00:00So tell us why you're going to go public through a spec as opposed to listing through an IPO.
00:04What was the thinking behind that?
00:06Well, the bigger picture is robotics and physical AI are hitting every worksite everywhere.
00:13And I'm a career robotics entrepreneur. I started my first robotics company 20 years ago
00:18and started it with the idea that robots really could transform life for humans.
00:23And that is starting to happen in big industries like construction and agriculture and warehousing
00:28and defense. And we see that through our 600 customers.
00:32And so we have grown tremendously through inbound demand.
00:35But there is so much more demand than we can handle with our current staffing and current resourcing
00:42that we needed to find a way to scale.
00:45And so in evaluating all of our options, doing this back deal made the most sense.
00:50So this is because of speed?
00:52Speed. It is mainly so that we can scale up our technology and our go-to-market and our operations.
00:58And our international expansion strategy.
01:01So it is to serve the demand that we have from our customers because everyone is scaling.
01:05But the spec is a great way to do it that is efficient.
01:10We have an incredible partner.
01:12Tom Bushy, the CEO of Newberry 2, has tremendous experience in robotics and physical AI already.
01:18And so it's an efficient way of scaling the company up.
01:22Let me ask you a question. You mentioned physical AI.
01:24What is the difference between robotics and physical AI?
01:26Well, I always say it together, robotics and physical AI, because it's kind of the same thing.
01:33So we have a super big tent approach.
01:36We think that anything that moves is getting controlled by AI.
01:40And so that's physical AI.
01:42Robotics is a subset of physical AI.
01:44So a lot of people think humanoids when they think robotics.
01:46But it really is anything that moves starting to get controlled by AI.
01:50I like that simple explanation.
01:51So AI can hallucinate in software.
01:54What does hallucinating look like in your world, like when there's a forklift?
01:58I mean, how can that hallucinate?
01:59And I can imagine that must be pretty scary if it does.
02:02Absolutely.
02:02I mean, you just said why we exist is because AI makes mistakes, and we've all experienced that, right?
02:08And so if you've ever been talking with a chatbot and thought, you know, that just doesn't make sense, imagine
02:13that in the physical world.
02:15Like people could die, property could be hurt, and so there needs to be an independent set of guardrails that
02:21sit between the AI driver and the machine itself that ensure that the machine follows the rules, whatever the rules
02:27are.
02:28Can you talk to us about maybe hallucinations or failures that maybe you didn't anticipate that happened in real life?
02:35I just find it hard to imagine.
02:36I haven't ridden a self-driving car, for instance, because I'm scared.
02:39No Waymo for Isabel.
02:40I mean, Waymo is a company that is very safety conscious, and they have done it incredibly well.
02:48And they would tell you that it took them 18 months to get through a demo, but then 15 years
02:53to get to scale.
02:54And a lot of that was getting the safety exactly right.
02:58And so if physical AI is going to go out across the economy, across all of these different kind of
03:03work sites, not everybody can spend 15 years with Google-funded dollars to get there.
03:08So that's, again, why we exist, is to be this independent source of trust that everyone can subscribe to rather
03:13than having to do it themselves.
03:15So you say trusted safety is a hidden bottleneck in scaling physical AI.
03:21How was that achieved before Fort Robotics came along, before you guys applied what you are able to do to
03:28the technology?
03:29Well, the field is called functional safety, and it involves analyzing the risk of any kind of machine and then
03:36designing mitigations and then building those mitigations.
03:39And this is a practice that applies to anything that moves.
03:43So planes get this, cars get this, industrial robots get this.
03:46And functional safety for robotics has really revolved around 1961, when we got the first industrial robot arm installed on
03:55an automotive assembly line.
03:59And it has not yet been adapted for this new class of machine that's moving, that's running AI, that's working
04:04around people and property, that's connected to the Internet.
04:07Totally different risk profile.
04:09So the legacy safety that worked for these machines that were stationary, surrounded by a cage, no longer works for
04:15this new kind of machine.
04:16So you need a new approach.
04:18That's why we're here, is to create this new approach.
04:20And you're pitching your company as a universal safety layers across all of these physical AIs.
04:25What is stopping other robotics companies from building that layer themselves?
04:29Well, this is a, well, it's really hard.
04:33Safety is really hard.
04:34Doing it well takes setting up a third-party audited development process and then getting your technology done and then
04:42getting that audited and then taking that through third-party certification.
04:45So it's kind of like developing a drug.
04:47And nobody wants to actually do that if this is not their business.
04:51They want to do the job of moving pallets or digging holes or picking apples or whatever it is.
04:56And so that's why safety is always outsourced.
04:59In every other machine industry, safety is always outsourced.
05:02But in developing a drug, there's an FDA.
05:04There's a government entity involved.
05:06Is there a government entity involved right now in physical AI safety?
05:11It is a bit balkanized.
05:13So the Europeans are doing it differently than the North Americans and doing it in different places in Asia.
05:18And we're a global company, so we're selling it to every region.
05:21And so we get a little bit of glimpse of all the different.
05:24And it's a mess.
05:24And it's a bit of a mess.
05:27But there are standardization efforts happening.
05:30We're sitting on several standards bodies.
05:31We're helping craft those standards.
05:33The international standards are really going to be the thing that guides technology development.
05:37And then the regulations will point to the standards in different ways.
05:40The American way that regulation will point to the standards will be different than the European way.
05:44But the Europeans have legislated that if a machine is running machine learning, it has to have, by law, an
05:50independent governance device.
05:52So that is why we exist.
Comments