Spatial Intelligence for Real-World Businesses: Alive, Autonomous, and AI-Driven.
A zero-friction, 8-bit digital twin platform connecting real-time business telemetry (POS, Google Sheets, Booking APIs, Telegram) to an interactive spatial world. Configurable and operable by AI Agents through Model Context Protocol (MCP) — without writing a single line of frontend code.
One universal 8-bit spatial engine. Endless real-world business environments.
Whether you operate a boutique spa, an artisanal roastery, a high-velocity supermarket, or a franchise fitness network, SpatialTwin transforms flat telemetry feeds into a self-rendering, living simulation.
- Daily Revenue: ฿14,500 (28 sessions)
- Room RevPASH: 78.5% Utilization
- Staff Stamina: Somchai: 74% · Yuki: 88%
- SLA Compliance: 94.2% on-time starts
- Queue Length: 4 orders pending
- Table Occupancy: 82% (14 tables seated)
- Barista Stamina: Kenji: 91% [NOMINAL]
- Espresso Telemetry: Machine 1: 9.2 bar pressure
- Checkout Flow: Lane 2 bottleneck: 3 in queue
- Chiller Telemetry: Zone 4: +3.8°C (Compliant)
- Average Basket: ฿420.00 / transaction
- Hotspot Density: Fresh Bakery: 8 shoppers
- Cardio Utilization: 6/8 treadmills active (75%)
- Locker Density: 35% capacity
- Trainer Duty: 2 active personal sessions
- Facility Climate: 21.5°C HVAC nominal
- Network Size: 5 connected business nodes
- Aggregated Rev: ฿142,800 today
- Roaming Staff: 14 employees on active shifts
- Street Footprint: Kamphaeng Phet 7 Roi / RCA
[1] Spas & Wellness
Session countdowns, amber heat glow, staff stamina depletion, and client wait-time SLA alerts.
[2] Cafes & Bakeries
Table occupancy tracking, queue bottlenecks, barista stamina during morning rush hours.
[3] Retail & Marts
Checkout lane congestion, cold-chain temperature telemetry, and foot-traffic density heatmaps.
[4] Fitness & Studios
Cardio and strength equipment utilization, locker room occupancy, and trainer schedules.
Frontline business operations and AI agents are trapped behind a wall of tabular abstraction.
Digital systems process oceans of data every second, yet frontline managers and AI models remain blind to what is physically happening on the floor in real time.
Rows & Columns Obscure Physical Reality
Traditional POS registers, booking sheets, and CRM lists dump operations into endless rows. A spreadsheet cannot show at a single glance who has waited on the sofa the longest, or why Room 3 sits empty while guests crowd the hallway.
LLMs Cannot Act Without Spatial Awareness
Modern AI agents (Claude, Cursor, custom assistants) query SQL tables with ease, but lack a physical spatial canvas to comprehend, test, and rearrange real physical environments. They cannot detect bottlenecks or spatial layout friction.
Traditional Game Engines Require Endless Art Sprints
Traditional 3D or 2D digital twins (Unity, Unreal, Phaser hand-rigs) require dedicated 3D modelers, animators, and weeks of developer iterations just to add a single station. Scaling to thousands of physical stores is prohibitively expensive.
Static Code Cannot Keep Pace With Real Daily Shifts
In actual hospitality and retail, operational parameters change daily. Wait tolerances shift, therapist break intervals vary, and seasonal room usage fluctuates. Hardcoded software requires hiring developers for tiny operational adjustments.
An autonomous, self-rendering 8-bit digital twin governed by natural language and MCP.
SpatialTwin converts raw business feeds into a thriving, gamified isometric simulation layer. Anyone—and any AI Agent—can monitor, optimize, and orchestrate physical spaces without writing frontend code.
The Living World Dollhouse
Telemetry transforms into ambient physical cues. Over-threshold client wait times trigger animated impatience steam. Active treatments cast warm pulsing amber light. Overtired staff show stamina bars. Happy checkouts trigger coin showers and sakura petals.
AI-to-Tile in Seconds
Type a natural language prompt (“Add an outdoor bamboo onsen soaking tub”) or upload a phone snapshot. The background pipeline removes backgrounds, snaps pixels to a cozy 16-bit Kairosoft aesthetic, and generates atlas spritesheets on the fly.
Model Context Protocol First
Governed via native MCP tools. Any AI client (Claude Desktop, Cursor, LangChain) can query layouts, spawn equipment, bind live sensor streams, and inspect staff exhaustion using pure tool calls.
Parameters Over Static Code
Business rules are dynamic JSON data, not hardcoded logic. Tell your agent: “Allow up to 25 minutes of wait time and mandate a 15-minute tea break every 3 sessions.” The engine adapts physics, timers, and alerts instantly.
Built and running in production today versus our enterprise horizon.
We distinguish clearly between what is already verified in our codebase right now and our upcoming product roadmap.
[★] VERIFIED IN CODE TODAY
Fully functional in apps/web-runtime & apps/cloud-bridge
-
[x]
4-Layer 2.5D Isometric Engine Phaser 3.87 runtime with 2:1 isometric grid, modular walls with continuous depth projection, sprite reflection, and drag-and-drop furniture rearranging.
-
[x]
Dynamic Lighting & Particle Physics Additive amber illumination (#FFAA33) with harmonic pulsing for treatment rooms; particles for tub steam, tea vapor, gold coin checkout fountains, and sakura review showers.
-
[x]
Live Staff Stamina Engine Real-time stamina worker tracking physical strain by treatment type (Thai Massage vs Aromatherapy) and calculating rest recovery automatically.
-
[x]
Model Context Protocol Server (MCP v1.1.0) 12 production tools for Claude Desktop and Cursor covering equipment CRUD, spatial layout queries, stream bindings, and dynamic business rule reconfiguration.
-
[x]
Autonomous Staff Navigation & Pathfinding Room waypoint patrol routes, obstacle avoidance, and interactive double-click floor walking to dispatch staff anywhere on stage.
-
[x]
Multimodal Voice & AI Copilot ElevenLabs streaming voice with Web Speech API fallback, Vercel AI SDK (@ai-sdk/openai), and a zero-dependency local NLP parser.
-
[x]
100% Client Standalone Mode Browser simulation runs independently with zero external server dependencies when deployed to CDN or Vercel.
[→] UPCOMING ENTERPRISE HORIZON
Architected blueprint & scheduled milestones
-
[+]
Two-Way Telegram & WhatsApp Staff Bot Click-to-message directly from staff sprites on the floor; staff reply via chat (/break, /ready, /done) to automatically clear exhaustion badges.
-
[+]
Avalanche L1 On-Chain Notarization Cryptographic anchoring of operational hours and verified customer ratings onto an Avalanche L1 subnet for trust verification between franchisors and investors.
-
[+]
Smart-Split Tip & Commission Contracts Automated, transparent distribution of tips and service fees to staff crypto wallets immediately upon POS transaction confirmation.
-
[+]
Multi-Tenant Federated City View Interactive regional map displaying 10–50 franchise stores as connected 8-bit buildings with live revenue, queue sizes, and staff availability badges.
-
[+]
Hardware IoT Sensor Mesh Infrared room presence sensors and magnetic door contacts for hands-free session detection; RFID wristbands for automatic sauna and pool occupancy tracking.
-
[+]
Granular Role-Based Access Control (RBAC) Dedicated permission tiers: Network Brand HQ, Branch General Manager, and Receptionist Kiosk Signage (view-only mode).
Open-core freedom. Turnkey cloud operations. Enterprise scale.
Start completely free on your own machine, or let us host, connect, and scale your spatial twins with guaranteed uptime and AI infrastructure.
Open Core
For developers, indie hackers, and open-source AI researchers.
- [+] 100% self-hosted & open source under MIT License
- [+] Full Phaser 3.87 2.5D isometric web runtime
- [+] Local MCP v1.1.0 server spec with 12 tools
- [+] Offline procedural asset generator fallback
- [+] Standard REST & WebSocket JSON API hooks
- [+] Community support via GitHub and Discord
Managed SaaS
For single-location service businesses (spas, cafes, gyms, clinics, retail).
- [+] Zero-infrastructure: Fully managed cloud hosting
- [+] Direct webhook connectors (Google Sheets, POS, Telegram)
- [+] Cloud-hosted MCP Server & Gateway for all AI clients
- [+] Visual SLA alerting system with Telegram/SMS push notifications
- [+] Unlimited multi-user view access (Owner, Staff, Kiosk)
- [+] 50 AI custom asset generation credits included monthly
- [+] 99.9% uptime guarantee with automatic backups
Franchise & Enterprise
For multi-location brand operators, hospitality groups, and consultancies.
- [+] HQ Control Dashboard: Monitor 100+ branches simultaneously
- [+] Custom AI style training for brand-exact 8-bit assets
- [+] Enterprise Role-Based Access Control (RBAC) & Audit Logs
- [+] Historical operational replay to optimize floor throughput
- [+] Custom legacy integrations (1C, SAP, Oracle, legacy POS)
- [+] Private VPC deployment or hybrid on-premises hosting
- [+] Dedicated Solutions Architect & 24/7 priority SLA support
[+] VALUE-ADDED REVENUE STREAMS & ADD-ONS
Modular extensions designed to power enterprise customization and developer ecosystems.
Governed entirely through Model Context Protocol (MCP v1.1.0).
Every room, station, and telemetry threshold is accessible to AI agents via declarative tool definitions. Connect Claude Desktop, Cursor, or your autonomous backend in 60 seconds.
12 Declarative Tools for AI Agents
SpatialTwin treats spatial geometry and business rules as mutable state. AI assistants read live layouts, spawn equipment, bind real-time streams, and reconfigure operational constraints via natural language.
- get_layout(facility_id) Returns coordinates, active stations, staff locations, and waiting clients.
- place_station(entity_type, grid_x, grid_y, orientation) Spawns interactive massage tables, onsen tubs, or tea stations on the 2.5D grid.
- bind_data_stream(entity_id, source_url, json_path) Connects POS or Google Sheet cells to live ambient lights and status bubbles.
- configure_business_rules(rule_id, operator, threshold, alert) Dynamically updates wait-time tolerances and staff rest requirements without re-deploying code.
// claude_desktop_config.json
{
"mcpServers": {
"spatialtwin": {
"command": "npx",
"args": [
"-y",
"@spatialtwin/mcp-server@latest"
],
"env": {
"SPATIALTWIN_BRIDGE_URL": "https://api.spatialtwin.ai",
"SPATIALTWIN_API_KEY": "sk_live_rca_alley_9281"
}
}
}
}
// Example Natural Language Prompt to Claude:
"Claude, inspect the salon layout. If any guest
has waited on the sofa longer than 15 minutes,
trigger the impatience smoke alert, notify Somchai
via Telegram, and light up Room 1 for service."