be way better. Um maybe not everything, but a lot of things. Um so there's like a huge calling for side quests. Uh but we we we just tried to keep everything in line with our vision which is like we want to make uh the American digital experience better

and then we we've kind of chosen a track to get there and on the way we built this model and on the way we built Trumper X but we're excited to see uh to see how it

develops from here and back on the show. Uh yeah, like diving more into that, do you have a reference point in in tech people might ship, you know, they might think in quarters, financial quarters, three month cycles. They also might think about a two pizza team, which I think is like 10 people. Uh do you have an idea of of where the sweet spot is from what you've experimented on how many people do you want to bring into a project and then how long do you want to spend there so you don't get stuck for a decade because you might not have a decade. Yeah, I mean there's definitely a lot of a lot of places to get stuck because the the visibility is super low in a lot of these projects and you don't know how broken they are until you're like you're really in it.

Sure.

Um that that you know being able to determine that in advance is like is definitely you know AGI level.

Yeah.

We have a really great team. We're we're very fluid. We're we're constantly trading responsibilities back and forth. You know someone might be you know better at doing you know one part of the tech stack than somebody else but they're on a different project. We'll just borrow them for a day. It's or even for an hour. It's it we we share a lot of responsibility at the studio. There's this is also definitely the only place in the government where people work seven days a week

consumed, you know, on Red Bulls. Um I I think the ideal amount of people per project if they're if they work super hard is two.

Like what one design person, one engineer, and they both have like, you know, full scope and then they're able to call on people as necessary.

Yeah. Yeah. two with the caveat of you're you're calling in your co-workers say, "Hey, can

yeah,

take a look at this over my shoulder quite frequently."

Yeah, that makes sense.

What What's your what's your guys's pitch to talent that that uh that you might want to recruit into the National Design Studio, I imagine. Uh uh lots of people that would join could get a go get a blank check from a venture fund or could go work at some of the best companies.

Everyone that has joints, you know, that's the case for turn that off. Yes. Um it's definitely more for people who are super missionoriented. Um you know who who the hell like what what great engineer wants to come work in the government? You know the answer is typically nobody unless it's you know like the IC where there's really interesting problems to solve. Um so I I think that we have like the a super golden opportunity. Um at least the way I evaluate problems I try to see how big the problem is in terms of like how many people will use it. uh the delta between what exists versus what our team can do and how fast we can do it. When you look across those three matrices, it's like a a home run place to work. So, I think that is its own natural kind of calling card for the right kind of talent that we need for the studio.

Awesome.

Complex, too. It's it's also a huge benefit. It's pretty sick.

Awesome.

Where is that? Is that where you guys are right now?

We're not there right now, but we're about to be there. Yeah.

Awesome. All right. All right. Well,