Deployment Guide

This section helps you choose and implement the right deployment strategy for ChatSpatial based on your specific needs and resources.

Why Two Deployment Options?

ChatSpatial supports two distinct deployment scenarios because different users have different needs:

1. Local Deployment (Individual Use)

What it is: Install ChatSpatial directly on your personal computer and use it with Claude Desktop, Continue, or other MCP-compatible clients.

Best for:

Individual researchers with sufficient local computing resources
Quick prototyping and small dataset analysis
Users who already have Claude Desktop or similar MCP clients
Privacy-sensitive data that cannot leave local machines

How it works: ChatSpatial runs as an MCP server on your machine, directly communicating with your AI assistant through the MCP protocol.

Get Started: Installation Guide

2. Server Deployment (Team Collaboration)

What it is: Deploy ChatSpatial on a centralized server with a web interface (Open WebUI), allowing multiple team members to access it through their browsers.

Best for:

Research teams sharing computational resources
Large datasets requiring 128GB+ RAM
Groups without individual high-performance computers
Centralized data management and collaboration

How it works: ChatSpatial runs on a server, accessed through Open WebUI. Since Open WebUI does not natively support MCP, we use an mcpo proxy to bridge the protocols.

Get Started: Server Deployment Guide

Quick Decision Guide

Ask yourself these questions to choose the right deployment:

Do you have Claude Desktop or another MCP client installed?
- Yes → Consider local deployment
- No → Consider server deployment
Will multiple people need access?
- Yes → Server deployment is better
- No → Local deployment is sufficient
Do you have a powerful personal computer (32GB+ RAM)?
- Yes → Local deployment will work
- No → Server deployment recommended
Is your data sensitive and cannot leave your machine?
- Yes → Local deployment required
- No → Either option works

Comparison Table

Aspect	Local Deployment	Server Deployment
Users	Single user	Multiple users
Access	MCP client (Claude Desktop)	Web browser
Setup Complexity	Simple (pip install)	Moderate (Docker, nginx)
Resource Sharing	No	Yes
Minimum RAM	16GB	64GB
Data Location	Local machine	Centralized server
Cost	LLM API only	LLM API + Server
Best For	Individual analysis	Team collaboration

Technical Architecture Differences

Local Deployment

Claude Desktop/MCP Client
        ↓
    ChatSpatial MCP Server
        ↓
    Local Computing Resources

Server Deployment

Team Members (Browsers)
        ↓
    Open WebUI (Web Interface)
        ↓
    mcpo Proxy (Protocol Bridge)
        ↓
    ChatSpatial MCP Server
        ↓
    Server Computing Resources

Next Steps

For individual use with Claude Desktop → Installation Guide
For team collaboration via web browser → Server Deployment Guide
Need help deciding? → Contact Support

Server Deployment