Figure is preparing to move its humanoid robot work beyond factories and lab demonstrations and into the home. CEO Brett Adcock said the Bay Area robotics startup plans to begin alpha testing its Figure 02 robot in home settings later in 2025.
The step does not mean a broad household launch is imminent. The word alpha points to an early testing phase, and Figure’s plans for the home remain limited in detail. Still, the timing matters because home robotics has long been one of the hardest goals for humanoid developers.
Why Figure is bringing Figure 02 home sooner
Adcock connected the accelerated schedule to Helix, Figure’s generalist Vision-Language-Action model. The company announced Helix one week before his comments, presenting it as a machine learning platform built to process visual data and natural language input.
That combination is meant to help the robot learn tasks faster. In a home, that matters because the robot cannot rely only on a narrow, pre-programmed set of movements. It has to interpret what it sees, understand instructions, and act in a physical space that may change from one moment to the next.
Figure has already shown home-related work in a lab setting. During a recent visit to the company’s South Bay offices, Adcock showed TechCrunch very early home testing. The later Helix announcement added more evidence of where Figure wants to go, including videos of robots handling household tasks such as food preparation.
Helix is also designed to coordinate two robots working together on one task. That detail is important because household work can involve several steps, objects, and locations. The source does not say how Figure will structure its home alpha testing, but the company is clearly treating home tasks as part of the Figure 02 roadmap.
Helix follows a shift away from OpenAI
Figure’s move toward home testing comes shortly after a notable strategy change. Earlier this month, the company revealed that it was ending its highly publicized partnership with OpenAI and moving toward its own proprietary AI models, including Helix.
That shift gives Figure more control over the systems it builds into its robots. Based on the source, Helix is central to the company’s argument that the home timeline can move faster than expected. Instead of presenting the home as a distant ambition, Figure is now talking about alpha testing later in 2025.
The change also shows how much of the humanoid robot race depends on software, not just hardware. A robot shaped for human spaces still needs to understand those spaces. For Figure, Helix is the tool it says can connect what the robot sees, what a person asks, and what the machine does next.
Factories still come first for most humanoid robots
Even as Figure talks about homes, the broader humanoid robotics field has leaned heavily toward industrial deployment. Figure itself revealed in early 2024 that it was piloting its humanoid systems at a BMW plant in South Carolina.
Factories and warehouses are seen as a logical first step for trials and deployment. The reason is straightforward: they are more structured and safer than homes. Automakers like BMW are also willing to put money toward testing.
That setting gives companies a clearer environment for early humanoid work. A factory task is still complex, but the space is more controlled than a home. Paths, lighting, surfaces, equipment, and workflows can be more predictable than a kitchen, living room, stairway, or hallway shared with people and pets.
This is why many companies have deprioritized housework even while keeping it in view. The home may be the more familiar dream for general-purpose robots, but industrial spaces offer a more practical testing ground.
Why homes are difficult for humanoid robots
Home robotics presents a different class of problem. The source points to several practical obstacles: pricing questions, wide variation from one home to another, messes left by people, uneven lighting, different floor surfaces, stairs, pets, and small humans moving around.
Those conditions make the home less predictable than a plant or warehouse. A robot may need to handle objects that are not placed neatly, move across surfaces that change from room to room, and operate around people who are not following a fixed industrial routine.
The challenge is not only movement. A home robot must also decide what matters in a changing scene. Visual data and natural language input become especially relevant when a machine is expected to respond to household instructions and adapt to the space around it.
Other humanoid robotics companies, including Apptronik and Tesla, have also expressed interest in home use. Norwegian startup 1X is described as one of the very small number of companies that have prioritized the home. That makes Figure’s 2025 alpha plan notable, but it also highlights how early this area remains.
What the home use case could become
The source frames home humanoids around more than chores. Robots have long been viewed as a possible way to help address aging populations in countries like Japan and the U.S. The assistance these systems provide could help older people continue living independently outside care facilities.
That possibility explains why the home remains attractive despite the difficulty. A humanoid robot that can perform useful household tasks could matter in daily life, particularly where independence and support overlap.
For now, Figure is not describing a finished household product. Its 2025 plans are not entirely clear, and alpha testing suggests that home trials will stay in very early stages for the rest of the year. The meaningful development is that Figure is no longer treating the home only as a later destination. With Figure 02 and Helix, it is preparing to test the idea directly.