General Intuition: Startup that teaches AI to navigate in time and space captures $300 million

General Intuition is about to become one of the hottest businesses in the artificial intelligence market. The New York startup negotiates a $300 million round that would raise its valuation to about $2 billion — quite a leap for a company that has not yet commercially launched its main product. The motive for enthusiasm? A foundation model designed to give AI agents the ability to navigate through space and time using real-world videos.

The company built a database with impressive 2 billion videos captured each year — approximately 230,000 hours of content per hour. This collection feeds a model that not only classifies images, but learns the basic physics of movement, the persistence of objects and the causality between actions. The promise is to turn AI agents from reactive automata into entities with world understanding.

General Intuition capture: what differentiates your AI model?

While most AI startups focus on isolated language or vision, General Intuition looks at a more fundamental level: understanding how the world behaves in time and space. Its foundation model is trained with continuous videos — not static frames — which allows predicting what happens before and after each scene. For Brazilian developers working with robotics, autonomous vehicles or internal navigation systems, this means that soon they will be able to use an API that already understands concepts such as "the car turned left and disappeared behind the building" without manually programming each object persistence rule. It's like giving common sense to a robot.

How does General Intuition train its innovative foundation model?

The differential is in the scale and the curatorship. General Intuition collects about 2 billion videos a year — a volume that covers from public security cameras to user-generated content on open platforms. Videos are preprocessed to remove inconsistent time metadata and normalize frame rate. Then the model learns a latent representation of 3D space over time: where objects are, how they move, which obstructs what. It's a step beyond what models like CLIP or DALL-E do with static images.

General Intuition capture: $2 billion valuation is justified?

The high valuation reflects the moment of the market. Large players like Google, Meta and OpenAI already invest billions in AI agents that can perform complex tasks in the digital and physical world. If General Intuition manages to package this model as a business-ready API, it can throw away home solutions that today require hundreds of computer vision engineers. For comparison purposes, startups focused exclusively on video foundation models, such as Runway, achieved similar valuations with still incipient revenues. The bet is that the market for AI agents will explode in the next three years and whoever has the best "visual operating system" will come out in front.

Frequently Asked Questions about General Intuition capture and technology

How much has General Intuition gotten up so far?

Startup does not disclose previous values, but sources indicate that the seed and series rounds A total of about $50 million, led by AI-focused funds.

When will the template be available for developers?

The company has not yet announced a public release date, but API access is expected to start in closed beta in the second half of 2026.

How does that fit into the race for AI agents?

General Intuition provides the layer of space-time perception that other language models do not have, acting as a complement — not a competitor — of architectures such as GPT and Claude. This is what Estonia did when creating official identity for AI agents, but at the level of visual perception.

Source: TechCrunch