Hugo Robot: Cute Prototype, Big Claim, Open Questions
“Introducing Hugo: the first human companion robot.” That is how Marc Theermann, CEO of Dynamic Creatures, opened a post on X on 16th October. It is a bold sentence, and it is one his own company does not use everywhere. His LinkedIn post about the same robot makes a smaller claim: Hugo is the company’s first living character robot.
The Boston Globe reporting on the same launch had something far more useful to say. The Hugo in the company’s video is teleoperated (i.e. someone is in control of it remotely) and it cannot yet do physical contact with people on its own.
Here’s the story as I see it. A lovable prototype from a serious team, dressed up in a label that is way ahead of anything we’ve seen so far. I want to take both halves of that seriously. I’ll cover what Hugo is, what it can’t do yet, why its cuteness is such a clever design decision, how it stacks up against the robots that have come before it, what the research tells us about robot companions and where I think this all is heading.
Meet Hugo
Hugo is a purple furry two legged character with large eyes, small horns, a flower on the head and a blue nose. Underneath the costume, the company says, is a robotic platform with 43 actuators, plus animatronic character design and expressive movement. A video of Hugo was released on 13th October. The press descriptions state the video shows Hugo walking, dancing, high fiving engineers and hugging children.
Image Credits: Humanoids Daily
Hugo runs on SnowJay (the company’s AI and robotics platform) which they describe as a combination of learning based control, real time perception, behaviour programming, character design and costuming. The published feature list includes safe stop, controlled shutdown and a controlled fall. That last one tells you what the engineers worry about when you’ve got a two legged machine operating around people. The company’s own site also states that the interaction layer uses third party reasoning models in addition to their own.
The people involved are legitimate. Dynamic Creatures is a spinout from Boston Dynamics which emerged from stealth on 8th September. Theermann was Boston Dynamics’ Chief Strategy Officer. Co-founder Farbod Farshidian was a Research Leader at the Robotics and AI Institute. The seed round was co-led by Kindred Ventures and Eniac Ventures with Heliad, Sunshine Lake and BlueGrass Ventures also participating. Boston Dynamics founder Marc Raibert is an angel investor. The amount raised has not been disclosed. The company positions itself as Boston Dynamics’ official entertainment and hospitality partner (Fortune have reported that Boston Dynamics has no equity in the company).
One further detail is important in relation to what comes next. Theermann has said the company is firmly in favour of a cartoonish design and will probably never produce a photorealistic robot. He has also said the aim is to make people at ease with robots by making them cute and cuddly. Keep that in mind. That is one of the company’s most deliberate decisions to date.
What the video does not show
Beyond the teleoperation the company says Hugo is still being developed to be able to understand each interaction and respond in real time. Humanoids Daily pointed out that the clip gives no indication as to how much of Hugo’s behaviour is autonomous or whether the robot is ready to be working unsupervised amongst guests.
There’s a loose end there. Runtime Wire reports that Theermann’s X thread states that Hugo runs on a Vision-Language-Expression model (the aim of which is to link what the robot sees and hears with its gestures and expressions). The video description says they’re developing that model. One account is in the present tense, the other in the future. I haven’t been able to find anything to resolve which is the case.
There’s nothing else in the way of specs. No published price, no height, no weight, no battery life, no sensor list. They say they’re working with one theme park and one retailer (neither of which are named). The hardware itself is unconfirmed. A monitor in the video showed a skeleton that Humanoids Daily felt resembled a modified Unitree G1 humanoid. It called that a visual inference and left it unconfirmed. There’s no independent hands-on test that I can find.
To the company’s credit it doesn’t hide the human in the loop. It’s already said that the initial deployments will be operator reliant, that it has hired puppeteers to work out the robot’s emotions and that it intends to move over time from one operator per robot to one operator looking after many. First pilots are due in early 2027. That’s a perfectly honest account of where the technology is at. It’s not a description of a companion.
Why cute is a smart bet
In 1970 the Japanese roboticist Masahiro Mori published a short essay in a journal called Energy. His contention was that we become more comfortable with robots as they become more and more human, up to a point. Beyond that point, as the human likeness gets very close but not quite there, our comfort drops into what he termed the ‘valley of eeriness’. The phrase ‘uncanny valley’ came about in a 1978 book.
Mori didn’t call it science. In a 2012 interview he described it as being advice to robot designers (to stop at the first peak, before the valley). A horned purple yeti will stop well short of the valley. You won’t mistake Hugo for a person and Theermann’s no photorealism line is the same instinct. That’s good design logic.
The idea has critics. A 2018 scholarly re-evaluation contends that the uncanny valley is regarded as a settled phenomenon when in fact it is not. Treat it as a rule of thumb rather than a law.
Image Credits: The Boston Guide
Cuteness has a second job, and this is my view, not a finding. A cute shell lowers your guard. Theermann is open about the aim being comfort. Comfort is a feature for the guest and a selling point for the company, and it’s interesting to note how well the two line up here.
Hugo against the field
Hugo’s closest rival isn’t a startup, it’s Disney. Olaf the walking snowman from Frozen made his debut on 29th March at Disney Adventure World in Disneyland Paris (Hong Kong Disneyland is next). His walk was learned via reinforcement learning in a physics simulation (same technique as used for the BDX droids in Galaxy’s Edge) with the Frozen animators providing the poses and motion as training data. A human operator controls what he says via a remote which is out of the guests’ view. Guests don’t currently get one on one time with him and Disney have no date for hugs (though that’s the stated long term goal). Disney Imagineering’s Kyle Laughlin has confirmed that Olaf will be going to Disneyland and Walt Disney World, just no timeline. In Paris he’s currently mostly in a boat show so as many people as possible can see him.
Put that alongside Hugo and the pattern is clear. Same kind of learned walk, same performance driven by animation, same human on the other side of the curtain, same stated aim of the character being to give hugs. The difference is who owns the character. Disney own Olaf. Theermann has said that if you own the rights to a character, his company is the one to call. Hugo is a case in point of what Dynamic Creatures could do for someone else.
The closest thing to a home companion with a human attached is 1X’s NEO which was announced in October 2025 at $20k or $499 a month. You can book in when a remote operator is going to take over the robot and what it is going to do. The buyer has to agree that the operator will be able to look into the house via the robot’s camera (1X say they can blur people and you can set no go zones). Basic stuff like opening doors and getting things will be expected to run autonomously at launch. 1X’s CEO has stated that they need footage of people’s homes in order to improve the product. NEO is a helper not a character. It is however the most obvious example of a company having written the human operator into the terms of sale.
That table answers the “first” question without any input from me. Aibo came out in 1999. Paro was in the Guinness Book of Records by 2002. Lovot was unveiled in 2018. Companion robots are an old category. “First” may well be intended in a more restricted sense (i.e. the first furry character to walk on two legs in a crowd). There’s nothing published to support that reading, and the company hasn’t offered it.
Two things are clear. Firstly, the newest entries here (Hugo, Olaf and NEO) all have a human in the loop. The difference is in who makes that explicit and where they do it: in a demo clip, in a terms of sale or on a remote in the hand of a cast member. Secondly, a lot of these robots are dependent on the company that made them. Aibo needs a cloud subscription. Moxie‘s robots ceased to function when the company that made them went out of business, as they relied on a cloud AI. A venue leasing Hugo would depend on one startup. I’m not saying Dynamic Creatures is going to go bust. I’m just saying a sensible operator should be asking themselves what happens to a character they’ve leased if the company that made it goes under.
Two meanings of “companion”
Hugo is built for theme parks, hotels, resorts, casinos and retail. That is a character at work, a mascot that moves among guests. A companion at home is a different job.
Theermann clearly has the second one in mind. He told the Globe he is certain every household will have a robot within ten years. In September he said the goal is to create companions, not only better robots. In a 2022 podcast he called a home companion his lifelong ambition, put it a decade or two away, and said he worried about lonely older people in big cities. Fortune’s reporter used the word companion for Hugo as well.
So the label is not the press’s invention, and it is not marketing fluff dropped in at the last minute. It is the company’s stated destination. But what the company offers today is managed pilots for venues, with leasing on the table. “Companion” describes where Dynamic Creatures wants to end up. I would call it a roadmap, not a product category.
What the research says about robot companions
The strongest case for Theermann’s vision comes from research, not from any tech demo. A 2025 meta-analysis in The Gerontologist pooled 19 studies with 1,083 older participants and found that social robots reduced loneliness, with a pooled effect of d = -0.59, which the authors call large. The reduction was bigger in institutions such as nursing homes than for people living at home. Robot type was not a significant factor.
The same authors flag the limits. Their own bias test suggests caution about generalising, there is little long-term follow-up, and most studies came from high-income countries. They also say robots should complement human relationships, not replace them.
That tells me two things. Firstly, robot companions are no joke. In care contexts with the right kind of user they appear to be of benefit. Secondly, nothing in that analysis tested anything like Hugo. The paper was published prior to Hugo’s existence and Hugo is a furry biped for the crowd which in the demo is operated by a person. A person operated Hugo is a puppet and puppets have a long and respectable pedigree. That is not the same claim as a companion.
The most forceful objection comes from outside the field of robotics. In a book published on 29th September Sherry Turkle (MIT) writes about chatbots and the way they can provide the illusion of empathy, which people then come to confuse with genuine empathy. Her book is about chatbots, not Hugo. The company would say that its Vision-Language-Expression model is designed to link perception and language to gesture, movement and emotional expression. To provide the appearance of feeling is precisely what Turkle is concerned about. The meta-analysis authors would point to the measured drop in loneliness. I think that both could be true. The benefit could be real and the cost could still be an open question.
Where this goes
Predictions are opinion, so here are mine, with the reasons attached.
- Human operators will stay for years. The company’s own roadmap runs from one operator per robot to one operator watching many, then to monitoring from off site. Disney still has a person typing Olaf’s lines. The number to watch is the ratio of operators to robots. It decides whether leasing a character pays for itself.
- Free-for-all hugging will come late. Disney and Dynamic Creatures both name hugs as a goal. Disney gives no date. Theermann has said humanoid safety is not fully figured out, and the Globe notes that humanoids rarely get close contact with the public because their actuators can hurt people. I expect the first deployments to use choreography, such as shows, photo poses and a boat in a lagoon, long before they use open arms.
- The home will come with a human attached. NEO have the template laid out: a subscription, a remote operator and a consent form as to who can see inside your house. If a companion robot gets into homes in the next few years I think the battle will be about who’s watching rather than about intelligence.
- The ‘companion’ label will head towards care. The best evidence for companion robots is in the area of elder care and Theermann’s own stated concern in 2022 was lonely older people. If the label is to mean anything it will be in that domain rather than in a casino. Nothing that has been announced suggests that is the plan. That’s the one prediction I’m least confident about.
- Customers will ask who owns the robot when the company goes away. Moxie is the case study. Leasing concentrates that risk on the operator.
I could be wrong on all five. Here is what would change my mind. An independent test showing Hugo working without a person steering it. A published price, battery life and sensor list. Named customers and real deployments. Safety evidence for robots that hug children. Any one of those would move me.
Verdict
Hugo is a well funded, well staffed, deliberately lovable prototype and I’d like to see more hardware startups be as open about the human behind the curtain. The cuteness is good design. The competition is real, with Disney obviously running the same show under a much more familiar name and 1X having put a human in their terms of sale. The research into robot companions is more positive than the sceptics will admit.
But “first human companion robot” is a future-facing slogan. Until Hugo works without a person steering it, in a real venue with numbers on the table, I’d still call it a nice character, not a companion. I’ll ask you again once the first pilots are running in early 2027.
