Regardless of how far removed I get from the days of summer vacation, there’s forever a bit of melancholy in the air in the wake of Labor Day weekend. Heck, my brain even likes to toss me the occasional school anxiety nightmare for old times’ sake.
While I’ve got no class to speak of (just ask my exes) this semester, things are certainly ramping back up. As I mentioned last week, I’ve got a quick trip to Pittsburgh for Robotics & AI Demo Day on September 16, where I’ll be conducting a fireside interview with Seegrid CEO Chris Baker.
And then, on September 22, Automated is hosting its first-ever West Coast live podcast and happy hour, where I’ll be interviewing Rodney Brooks (photo from our Boston happy hour featuring recent pod guest Kate Darling above). I believe we bumped up our initial capacity cap, but as I write this, we’re very close to selling the thing out, so be sure to secure a spot here.
You can also catch me hosting one of the stages at A3’s Advanced Vision & AI Conference (AVAC) in Santa Clara on the 23 and 24. 50% off code here -- just tell them Automated sent you. Or don't. It won't affect my hosting fee either way.
Something I didn’t mention last week is our plans for Automated's first birthday. Simply put, we are podcasting. A lot. We’re dropping a full, brand-new episode of the show every day next week, Monday-Friday.
It’s a killer lineup:
Amanda Prorok of the University of Cambridge
Dhruv Batra of Yutori
Clara Vu of Magnendo
Carolina Parada of Google DeepMind
Tony Zhou of Sunday
In fact, we’re about to drop so much content on your heads that we need to get creative with this newsletter’s formatting over the next couple of weeks. But hey, we’ve got a couple of days to figure that out. I’m sure it will be fine.
Here's a nice one-year reel our editor put together for you to watch on loop until next week. Or, you know, check out this week's episode with OpenMind below.
Founders spend years crafting origin stories. A great one can be foundational to a successful pitch deck and the eventual memoir and major motion picture.
OpenMind’s tale won’t do much in the way of scoring the physical AI startup its own Hulu miniseries, but there’s something nearly universal in its portrayal of everyday disappointment.
Jan Liphardt, a Stanford professor of bioengineering, bought three robots for himself and two sons. Decades of movies and television set high expectations for the impending arrival of the bipedal and two quadrupeds.
“When all the boxes arrived and you open them,” says Liphardt, “first you start charging lots of batteries, then you plug in the batteries, and then with a lot of luck, the robot may be able to stand up and then it's standing there. And it's not doing anything at all.”
Upon much study of standards around dynamically stable robots, this is, perhaps, one of those rare examples where one can unequivocally say that a humanoid landed with a dull thud.
“That then immediately led to, ‘what's the right software architecture here?’" Liphardt posits. “And what kind of chips are in the system? How do we start turning this inert piece of plastic and metal into a friend, companion, helper, coworker? How do we make it speak 50 languages? How do we make it navigate my neighborhood? How do we make it remember everything I've told it? And how do we personalize its behavior for specific use cases and scenarios?”
The CEO quickly reiterates the OpenMind’s founding story even more succinctly, as “having a bunch of hardware drop shipped to my living room and then realizing that it basically just didn't do anything at all.”
Liphardt founded the company in 2024 with the seemingly simple premise of helping robots be helpful right out of the box. OpenMind describes its core value add as, “[making] it easy for people to access and configure robot workflows, and we make sure every robot has the base intelligence to operate reliably and safely around humans — turning capable hardware into actual robot outcomes.”
“The sky’s the limit,” Romain Moulin laughs. The metaphor is, perhaps, a bit on the nose for the company behind vertical AS/RS (automated storage and retrieval system), Skypod. The CEO is responding to a question about Exotec’s decision to expand internationally early on in the fulfillment automation service’s existence.
The more terrestrial answer is one having to do with corporations. If you want big customers, you need to be international. Regardless of the depths of one’s pockets, no one wants the headache of adopting an entirely different, non-compatible automation system across geographies.
“As soon as you start being at a level when you discuss with big brands,” Moulin adds, “they expect that you follow them everywhere in the world.”
It’s a quick ramp up from humble beginnings. “When we went to sell the first system, we had not much to show,” he notes. “Really, we had a small robot. We could prove that we could make robot move, but we did not have the Skype pod as you can see today on our site. So, the first page of our PowerPoint was our resume. Our experience was the biggest stuff we had to show.”
Moulin and cofounder/CTO Renaud Heitz weren’t titans of industry in those early days. They did, however, both sport CV bullet points worthy of pitch deck inclusion, including shared histories at GE and French AGV (autonomous guided vehicle) firm, BA Systemes (now Alstef Mobile Robotics, courtesy of a 2018 merger).
Founded in Lille, France in 2015, Exotec was one of several robotics firms whose existence was spurred on by Amazon’s Kiva acquisition three years prior. The creation of Amazon Robotics quickly shifted expectations around online commerce. Global retailers suddenly found themselves competing with the promise of same- and next day delivery.
Exotec and its ilk formed to serve customers hoping automation could help them keep up with the Bezoses. Moulin notes that the company’s European origin had a direct impact on Skypod’s creation. Kiva’s AMRs, he explains, are the product of a geographically massive country with equally massive distribution centers. For Exotec, the sky had to be the limit, because there was simply no other direction to build.
The final 3% of humanoid deployment can consume nearly all the work. A robot that performs well in a lab still has to handle traffic, animals, wet pavement, moving leaves, different languages, and countless other edge cases before it can operate safely in an unstructured environment. Watch on YouTube>
Rajat Bhageria (Chef Robotics) - Restaurant economics ultimately didn't make a lot of sense for Chef Robotics, so the startup pivoted to an entirely different sector of food service.
Kate Darling (RAI Institute) - Is emotional attachment to robots actually a good thing? Can we give human workers a stronger voice in the conversation around workplace automation? These big-picture questions and more.
Nic Radford (Persona AI)- A lot of companies claim to be building humanoids for the real world, but few walk the bipedal walk quite like Persona AI's ultra-rugged welding systems.
Humanoid reliability concerns extend beyond how systems operate on factory floors. As Unitree's recent Shanghai market debut demonstrates, these things can be even more volatile where investment is concerned. A 45% drop in share prices plunged the Hangzhou manufacturer's valuation from a high of $66 billion down to $30 billion. Still nothing to sneeze at, of course, but Chinese regulators are understandably looking for ways to avoid replaying that particular rollercoaster ride. According to a recent report from The Information, new rules would require companies to prove innovation, demonstrate revenue generation, or show a steady reduction in loss prior to IPO approval. Such regulation is likely going to be frequently put to the test in the coming years. Volatility or not, fellow Chinese firms with far less established track records than Unitree will no doubt be looking to make their mark on the market.
On a completely related note, here's Agility's pre-merger SEC filing. I'm sure there are those moment when going public feels like a bit of a Faustian bargain -- a big influx of investment that requires a company to pull the current back on its internal machinations. When Agility announced its own SPAC plans the week of Automate, it pitched these both as net positives, positioning itself as the only humanoid maker with the chutzpah to expose itself in such a manner. There's certainly truth to that — and to its claims of having a meaningful head start on a number of fronts. Another truth is that head starts can also be a mixed bag in the tech world.
The Robot Report crawled through the filing, and rightfully notes that — much like in the case of the above discussion — valuations are predictive. Investors are betting on what companies will become, and there are far too many factors determining that outcome to mention. If I had to hazard a guess, sunlight would be extremely harsh at this stage of the game for a vast majority of humanoid makers. Much of this filing concerns the imminent launch of the still unannounced Digit V5 robot.
"Agility’s commercial deployments are concentrated with a limited number of warehouse, logistics, and manufacturing customers and a significant portion of Agility’s revenue is concentrated with a limited number of customers," per the filing. "For example, we currently have one customer that has committed to deploy $300 million of Digit v5 over a multi-year period, subject to the realization of certain contractual milestones, product features and specifications. Agility may be unable to meet such contractual milestones or deliver such product features and specifications and realize such revenue, which would materially adversely affect Agility’s business and operating results."
Other interesting V5 tidbits included therein are the $ 8,500-a-month RaaS licensing model and "a weighted adoption curve of approximately 800 [units] in 2027, approximately 7,000 in 2030, and up to approximately 25,000 in 2035."
After I quickly touched on Andreessen Horowitz's Machine Age Fund, Becca has taken a deeper dive into the $1.1 billion hardware bet. She spoke with MassRobotics executive director Tom Ryden and investor Aaron Tainter about some of the wider impact a16z's first all-out foray into robotics could have for the broader physical AI space. After all, $1.1 billion doesn't get you all that far in today's investing world, but a16z remains one of tech's most influential venture firms. I certainly wouldn't say they're the tip of the spear here (though they've already dipped their toes in the water through investments like Skydio and Mind Robotics), but when they pen a good, old-fashioned blog manifesto, people pay attention.
Something a bit different this week, as Liam takes a look at a Washington State University study utilizing AI to improve how high-performance copper–chromium–niobium alloy, GRCop-42, is 3D printed. The material has proven an important one in the construction of NASA rockets, and this method could dramatically reduce the resources, cost, and time required to build these systems via additive manufacturing.
I spotted a familiar face tucked away, hanging out in a corner, when I visited Boston Dynamics' HQ back in April. Sparkles had its moment in the sun two years prior, racking up a respectable 2.6 million YouTube views before retirement. The sheepdog's legacy is set to live on in the form of a spinoff startup from former chief product officer Marc Theermann. Dynamic Creatures looks to be operating in a similar space as Disney Research, building autonomous robot puppets for real-world settings. "The company is creating a new category of mobile characters that move naturally through guest environments, understand the context of an interaction, and respond in real time," it writes in its announcement post. "As Boston Dynamics’ official entertainment and hospitality partner, the company is already working with a large theme park and retailer to develop guest-facing character experiences designed to feel personal, attentive, and memorable. "
The Association for Advancing Automation (A3) is North America’s largest automation trade association representing more than 1,400 organizations involved in robotics, artificial intelligence, machine vision & imaging, motion control & motors, and related automation technologies.
.jpeg?upscale=true&width=1120&upscale=true&name=images%20(14).jpeg)
