Generalist AI
Generalist AI is a privately-held embodied AI company developing foundation models (GEN-0, GEN-1) that enable general-purpose robots to perform dexterous physical tasks. Founded in 2024 by DeepMind and Boston Dynamics veterans, it licenses these models to robotics OEMs and enterprise customers.
- Company typePrivate
- Founded2024
- HeadquartersSan Mateo, United States
- Headcount11–50
- GTM typeB2B
- OfferingSoftware
What Generalist AI does
Generalist AI is a privately-held embodied AI company developing foundation models that enable general-purpose robots to perform dexterous physical tasks in real-world environments. Founded in 2024 by veterans of Google DeepMind, OpenAI, and Boston Dynamics—including co-founder and CEO Pete Florence, who led work on PaLM-E, RT-2, and Gemini Robotics—the company is headquartered in San Mateo, California with engineering offices in San Francisco and Boston. Its product portfolio centers on two generations of embodied foundation models: GEN-0 (released November 2025), which established the first empirical demonstration of scaling laws in robotics, and GEN-1 (released April 2026), which the company claims achieves 99% success rates on dexterous manipulation tasks (kitting auto parts, folding boxes, packing phones, servicing robot vacuums, folding t-shirts) at approximately 3x the speed of prior state-of-the-art, with only ~1 hour of task-specific robot data required per new task.
The technical approach is distinguished by training models from scratch on proprietary physical interaction data, rather than fine-tuning vision-language-action (VLA) models or relying on world model paradigms. The company operates a Physical Data Engine that has accumulated over 500,000 hours of real-world manipulation data—growing at more than 10,000 hours per week—collected through custom handheld ergonomic devices that preserve natural human reflexes and closed-loop sensorimotor behavior. Its novel Harmonic Reasoning architecture enables models to think and act simultaneously through asynchronous continuous-time streams of sensing and acting tokens, supporting scaling to 10B+ parameters without System1-System2 inference-time guidance.
Generalist AI generates revenue by licensing its embodied foundation models to robotics OEMs and industrial automation companies for integration into commercial robot products (humanoid robots, warehouse machines, industrial robotic arms, autonomous platforms), supplemented by enterprise partnerships and custom deployment engagements accessed via direct outreach. Distribution is amplified through strategic ecosystem partnerships—NVIDIA (Cosmos Coalition founding membership and financial investor), Universal Robots (GTC 2026 live demo on UR7e/MiR platform), Teradyne Robotics (Automate 2026 ecosystem showcase), and Scale AI (data infrastructure). The company has raised over $500 million total, most recently a $400 million Series A in June 2026 at a $2 billion post-money valuation led by Radical Ventures, with NVIDIA, Bezos Expeditions, and angels Fei-Fei Li and Eric Yuan among participants.
Generalist AI firmographics
Firmographics- Name
- Generalist AI
- Legal name
- Generalist AI, Inc.
- Website
- https://generalistai.com
- Company type
- Private
- Founded year
- 2024
- Operating status
- Operating
- Headcount range
- 11–50 employees
- Short description
- Generalist AI is a privately-held embodied AI company developing foundation models (GEN-0, GEN-1) that enable general-purpose robots to perform dexterous physical tasks. Founded in 2024 by DeepMind and Boston Dynamics veterans, it licenses these models to robotics OEMs and enterprise customers.
- Ownership category
- akta.pro rank
Generalist AI industry classification
Industry- Product category
- Embodied AI Foundation Models
- NAICS
- Navigational, Measuring, Electromedical, and Control Instruments Manufacturing (33451)
- SIC
- Services-Prepackaged Software (7372), Services-Computer Integrated Systems Design (7373), Services-Computer Programming Services (7371)
- akta.pro industry
- Foundation Model Developers (LLM/Multimodal Model Labs)Hardware, Technology & Telecom/Artificial Intelligence & Machine Learning/Foundation Models & Generative AI (LLMs, multimodal)Humanoid & General-Purpose RobotsHardware, Technology & Telecom/Artificial Intelligence & Machine Learning/Robotics & Autonomous Systems
Keywords
Where Generalist AI is headquartered
LocationHeadquarters
- HQ city
- San Mateo
- HQ country
- United States
- HQ region
- North America
Offices3 records
Markets served
Generalist AI business model
Business model- GTM type
- B2B
- Offering type
- Software
- Cost components
- Personnel, Technology or R&D, Infrastructure, Operations, Marketing or Sales, Supply Chain
Revenue model
- Foundation Model Licensing to Robotics Manufacturers: Generalist AI licenses its embodied foundation models (GEN-0, GEN-1) to robotics companies for integration into commercial robot products. Early-access partners gain model access starting April 2026. The company generates revenue by enabling robot OEMs and industrial automation companies to embed physical AI capabilities into their hardware platforms, with partnerships established across humanoid robots, warehouse machines, industrial robotic arms, and autonomous platforms.
- Enterprise Partnership and Custom Deployment: Generalist offers enterprise partnerships for companies with specific use cases for robotic automation. The company accepts inquiries from businesses with particular automation needs (via partnerships@generalistai.com) and likely structures custom commercial agreements for access to models, fine-tuning on customer-specific robot hardware, and deployment support.
Go-to-market motion1 record
Distribution channels3 records
Marketing channels7 records
Generalist AI product offering
Product offeringCore offering
Generalist AI builds embodied foundation models—GEN-0 and GEN-1—that are trained from scratch on physical interaction data to enable robots to perform general-purpose dexterous tasks in the real world. The company licenses these models to enterprise robotics manufacturers and automation companies, and operates a proprietary Physical Data Engine that has collected over 500,000 hours of real-world manipulation data via custom handheld ergonomic devices. GEN-1 achieves 99% task success across tasks such as kitting, folding, packing, and robot-vacuum servicing, completing them roughly 3x faster than prior state-of-the-art.
Product overview
Generalist AI is a frontier AI research and product company building general intelligence for the physical world. Its product portfolio centers on embodied foundation models—GEN-0 and GEN-1—trained from scratch on the company's proprietary physical data engine containing over 500,000 hours of real-world manipulation data. These models enable general-purpose robots to perform dexterous physical tasks including packing, folding, assembly, and servicing across manufacturing, logistics, healthcare, agriculture, hospitality, and space applications. The company also offers early-access partnerships for commercial deployment of its models. The founding team comes from Google DeepMind and Boston Dynamics, bringing expertise from prior work on PaLM-E, RT-2, and Gemini Robotics.
Differentiator
Problem solved
Functional benefit
Brands
- GEN-0: First generation embodied foundation model demonstrating scaling laws in robotics, released November 2025.
- GEN-1
Products and services
- GEN-1 Embodied Foundation Model GEN-1 is Generalist AI's flagship large multimodal embodied foundation model that emits real-time robot actions for autonomous dexterous manipulation. It achieves 99% average success rates on multiple physical tasks, completes them roughly 3x faster than prior state-of-the-art, and exhibits improvisational intelligence for recovering from unexpected scenarios. The model is trained from scratch on over 500,000 hours of real-world physical interaction data and requires approximately 1 hour of task-specific robot data to adapt to new tasks, enabling commercial deployment of general-purpose robots for enterprise customers.
- GEN-0 Embodied Foundation Model GEN-0 is Generalist AI's first embodied foundation model, designed to establish that scaling laws exist in robotics analogous to language models. It introduced the Harmonic Reasoning architecture enabling simultaneous thinking and acting via asynchronous continuous-time streams of sensing and acting tokens, was pretrained on 270,000+ hours of real-world manipulation data, and demonstrated fast adaptation to new robots and environments across 6DoF, 7DoF, and 16+DoF semi-humanoid embodiments.
- Physical Data Engine The Physical Data Engine is Generalist AI's proprietary infrastructure for acquiring high-fidelity physical interaction data. It consists of custom handheld ergonomic devices that allow operators to perform tasks naturally while preserving reflexes, micro-corrections, and real-time recovery—capturing the closed-loop sensorimotor loop intact—and the supporting data processing and storage systems. The Engine has collected over 500,000 hours of real-world manipulation data and underpins training of the company's GEN-0 and GEN-1 foundation models.
Quantifiable outcome
- 99% average task success rates across diverse dexterous tasks (vs. 64% for prior models on identical tasks)
- +4 more outcomes
Companies that use Generalist AI
Customer profileNamed customers1 record
Segments4 records
Ideal customer profiles1 record
Generalist AI technology and API
TechnologyTechnology focussed Yes
API detail
- Has API
- No
- API docs
- API detail
Core technology
AI maturity
App detail
Integration5 records
AI capability9 records
Feature6 records
Generalist AI partnerships and signals
Strategic signalPartnerships
Four partnerships are on record, tiered core and strategic.
- NVIDIA (NVentures and Cosmos Coalition)coreGeneralist is a founding member of the NVIDIA Cosmos Coalition alongside Agile Robots, Black Forest Labs, LTX, Runway, and Skild AI, advancing open world models for physical AI. NVIDIA's NVentures also invested in Generalist's $400M funding round, making NVIDIA both a technology partner and a financial backer. NVIDIA provides the Cosmos 3 platform and ecosystem infrastructure within which Generalist's models operate.
- Universal RobotsstrategicUniversal Robots invited Generalist to conduct its first public live demo of GEN-0 at NVIDIA GTC 2026 in their booth, on their new mobile manipulation platform (UR7e arms on a MiR base). Generalist was able to set up the demo on a completely new robot platform within days, demonstrating the model's cross-embodiment generalization capabilities. Universal Robots is the world's #1 cobot manufacturer by volume.
- Teradyne RoboticsstrategicGeneralist AI participated as an ecosystem partner in Teradyne Robotics' showcase at Automate 2026 in Chicago, part of a live demonstration ecosystem of 11 partners including Scale AI, NVIDIA, Cambrian, AICA, beRobox, ROEQ, Maple Advanced Robotics, and others. The demonstrations emphasized real-world deployability of physical AI across AI data centers, electronics manufacturing, and unstructured industrial environments.
- Scale AIstrategicScale AI and Generalist AI are both ecosystem partners in the robotics space. Scale AI is expanding its Data Engine for Physical AI, having completed over 100,000 production hours at its San Francisco prototyping laboratory, and has highlighted Generalist AI as one of its leading robotics firm partnerships. Scale AI's data infrastructure and annotation services complement Generalist's foundation model development.
Scale indicators9 records
Recent moves6 records
Expansion highlights6 records
Generalist AI competitors and assessment
Company assessmentDirect peers
- Physical Intelligence: Physical Intelligence builds embodied foundation models for general-purpose robots, pursuing nearly identical technical thesis (large models trained on physical interaction data, licensed to robot OEMs and used for dexterous manipulation). It is the most direct comparable competitor to Generalist AI's GEN-0/GEN-1 stack.
- Skild AI: Skild AI develops embodied foundation models for robotics with a focus on general-purpose physical intelligence across diverse robot embodiments. Co-founder of the NVIDIA Cosmos Coalition alongside Generalist, targeting the same OEM and enterprise robotics customer base with a comparable foundation-model product.
- Covariant: Covariant built AI foundation models for warehouse robotics (picking, sorting, kitting) and was acquired by Amazon in 2024. It is the closest historical analog to Generalist's industrial-manipulation thesis and continues to operate as Amazon's in-house robotics AI effort, competing for the same enterprise robotics OEM relationships.
Broad incumbents
- NVIDIA: NVIDIA is both an investor (NVentures) and a competitor through its GR00T humanoid foundation model and Cosmos 3 physical AI platform. It controls the underlying compute and orchestration layer (Isaac, Cosmos) on which Generalist's models run, and competes directly in providing embodied AI to robot manufacturers.
- Google DeepMind: DeepMind is the origin of the RT-2, PaLM-E, and Gemini Robotics work that Generalist's founders helped pioneer, and now ships its own Gemini Robotics foundation models for dexterous manipulation. As a frontier lab with vastly larger compute and data resources, it competes with Generalist across the same embodied foundation model category.
- Tesla (Optimus): Tesla is developing the Optimus humanoid robot using in-house foundation models trained on data from its autonomous driving and manufacturing operations. It competes for the same humanoid-robotics customer and OEM mindshare while bringing much larger capital, manufacturing, and data collection infrastructure than Generalist.
- Boston Dynamics: Boston Dynamics is the originator of Atlas, Spot, and Stretch robots and a former employer of Generalist's founders. Through its Hyundai ownership and Reinforcement Learning Research Kit, it develops its own manipulation AI in-house, making it both a potential customer and a competitor for embodied AI licensing in industrial and logistics settings.
Emerging players
- Figure AI: Figure AI builds humanoid robots for industrial and commercial use cases and is developing its own Helix foundation model for robot control. It is an emerging player in the same embodied AI category that overlaps with Generalist's humanoid-robot deployment thesis, particularly for warehouse and manufacturing customers.
- 1X Technologies: 1X Technologies develops the Neo humanoid robot with an in-house AI stack focused on safe, dexterous operation in home and light-commercial environments. It is an emerging competitor in general-purpose humanoid robotics with partial overlap to Generalist's consumer/domestic robotics use cases.
- Agility Robotics: Agility Robotics builds the Digit humanoid robot for logistics and warehouse applications, with an in-house autonomy stack that competes for the same industrial manipulation workloads Generalist targets. It is an emerging player with deep expertise in bipedal locomotion and a growing footprint in commercial deployments.
Market position
Strengths5 records
Weaknesses5 records
Competitive moat6 records
Key risks6 records
Key highlights7 records
Customer concentration
Generalist AI social profiles
Digital presenceGeneralist AI financial estimates
Financial estimateRevenue estimate
Valuation estimate
Generalist AI leadership team
Management profileNumber of profiles
Profiles3 records
Generalist AI funding detail
Funding detailFunding overview
Funding rounds5 records
Investors14 records
Funding detail is available on the Subscription and Enterprise plan.Contact sales →
Generalist AI M&A and investment
M&A and investmentM&A
Investments
M&A and investment is available on the Subscription and Enterprise plan.Contact sales →
Frequently asked questions about Generalist AI
What does Generalist AI do?
Generalist AI builds embodied foundation models—GEN-0 and GEN-1—that are trained from scratch on physical interaction data to enable robots to perform general-purpose dexterous tasks in the real world. The company licenses these models to enterprise robotics manufacturers and automation companies, and operates a proprietary Physical Data Engine that has collected over 500,000 hours of real-world manipulation data via custom handheld ergonomic devices. GEN-1 achieves 99% task success across tasks such as kitting, folding, packing, and robot-vacuum servicing, completing them roughly 3x faster than prior state-of-the-art.
Is Generalist AI a public or private company?
Generalist AI is a private company. It is classified as venture growth investor backed and is currently operating.
When was Generalist AI founded?
Generalist AI was founded in 2024. It employs 11 to 50 people.
Where is Generalist AI based?
Generalist AI is headquartered in San Mateo, United States, in the North America region.
How does Generalist AI make money?
Two revenue lines are on record. Foundation Model Licensing to Robotics Manufacturers are the primary driver. The others are enterprise Partnership and Custom Deployment.
Who are Generalist AI's main competitors?
Direct peers on record are Physical Intelligence, Skild AI and Covariant. Broad incumbents are NVIDIA, Google DeepMind, Tesla (Optimus) and Boston Dynamics. Emerging players are Figure AI, 1X Technologies and Agility Robotics.
Does Generalist AI have an API?
No public API is recorded for Generalist AI.
What industry is Generalist AI in?
Generalist AI's product category is Embodied AI Foundation Models. Its primary akta.pro industry code is HDAAACAA, Foundation Model Developers (LLM/Multimodal Model Labs), with a secondary code of HDAAAIAJ, Humanoid & General-Purpose Robots. Its NAICS code is 33451 and its SIC code is 7372.