Munich’s Kyle Walks Unblurred, Product Still a Year Out
A silver, square-framed biped with its guts hanging out is not the usual launch render. On 14 September, Nikolai Ensslen posted locomotion footage of the mule his stealth team calls Kyle, unblurred for the first time. Humanoids Daily picked up the clip the same day.
Ensslen co-founded motion-control firm Synapticon. He is explicit that Kyle is an internal development platform, not the first-generation product.
The mule, not Generation 1
“I have watched a lot of humanoids walk over the last years,” Ensslen wrote. “In my honest assessment, this is one of the better gaits out there — and it is still our very early stage.” He credited Martin Riedmiller and the team for the walking.
The box is the point. Ensslen said the rectangular frame is what you get from rapid prototyping and a short design-and-build cycle. Earlier clips hid that body because it was a development mule. Generation 1 had progressed far enough that the team was willing to show the machine they learned on.
On-screen text in the footage reads “Locomotion testing” and “Alpha internal development platform.” A gantry and catch cable stay in frame. There is no published height, payload, DoF count, or task list. Early 2027 is the stated unveiling target for Generation 1, not a ship date.
Who is in the room
Synapticon’s 19 December 2025 leadership note said Ensslen would leave operational duties to build a new Physical AI company, remaining a shareholder. Simon Fischinger, previously COO, became CEO on 1 January 2026. The company framed the split as a way to keep Synapticon a neutral motion-control supplier.
Humanoids Daily identifies other names around the footage. Oliver Groth’s X profile lists him as a former Google DeepMind research scientist now at a stealth startup; his research bio describes robotic exploration and control in Riedmiller’s DeepMind team after a PhD at Oxford. Groth, in a reply covered by Humanoids Daily, named the prototype Kyle and said the team built it from scratch in a few months with ownership of hardware and the AI stack.
Boris Belousov also shared the walking clip, crediting hardware, software, and integration, and calling reinforcement learning the finishing ingredient. His current profile, per Humanoids Daily, is AI engineer at a Munich stealth startup; his TU Darmstadt / DFKI bio documents earlier Systems AI for Robot Learning work.
Riedmiller’s public bio, linked from that coverage, describes leading the Controls team at Google DeepMind. Ensslen’s post does not state his formal role at the new venture.
The company is recruiting. Ensslen asked people who have built humanoids or shipped hardware at scale to get in touch.
A Human’s Take
I’m so here for a team that shows the ugly mule instead of a matte-black teaser. A gait you can watch, a gantry you can see, and a 2027 product they are not pretending to sell today is a cleaner pitch than most stealth decks. The missing receipts are still the boring ones: how much of that walk is the cable, how repeatable the gait is off the gantry, and whether Generation 1 keeps any of Kyle’s honesty once the covers go on.