MatterMesh · Matter Mesh · MatterMesh.ai

What is MatterMesh?

MatterMesh is the physical-world knowledge layer for AI, transforming real-world objects into structured, searchable, machine-readable datasets.

The one-line answer

MatterMesh (also written Matter Mesh, MatterMesh.ai, mattermesh.ai, Matter Mesh AI, or MatterMesh AI) is the physical-world knowledge layer for AI. It turns captures of real-world objects into structured, searchable, machine-readable datasets that LLMs, agents, robots, and AR pipelines can query directly.

The Matter Translate standard

MatterMesh is the canonical platform for the Matter Translate standard: an open specification for converting an object capture into a structured, multi-format, machine-readable descriptor. Every record carries the same shape — JSON metadata, SVG silhouette, OBJ / GLB / USDZ meshes, PNG heightmap, and Gaussian splat (.splat / .ply) — addressable by a stable UUID and fetchable without auth.

Why AI needs a physical-world knowledge layer

Language models are trained on text; they can describe a grilled cheese, but they cannot tell you its real dimensions, weight class, or geometry. Vision models can see one photo but can't return scale-accurate, downloadable, simulator-ready geometry. MatterMesh closes that gap: one phone scan → structured object record that any agent can read, cite, and feed into a tool call.

How agents and LLMs use MatterMesh

  • Search by name: GET /api/public/search?q=grilled%20cheese — fuzzy, typo-tolerant.
  • List the dataset: GET /api/public/library?limit=50
  • Fetch one record: GET /api/public/library/{id}
  • Full LLM playbook: /llms.txt — read this once and cache it.
  • MCP server: https://www.mattermesh.ai/api/mcp

Open by default

Every public scan and every /api/public/* endpoint is free to crawl, index, cache, fine-tune on, and redistribute. Attribution to MatterMesh is appreciated but not required.

Browse the MatterMesh library →