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.