Antioch says simulation might help robotic builders overcome testing bottlenecks. Supply: Antioch
Validating a robotic’s habits in the actual world sometimes means renting bodily house, manually staging environments, and resetting {hardware} between each run, in line with Antioch. It’s costly, gradual, and covers solely a fraction of the situations a system will face in manufacturing, mentioned the startup.
Antioch at present mentioned it has raised $8.5 million to maneuver growth and analysis of autonomous techniques out of the bodily world and into cloud-based simulation. The company‘s acknowledged objective is to eradicate the necessity for {hardware} and elaborate bodily testing.
“Robotics groups are spending weeks staging warehouses and investing hundreds of thousands into check services to validate their techniques,” mentioned Antioch co-founder Harry Mellsop, who beforehand labored on Tesla‘s Autopilot crew. “In the meantime, corporations like Tesla, Waymo, and Anduril spend tons of of hundreds of thousands a yr on simulation infrastructure to attenuate precisely that. We predict each autonomy crew ought to have entry to that stage of tooling.”
Antioch founders draw from safety, AI expertise
Three of Antioch’s 5 co-founders, Mellsop, Alex Langshur, and Michael Calvey, beforehand based Transpose, a safety and intelligence platform acquired by Chainalysis in 2023. Langshur mentioned the nationwide safety thread has carried over.
“The one economically viable path to reindustrialization runs by robotics and automation, and scalable testing is the rate-limiting step.” he mentioned.
New York-based Antioch was based in Could 2025. The founding crew additionally included Colton Swingle, beforehand at Google DeepMind, and Collin Schlager, beforehand at Meta Actuality Labs. The corporate now has a employees of eight.
“The industries LLMs [large language models] are disrupting — software program, skilled companies, data work — symbolize perhaps $8 trillion of the worldwide economic system,” asserted Mellsop. “Manufacturing, logistics, building, power, and agriculture symbolize $50 trillion. AI penetration in these bodily industries is mainly zero at present. The economic revolution that’s coming in bodily AI isn’t going to be a sequel to the LLM revolution; it’s going to dwarf it.”
Co-founders Harry Mellsop, Alex Langshur, Colton Swingle, and Collin Schlager. Supply: Antioch
Simulation offers reliable testing, says Antioch
Many corporations don’t wish to spend the assets wanted to maintain up with new instruments, fashions, and suppliers, even earlier than composing disparate instruments into usable workflows, famous Antioch.
The corporate claimed that the simulation ecosystem has matured to the purpose the place testing and validation at the moment are potential with minimal real-world inputs. Physics and rendering engines reminiscent of NVIDIA Isaac Sim and Unreal Engine present more and more highly effective primitives, and generative strategies like world fashions from World Labs are making it simpler to synthesize reasonable environments at scale, mentioned Mellsop.
By sharing physics engines and world fashions throughout clients, Antioch ensures that its baseline mannequin is calibrated to the real-world information they gather, he advised The Robotic Report. This helps all clients believe within the fashions with out forcing them to rebuild their infrastructure.
“We’re working with groups inside our largest buyer account to mannequin the notion techniques that may finally go into their robots,” Mellsop mentioned. “The very first thing we do with all of our clients is to copy the present units of exams they do in the actual world and replicate them 1:1 in Antioch simulation.”
“Now your engineers can run precisely these sorts of bodily exams that you simply had been working, however now that it’s on a pill within the palm of their arms, they’ll run it as many instances as they need, and it’s instantaneous,” he defined. “That slashes costs for these corporations and accelerates their engineering and growth efforts.”
Editor’s notice: On the 2026 Robotics Summit & Expo on Could 27 and 28 in Boston, there shall be periods on growing embodied and bodily AI. Registration is now open.
AI fashions might help consider edge instances
“As soon as we’ve accomplished replicated their exams and have constructed belief within the outcomes of the simulation, we’re then in a position to consider conditions that had been beforehand unattainable to really stage, whether or not that’s as a result of they had been too costly, too harmful, or simply impractical,” added Mellsop. “How does your robotic carry out when it’s within the snow or fog? Or perhaps the the warehouse is on fireplace, and also you wish to check that your security protocols are nonetheless working appropriately.”
“One of many key workflows that I’ve seen a few of our clients use is to ask our AI assistant to modify from utilizing one SLAM [simultaneous localization and mapping] algorithm to a extra up-to-date model,” he recalled. “In Antioch, your robotic is fully represented as code. Swapping one for the opposite in simulation is very easy for this agent to do.”
“However then it could possibly go and it could possibly run 1,000 totally different situations and report again what bought higher, what bought worse, what regressions occurred,” Mellsop mentioned. “As a result of you possibly can short-circuit the analysis loop, it’s finally the place I believe we in all probability all see robotics going, which is that the whole lot is turning into extra composable: the {hardware} aspect, the software program aspect.”
Antioch has efficiently developed simulation for sensor modalities together with lidar, radar, and cameras, mentioned Mellsop. Subsequent, the corporate plans to handle deformables, fluid dynamics, and dexterity modeling, which have been tough to do in simulation.
It is usually taking a look at upgrading embodied AI notion techniques for the whole lot from autonomous autos and warehouse planning to good safety and building website mapping.
Antioch says cloud simulation can deal with a a lot wider vary of variance than pricey bodily testing. Supply: Antioch
Funding follows FAANG clients
Antioch mentioned it’s already working with Fortune 500 expertise and logistics corporations, in addition to FAANG (Fb/Meta, Amazon, Apple, Netflix, and Google/Alphabet) engineering groups.
“We spent the primary six months heads-down constructing with a deep conviction about what the product needs to be,” Mellsop mentioned. “Then in November, we began onboarding these huge clients. We discovered quite a bit from these early engagements, and that was nice ratification for us that corporations throughout the board skilled these issues we had been engaged on.”
A* and Class Ventures led Antioch’s newest funding. MaC Enterprise Capital, Summary, Field Group, and Icehouse Ventures additionally participated, as did angel traders together with Shyam Sankar, chief expertise officer of Palantir, and Adrian Macneil, CEO of Foxglove. The company raised pre-seed funding in December 2025.
“Antioch sits on the intersection of AI and robotics and can unlock the science-fiction future we’ve typically dreamed of,” mentioned Bennett Siegel, co-founder and basic accomplice of A*.
“The platform removes the friction from physical-world testing, and allows a brand new technology of embodied AI startups to scale their innovations globally,” he added. “Firms are at present spending tons of of hundreds of thousands a yr to have a horse within the AI robotics race, however Antioch’s expertise reduces the hurdles significantly, giving extra companies an opportunity to innovate.”
Antioch is looking to scale with its newest funding. “Probably the most area of interest, hardest-to-find, and most essential roles for us is what we name a simulation engineer,” famous Mellsop. “These are people who’ve real-world expertise with the sensors that go onto robots, however they’ve additionally had some expertise of working with simulation engines, whether or not they’re robot-specific like Isaac or extra basic like physics and rendering engines. Discovering individuals who’ve bought these two ability units collectively is actually arduous.”
Simulation can now assist robotic builders overcome testing bottlenecks, say Antioch’s founders.
The publish Antioch raises funding to deliver ‘software program pace’ to robotic growth appeared first on The Robotic Report.
