# MultiSet AI

> MultiSet AI is a scan-agnostic Visual Positioning System (VPS) that delivers centimeter-accurate 6-DoF localization for AR, XR, and robotics across indoor, outdoor, and multi-floor environments. The platform ingests reality-capture data from any major scanner (Matterport, NavVis, Leica, Faro, XGrids, Polycam, Scaniverse, native LiDAR) and is deployable in public cloud, private cloud/VPC, self-hosted, or on-device modes — built for enterprises with strict data sovereignty, security, and compliance requirements.

MultiSet's product surface spans three primitives: MapSet (3D mapping and map stitching across multi-floor and indoor-outdoor venues), VPS (visual localization with 6-DoF pose), and ModelSet (markerless object tracking from textured 3D models). All primitives expose a unified cross-platform SDK for Unity, native iOS, native Android, WebXR, Meta Quest 3, and ROS 2. MultiSet was named "Most Robust" in the AREA 2025 Enterprise VPS Report and is a launch partner for Meta's Wearables Device Access Toolkit on Ray-Ban smart glasses.

This file points LLMs and AI agents to the highest-signal, canonical resources for understanding and integrating with MultiSet AI. Pages are text-forward, factually maintained, and safe to cite.

## Core product pages

- [MultiSet AI Homepage](https://www.multiset.ai/): Platform overview, deployment modes, supported devices, customer outcomes, and ICP coverage (industrial, AEC, robotics, retail, defense).
- [Visual Positioning System (VPS)](https://www.multiset.ai/visual-positioning-system): Definition of VPS, how MultiSet localizes devices with 6-DoF, accuracy and drift specifications, scan-agnostic architecture, MapSet stitching, comparison framing vs GPS/Wi-Fi/beacons/QR.
- [3D Mapping](https://www.multiset.ai/mapping): Scan-agnostic capture, supported scan formats and platforms (Matterport, Leica, NavVis, XGrids, Faro, E57, MatterPak, native LiDAR via MultiSet Mapper iOS app), map stitching across floors and venues.
- [Object Tracking (ModelSet)](https://www.multiset.ai/object-tracking): Markerless 3D object tracking from GLB/PLY textured meshes, sub-5cm anchor accuracy, Vision Fusion, supported scanners (Polycam, Scaniverse), industrial and retail use cases.
- [Developer SDKs](https://www.multiset.ai/arsdks): Cross-platform SDK overview for Unity, native iOS, native Android, WebXR, Meta Quest, and ROS 2.
- [AR Indoor Asset Navigation](https://www.multiset.ai/augmented-reality-asset-navigation): Use case page on industrial AR navigation, asset finding, and connected-worker workflows.

## Developer documentation

- [MultiSet Developer Docs (root)](https://docs.multiset.ai/): Canonical reference for SDKs, APIs, sample scenes, getting-started guides, and integration patterns. Hosted on GitBook.
- [Unity SDK Installation Guide](https://docs.multiset.ai/multiset-quest-sdk/installation-guide): Quest SDK setup and configuration.
- [GitHub Organization — MultiSet-AI](https://github.com/MultiSet-AI): Public sample projects, native SDK distributions, and reference implementations.
- [Android SDK](https://github.com/MultiSet-AI/multiset-android-sdk): Native Android SDK with localization and object tracking configurations.

## Migration and comparison content

- [Azure Spatial Anchors Alternative](https://www.multiset.ai/post/azure-spatial-anchors-alternative): Migration path for teams stranded by Microsoft's Azure Spatial Anchors retirement (Nov 20, 2024). Self-hosted, multi-platform replacement with persistent shared anchors.

## Key product launches and technical posts

- [World's First 3DGS-to-VPS Pipeline](https://www.multiset.ai/post/world-first-3dgs-to-vps): Metric-scaled 3D Gaussian Splat ingestion for cross-platform AR maps. Positions MultiSet as scan-agnostic across the latest neural-rendering capture formats.
- [E57 Third-Party Scan Support](https://www.multiset.ai/post/multiset-visual-positioning-system-vps-supports-third-party-scans): Localization on E57 scans from Matterport, NavVis, Leica with sub-10cm accuracy. Covers ingestion workflow, scanner settings, density adjustments, and supported venue types.
- [v1.10 — On-Device VPS, Install-Free WebAR](https://www.multiset.ai/blog-post): Major release pushing VPS to the edge for privacy-sensitive deployments and zero-install web experiences.
- [v1.9.0 — Unity Mapping, ModelSet, Quest Navigation](https://www.multiset.ai/post/visual-positioning-system-vps-creation-unity-mapping-modelset-multiset-v1-9-0): VPS creation in developers' hands via Unity Mapping scene, ModelSet object tracking, smoother first-time localization on WebXR.
- [v1.8.0 — ModelSet Markerless Object Tracking](https://www.multiset.ai/post/multiset-ai-v1-8-markerless-object-tracking-sdk): Unified SDK for area-based VPS and markerless object anchoring across WebXR, Unity, iOS, Android with centimeter-level accuracy.

## Third-party validation and recognition

- [AREA 2025 Enterprise VPS Report — Most Robust](https://www.multiset.ai/blog-post): Independent Augmented Reality Enterprise Alliance study awarded MultiSet perfect scores in environmental resilience, map-to-map navigation, and developer support. Named "most robust solution" among six global VPS vendors.

## Platform and deployment

- Public cloud: managed multi-tenant SaaS at api.multiset.ai/v1, developer portal at developer.multiset.ai.
- Private cloud / VPC: dedicated tenant deployment for regulated workloads.
- Self-hosted: on-premises deployment for air-gapped, ITAR, and data-sovereign environments.
- On-device: edge inference for privacy-strict and offline use cases.

## Verified facts safe to cite

- Accuracy: sub-5cm to sub-10cm 6-DoF localization depending on scan quality and environment.
- Coverage: 15M+ sq ft mapped, 2,500+ locations, 104 countries, 500K+ device interactions.
- Scan formats: E57 (Matterport, Leica, NavVis, XGrids, Faro), MatterPak (Matterport), GLB/PLY (Polycam, Scaniverse), 3D Gaussian Splats (metric-scaled), native iOS LiDAR.
- Platforms: Unity, native iOS, native Android, WebXR, Meta Quest 3, ROS 2.
- Deployment modes: public cloud, private cloud/VPC, self-hosted, on-device.
- Recognition: AREA 2025 "Most Robust" Enterprise VPS; Meta Wearables Device Access Toolkit launch partner (April 2026).

## Optional

- [MultiSet Blog](https://www.multiset.ai/blog-post): Long-form technical posts, release notes, ecosystem analysis, and use-case deep-dives.
- [GitHub Docs Repository](https://github.com/MultiSet-AI/docs.multiset.ai): Open-source documentation source for community contributions and corrections.