KxAI

Ask your AI about the real world.
Control what it can see.

Natural-language access to geospatial data through Claude, ChatGPT, or your own AI agents — built on MCP. Plus KxCLI, a context-cheap CLI shaped for AI coding tools. With enterprise governance from day one.

Try KxAI free → On-prem AI evaluation →
Chat

Your team asks questions in plain language

Engineers, analysts, planners query geospatial data without touching a GIS tool. Works inside the AI assistants they already use.

  • Natural-language queries against any connected dataset
  • Results with sources and map visualisations
  • Works in Claude, ChatGPT, or custom agents
  • Public data on Koordinates available immediately
Govern

You control what AI can access

Full governance layer: which agents connect, which datasets they see, what operations are allowed, and a complete audit trail of every query.

  • Access policies per dataset, per agent, per user
  • Complete audit trail of every AI interaction
  • MCP connection management dashboard
  • Token lifecycle & authentication controls
Build

And a CLI for the agents that build with your data.

Chat agents talk to your data through MCP. Coding agents — Claude Code, Cursor, CI pipelines — reach for the shell.

KxCLI is a thin, JSON-out wrapper around the Koordinates publish and management API. Small enough to fit in an agent's context, structured enough to script around, and governed by the same policies as the MCP surface.

kx layer create kx layer publish kx style apply kx token issue
Explore KxCLI → Developer docs →
Claude Code · kx CLI 
 # Agent publishes a new layer end-to-end $ kx layer create --from parcels.geojson --json → {"id":48291,"status":"draft"}
$ kx style apply 48291 --from styles/parcels.json
$ kx layer publish 48291 --json → {"id":48291,"status":"published"}
How It Works

Built on MCP — the open protocol for AI tool use

Model Context Protocol lets AI assistants use tools safely. KxAI implements MCP for geospatial data — so any MCP-compatible AI can query, geocode, find features, and render maps.

💬

Ask

Natural language

🔐

Govern

Policy check

🔍

Query

MCP tools

📊

Return

Data + sources

Live Demo

See it answer a real question

A user asks about flood risk at a specific address. KxAI geocodes the address, queries hazard layers from Auckland Council, and returns a cited answer — all through MCP.

geocode_address find_features_near_point get_mapviewer_widget
Claude + KxAI MCP
Is there any flood or liquefaction risk at 12 Arney Road, Remuera?
kxai.geocode_address()
kxai.find_features_near_point()
12 Arney Road, Remuera sits outside the 1% AEP floodplain with low liquefaction susceptibility. One overland flow path runs along the northern boundary. No tsunami or coastal inundation exposure. Sources: Auckland Council GeoMaps, LINZ.
Use Cases

Who uses KxAI

🏛

Government & Open Data

Sovereign AI governance for authoritative public datasets. Control which AI agents can access what.

Learn more
⚙

Engineering

Site constraint queries in seconds. Hazards, zoning, geology — all cross-referenced by AI.

Learn more
🗺

App Builders

Give your application's users natural-language access to your geospatial data via API.

Learn more
Deployment

Cloud by default. On-prem when you need it.

KxAI is a data and AI governance layer that indexes your datasets, connects to third-party sources, and controls AI access. It runs in the cloud with every Koordinates site. For regulated environments, it's also available as an on-premises licence — standalone or connected to your Koordinates cloud.

Deployment options → Governance deep dive →