SMAC Technologies raises $32M to build deep reinforcement learning models for military decision-making

Let me tell you about Sentry. Century shows developers what's broken and helps them fix it fast. That's why 150,000 organizations use it to keep their apps working. And up next, we have our last guest, Andy Markoff from Smack Technologies. He's the co-founder and CEO.

Andy, how you doing?

Great. Thanks for having me on to the show. Uh, first time in the show. Please introduce yourself and the company.

So, Andy Markoff, co-founder and CEO of Smack Technologies. I'm a what I like to say a recovering member of a cult known as the United States Marine Corps. Spent the first 10 and a half years of my professional career there. Got out um tried to figure out what I wanted to do when I grew up. And uh luckily one of the Marines I served with, my co-founder Clint Elenise got out and we both decided to work on building the the decision-making tools we wish we'd had when we were running the kill chain during our careers as you know Marine Raiders at Mars and like try and give back and like find a way to serve again after we've both gotten out. So we talked about this idea you know really fall of 2023 and then started the company in January 24 really. So what smack is we are the first frontier AI lab solely focused on building for national security and our mission is what we would call decision dominance but how do we take you know pabytes of sensor data multimodal data streams that you know too much for a human to actually analyze and then convert that analysis in real time into the right decisions across the range of military operations whether that's trying to figure out what are we doing for the next one to six months what are we doing for the next one to four days and like literally right now what are we doing been trying to make that process seamless and what right now is a pretty siloed and slow system.

What are you building on top of? Like it feels like Palunteer exists. There's databases like it's not like technology was just invented but AI was obviously new. So uh like how how are you integrating with systems? What what are the tailwinds here that allow you to actually deploy systems effectively and quickly? Yeah. I think I mean at the end of the day we are not building on top of another like AI model. We're building our own model. Yeah. Um and I think that's you know the reason for that is general purpose like large language models are are very useful and they're very good at analyzing massive amounts of textual data. They have a lot of really you know important use cases but they are not the right tool for 80% of military decision-m like and some of that's just they're trained on labelled data sets. There's no label data set for World War II. You know, hopefully we never have that data set, but but you need a different type of model to

make decisions on multiple time horizons, do the type of hard, you know, physics grounded geospatial reasoning required for most military decisions. And that's, you know, deep reinforcement learning is a is the right approach for that problem set. So we're we're building a deep reinforcement learning model and instead of you know labeled data sets we're we're training that model in an environment that's been built by you know physicists who are grounded in like the physics of a peer level conflict but also domain experts and I think that's a lot of our advantage and secret is there's there's a lot of information and expertise about how to fight wars that are in people's heads. They're not it's not in a document that can be read. that need to be encoded and put in an environment to to make a system that can actually function in that type of environment.

Interesting. So, uh I mean a lot of the foundation labs are going to uh data providers, data brokers, hiring experts to write down their SOPs. You're you're obtaining data from the US government. Is that correct? Like what is that what does that pipeline look like?

Sure. So so building this data is you know some of the domain expertise we have hired on the team. I mean, you know, Clint and I's background, I was a I was a joint fires instructor for a while at the Marine Corps's version of Top Gun. Clint kind of spent his whole career like in the fire space as we did at Mars.

Um, we we've hired a lot of people on the team that were the absolute best at their military discipline. They were instructors and they did it on deployment. And then we work with our end users. And I think in some ways this gets to how do you trust how do you trust AI? How do we get users to trust AI? Like we the users are part of our training data generation but like taking their knowledge their understanding of the right way to do things today and helping bring that into the environment in a way that you know they can trust that like their knowledge and their understanding is actually being used to help the model get to a better start point.

What uh ways in which uh conflict is evolving are you guys like willing to lean on and effectively bet the company around? It's obviously great to have the uh all the the uh time actually enlisted and and serving the country, but uh the battlefields like evolving. You know, I'm sure you've been uh you and the team have been watching all the footage coming out of the last few days. Uh but but where where is this all kind of going in your view

as far as where is like kind of conflict moving? Well, an example is like I think we saw like the US version of a shahed or I think uh like so so uh lowcost autonomous systems uh seem to be you know coming online in mass and and uh all these things are going to impact how you build the product.

Sure. So I mean I think at the end of the day you know the scaling and oper operationalizing autonomous systems is a large part of the future of warfare right and so I think that the concept that we talk about internally that we need to build and enable is what we would call intelligent autonomy but how do you orchestrate all of these autonomous systems not just against tasks that they've been assigned to but with each other and with the more exquisite expensive systems and how do you do that in a way where you still have a human in the loop right I think there's a lot of discussion about fully automating the killchain. No one wants that. That's not even really something that I've heard anyone talking about. What what fundamentally people are trying to do is have the right amount of human in the loop. Have humans for high value human touch points. It's not everywhere humans are making decisions today. Like we can't have humans involved in every single thing that they're involved in today, but a lot of the a lot of the decisions are not high value decisions that are humans uniquely positioned to do. So it's intelligent time is about removing humans from low value human touch points, keeping them and bringing them back into the system for those touch points that they need to make the decision whether for ethical reasons or for tactical reasons and enabling them to make decisions that help move you know hundreds of thousands of you know autonomous systems and man platforms and you know other types of unmanned platforms towards common goals across a you know what could be a 100 million square mile theater and like I think that's That's really the spec that we have to build towards.

Uh, makes sense. What what's the shape of the company today? Where are you guys based? Uh, what are what are you uh what are your plans with this uh with this new capital?

Sure. So, we are, you know, team of 18 today. Um, really split mostly between Elsaundo. We're like the majority of the tech team is Elsagundo and like a headquarters in in Texas. We have we have a physicist, a physics lab like down in Texas. part of the tech team is there and and really our product the way we think about it is like we need to have a model like a heavier weight model that is available for people that work out of an operation center or not like compute limited and then you know lighter like lighter versions of that model that can work at the edge at the front lines where people are going to be bandwidth constrained and so a lot of this year is getting what currently we're deployed uh under contract with the air force the Marine Corps and the Navy is to expand across all six services both at the command level but also to take what right now we're prototyping is the edge version of the model we built and get that deployed to frontline units across all six services by the end of the year.

Take us through the fundraising round. How much did you raise?

So we've raised 32 million to date um 26 million. today. Um, you know, we've been, frankly, just really fortunate to have the investor team that we had. I mean, like like I was saying earlier today, you know, Clint and I knew nothing about fundraising, procurement, government acquisitions, like we didn't know. We just like were wandering around in the dark, you know, back in January 2024 trying to to get moving. And having uh 72 kind of take us early on, believe in us, kind of teach us the ropes of the defense tech space was tremendously valuable. And you know, we've been really fortunate to have, you know, Geodess and Kosanoa co-lead the series A um to really help us figure out how do we get to the next how do we get to the next scale? How do we start scaling the team? How do we build the systems that allow us to grow rapidly? And we've just been really fortunate um to have supportive investors that understand the vision and and kind of knew what we need and knew what they could teach us at different phases.

What's the origin of the name? So smack is actually a tactical task that you would call over the radio to strike a target. So when I was, you know, we'd be out on a raid, you know, Afghanistan, Iraq, and like you'd have a hostile target that you were going to call in an air strike on, you know, the the conversation that you just smack that target is actually like a task.

Um, so that that was and I think, you know, fires fires is one of the military's functional areas, but I think that's an area that's, you know, Clinton's specialty. That was a lot of our initial go to market was in the fire space and we were initially heavily focused on that but obviously we're expanding this year to kind of the range of military decision- making.

Is it is it true that that you hired someone named John Kugan?

It is uh it is true. I actually was getting really confused earlier today when like I I saw yours. I was like wait why why is John why is my John on Twitter? He's also a Marine too. I was like Marines don't even use Twitter

for X.

Very cool. Yeah. Go around and tell everyone. Yeah, I hired John Kugan.

Here you go.

You're like, "What?"

That's very fun.

Uh, awesome. Well, great, great to meet you and, uh, I'm sure you'll be back on soon.

We'll talk to you soon.

Awesome. Thanks for having me. Cheers,