Figure AI is a company operating in the fields of advanced artificial intelligence systems and robotic technologies, established in the United States in 2022 to develop general-purpose humanoid robots. The company seeks to address industrial labor shortages and perform human-like physical tasks through autonomous systems. The robots developed by Figure AI are designed to execute complex tasks using AI-powered motor control systems and visual perception modules. Figure AI is an integrated technology company that manages all processes internally, from design to production, software infrastructure to field applications.

Establishment and Purpose

Figure AI was founded in 2022, headquartered in San Jose, California, United States. The founder of the company is Brett Adcock, who had previously been involved in ventures such as Archer and Vettery. The company's mission is to enhance human capabilities through artificial intelligence (AI) technologies and to deploy general-purpose humanoid robots on a global scale.

Developed Robot Models

To date, Figure AI has developed three different humanoid robot models:

• Figure 01 was the initial prototype, developed for fundamental system trials.
• Figure 02 is a prototype featuring humanoid attributes such as visual perception, balance, and walking, and was tested for use in industrial environments.
• Figure 03 was designed as a production-focused model suitable for low-cost and high-volume manufacturing.

For the development of Figure 03, rapid manufacturing techniques such as injection molding, metal injection molding, and pressing were preferred over CNC (Computer Numerical Control) machining. Additionally, the number of parts was reduced, lowering production time and costs.

Helix: Vision-Language-Action Model

Helix is a general-purpose Vision-Language-Action (VLA) model developed by Figure AI. It enables robots to receive natural language commands, perceive environmental data, and act accordingly.

^{Helix (Figure AI)}

Helix consists of two main components: System 2 (S2), which analyzes visual and language inputs to set high-level task goals, and System 1 (S1), which provides low-latency motor control capable of generating 200 commands per second. Helix also allows multiple robots to perform synchronized collaborative tasks.

The model was trained using 500 hours of multi-robot and multi-operator data, ultimately achieving various tasks with a single set of neural network weights.

Walking Control via Reinforcement Learning

To achieve human-like walking in the Figure 02 model, Figure AI employed reinforcement learning (RL) methods. Robots were trained in simulated environments through trial and error, learning to walk under diverse scenarios.

Walking policies derived from simulations were transferred to physical robots without manual fine-tuning (zero-shot transfer). During training, the stylistic features of human walking (such as heel-strike initiation, toe-off completion, and coordinated arm-leg movement) were incorporated through reward functions.

BotQ: Manufacturing Facility

BotQ, introduced in 2025, is Figure AI’s high-volume humanoid robot production facility with an annual production capacity of 12,000 units. The production processes at BotQ are integrated with MES (Manufacturing Execution System), ERP (Enterprise Resource Planning), PLM (Product Lifecycle Management), and WMS (Warehouse Management System) software platforms.

BotQ also utilizes Figure robots in its production lines to perform tasks such as material handling and part assembly, thereby increasing automation within the manufacturing process.

Supply Chain and Production Strategy

Given the absence of an established supply chain for the humanoid robotics industry, Figure AI designs many of its components internally. Actuators, motors, battery packs, sensors, and electronic boards are developed in-house, while strategic external suppliers are engaged for components requiring specialized manufacturing processes. The supply chain has been structured to support the production of 100,000 robots or 3 million actuators within four years.

Collaboration with BMW

In 2024, Figure AI initiated a commercial collaboration with BMW. While announced publicly as a "commercial partnership," the actual field application was limited. The company reported that multiple robots were engaged in end-to-end tasks at BMW’s South Carolina plant, though BMW officials clarified that only one robot was tested on a specific assembly task, initially conducted outside standard production hours.

^{BMW (Figure AI)}

Funding and Industry Position

Since its founding, Figure AI has secured over USD 700 million in funding from major investors, including Microsoft, Nvidia, Intel Capital, and Jeff Bezos’ investment firm. As of 2025, the company’s market valuation is projected to reach USD 40 billion. Figure AI operates within the humanoid robotics sector alongside companies such as Tesla (Optimus), Agility Robotics, 1X, Apptronik, and Unitree.

Corporate Culture and Values

According to its statements, Figure AI’s corporate culture is based on five core principles: technical courage, product priority, long-term optimism, future impact focus, and a championship mindset. The company maintains a distant stance toward social activism and prioritizes dedicating its resources solely to its core mission. Figure AI does not have a formal structural partnership or investment relationship with OpenAI. However, in the development of models like Helix, open-source and open-weight visual-language models have been utilized.