Serverless. Spatial-native. Pay as you go.
Pure spatial database. Built for AI agents.
Not a wrapper. Not an extension. A spatial engine built from the ground up — native CRS support, serverless provisioning, and built for AI agents.
One product. Two tiers.
The same spatial engine runs both. Start local, scale to cloud.
Plat
Free. Always.- CLI + Python interface
- Local .plat files
- Your machine
- Includes plat-mcp (AI agent tools)
Plat Cloud
Pay as you go- REST API + MCP interface
- S3-persisted datasets
- Serverless — provision in milliseconds
- Includes plat-mcp + full API
How it works
Three steps to a spatial query.
Provision
curl -X POST https://api.platlake.io/v1/sessionsQuery
SELECT * FROM wells WHERE ST_Within(geom, boundary)Terminate
curl -X DELETE https://api.platlake.io/v1/sessions/{id}Why pure spatial
Purpose-built beats bolt-on. Every time.
Built from the ground up
Spatial isn’t an extension or a plugin. It’s the foundation. Geometry, H3 indexing, and coordinate transforms are core engine capabilities — not afterthoughts.
Native CRS support
Work across coordinate systems without fighting your database. Plat understands spatial reference systems natively and guides you through transforms with intelligent suggestions.
Agent-ready
Provision a spatial database in milliseconds via REST API or MCP. Built for the speed AI agents require — not the minutes traditional databases take.
Simple pricing. No surprises.
Serverless. Pay as you go.
Plat Cloud — Compute
- REST API + MCP
- Serverless provisioning
- S3 persistence
- 30-day free trial
Plat Cloud — Storage
- Persistent datasets
- S3-backed storage
- Access from any session
- No egress fees
30-day free trial on Plat Cloud. No credit card required.
Frequently asked questions
Get started with Plat
Free. Always. CLI + Python.
pip
pip install platdbHomebrew
brew install platdbBinary
# Download from GitHub Releases
# Linux / macOS / WindowsQuick start
# Install
pip install platdb
# Query a spatial file
plat query 'SELECT * FROM wells.shp WHERE ST_Within(geom, boundary)'
# Connect to Claude Desktop
plat-mcp --database wells.plat