Fieldwork runs autonomous AI that captures, builds, and hosts navigable 3D experiences — from a text prompt to a live scene, without a human in the loop.
Spatial computing has great capture tools. It has great renderers. It doesn't have autonomous agents that run the whole pipeline.
Fieldwork fills that gap. One prompt, one agent, one live scene. Every time.
Agent ingests images, video, or LiDAR scans. Runs photogrammetry or monocular depth estimation. Outputs a raw 3D point cloud or Gaussian splat.
Scene reconstruction, semantic labeling, lighting estimation. Agent assembles a navigable world from raw capture data — no human in the loop.
Atmospheric audio, ambient motion, lighting. Agent adds the elements that turn a static scan into a living space.
World packaged as a browser-native scene. Agent deploys to edge, generates the public URL, and reports back. Done.
A prompt like "an abandoned library at dusk, dust particles, distant rain sounds" tells the agent what to build.
Fieldwork runs the full spatial pipeline: capture, reconstruction, synthesis, deployment. You watch from the dashboard.
The agent deploys the result, sends you the URL, and logs what it built. No infrastructure knowledge required.
Spatial experiences shouldn't require a render farm, a team of artists, and three months of iteration.
They should be as easy as hitting send.