SPATIAL PIRegister interestTechnology
Overlaying content on a camera feed is a solved problem. Knowing what is in front of the camera, where every component sits in 3D, what state it is in, and whether that state is correct is not. That is the technology Spatial PI is building.
Early-stage product · Capabilities and compatibility are still being validated
Direction
Each one exists to make guidance trustworthy: the system should understand the object well enough to guide, correct, and verify the work on it.
Recognizing machines, objects, and workspaces from the camera, and reading their current state. This is the foundation for checking work rather than just displaying instructions.
For the workerThe system knows what it is looking at before it says anything.
Keeping instructions locked to the real component as the person and the device move around the asset, so guidance stays where the work is.
For the workerThe instruction stays on the part, not floating somewhere near it.
Building a persistent understanding of the asset and its surroundings: what belongs where, how it fits together, and what has changed since the last time it was seen.
For the workerThe system remembers the machine, not just the frame in front of the camera.
Connecting perception to procedure: deciding whether a step is complete, what is wrong, and what the corrected state should look like.
For the workerThe system can tell right from wrong — and show the difference in place.
Research and engineering direction. These describe what Spatial PI is building toward, not benchmarked or shipped capabilities.
Why proprietary
General-purpose AR frameworks are good at placing content in a camera view. Guided, verified work needs more: perception that understands industrial objects, tracking that holds up on a shop floor, and reasoning that can tell a correct state from an incorrect one.
Owning that stack — rather than assembling another AR application on top of someone else’s — is what lets Spatial PI move from showing instructions to understanding work. It is a longer road, and it is the one we have chosen.
From assistant to platform
The order matters. Physical intelligence earns trust by helping people do real work well before it is asked to do more.
An AI coworker for people doing physical work: it sees, understands, guides, and verifies, while the person performs the task and stays accountable for it.
Move SOPs, paper manuals, and 2D instructions off the page and onto the object, with verification and documentation built into the work itself.
Connect AI with the real world across an organization's equipment, procedures, and people — reducing the gap between the digital and physical worlds.
Register interest
If your team has equipment, procedures, and a workflow worth validating against, we would like to hear about it.