Dream Labs Unveils Dream API: The Brain for Everyday Robots, Powered by $90M Seed Funding

For years, the robotics industry has grappled with a fundamental challenge: while robotic hardware has become increasingly affordable and capable, the software intelligence to make these machines truly versatile has lagged. Factories worldwide operate millions of industrial robots, yet most are programmed for a single, repetitive task, with learned capabilities rarely transferable to different machines. Introducing a new task or hardware often means starting the programming process from scratch, a highly time-consuming and expensive endeavor. Dream Labs was founded in 2026 by former NVIDIA robotics researchers, including those who contributed to projects like GR00T, DreamZero, and World Action Models, specifically to tackle this problem.

The core innovation lies in Dream Labs' flagship model, Dream-1, a powerful robot foundation model. Unlike traditional programming methods, Dream-1 is trained on vast amounts of human egocentric data – videos of people performing everyday tasks. By observing human actions from a first-person perspective, the model learns to predict what will happen next and subsequently determines the appropriate robot actions. This video-driven learning approach allows Dream-1's capabilities to transfer across different hardware platforms, a significant departure from the siloed intelligence of previous robotic systems.

Joel Jang, co-founder and CEO of Dream Labs, eloquently articulated this vision: “The hardware is ready. What’s been missing is a brain any robot can use.” He draws an analogy to driving a rental car: you don't relearn how to drive, you simply find the controls, and your existing driving experience takes over. Dream-1 applies this same principle to robots, enabling a new machine to become useful in minutes rather than months of specialized programming.

The Dream API is designed to be the conduit through which developers can tap into this general-purpose robotic intelligence. It functions as the first publicly available API for a robot foundation model, allowing developers to connect supported robots, provide visual inputs (from a robot's camera feed), and issue plain-English instructions. Dream Labs' cloud-based models then process these inputs and generate the precise motor commands needed for the robot to execute the task. This abstraction layer dramatically simplifies robot programming, shifting the focus from low-level motion control to high-level task definition.

Currently, Dream-1 can control any robot equipped with grippers, offering 'zero-shot' capability, meaning it can operate previously unseen compatible hardware without additional task-specific training. Dream Labs is already collaborating with over 20 hardware partners, supporting a range of robots typically priced between $2,000 and $10,000. The initial capabilities focus on tabletop activities such as picking up and placing objects, and even folding laundry. The company has ambitious plans to extend support to wheeled mobile robots and systems featuring more advanced, dexterous robotic hands in the near future.

Beyond the core API, Dream Labs offers additional services to further empower developers and robot manufacturers:

• Fine-tuning as a Service: For specific hardware not yet natively supported or for new, highly specialized capabilities, developers can submit a small batch of demonstration data. Dream Labs then provides a tailored model within hours, allowing for rapid customization and deployment.
• Robot Playground: This innovative browser-based environment allows anyone to type commands to a robot located in Dream Labs' San Francisco office and observe its responses live and unedited. This offers a low-barrier entry point for experimentation and understanding the API's capabilities without the need for physical robot ownership.

The pricing structure for the API is designed for accessibility, with the flagship model costing $3 per hour and a less capable model available for 75 cents per hour. This model aims to make advanced robotic intelligence accessible for a wide range of applications and development budgets.

The $90 million in seed funding, led by prominent venture capital firms B Capital and Kindred Ventures, underscores the significant market potential and confidence in Dream Labs' approach. The funding round also saw participation from A.Capital, Unusual Ventures, SV Angel, Wing VC, Scribble VC, and individual investors like Adrian Aoun. A substantial portion of this investment will be allocated to expanding Dream Labs' computing infrastructure, which is crucial for scaling the range of tasks its AI models can perform, supporting additional robotic platforms, and accommodating a growing developer base.

The launch of Dream API is a significant step towards realizing a future where robots are not just tools, but intelligent, adaptable assistants integrated seamlessly into various aspects of life and industry. For developers, this means new opportunities to innovate in areas previously constrained by complex robotics programming. Imagine creating applications for elder care robots that understand nuanced requests, logistics robots that adapt to changing warehouse layouts on the fly, or even personal assistant robots that learn household preferences through simple conversation. The abstraction provided by Dream API allows developers to focus on application logic and user experience, rather than the intricate mechanics of robot control.

As Dream Labs continues to expand its model's capabilities and hardware compatibility, the impact on industries ranging from manufacturing and logistics to healthcare and consumer services could be transformative. The ability for a single AI brain to control diverse robotic hardware promises to accelerate innovation, reduce operational costs, and unlock unprecedented levels of automation and flexibility in the real world.