Ask a question, get a map. A natural language interface for querying geospatial data.
📘 Latest: Version 0.3.1 - Query Builder Enhancement
Recent Changes:
- Aggregations & Analytics: Full GROUP BY support with all major aggregation functions (COUNT, SUM, AVG, MIN, MAX, STDDEV)
- Advanced SQL Features: CTEs (WITH clauses), subqueries, UNION operations, computed columns, and comprehensive WHERE operators (BETWEEN, IN, IS NULL)
- Spatial Calculations: ST_Area, ST_Length, ST_Centroid, ST_Perimeter in SELECT clauses
- Sorting & Pagination: ORDER BY and LIMIT for top-N queries
See the full CHANGELOG for details.
How many questions can be answered with a map? Questions from urban mobility to planning to climate response and more can be answered with geospatial data, but can require complex spatial SQL queries and data cleaning that is non-trivial for non-technical users.
Geoff takes a prompt in natural language, converts it into a spatial SQL query, and displays the result on a map - shortening the time from question to insight for planners, geographers, and more.
Problem Statement: Planners, NGOs, activists, public employees and more have questions with geospatial answers but are often limited by the collection and querying of spatial data.
- Natural Language Queries: Ask questions in plain English and get answers mapped to real geospatial data.
- Dynamic Schema Selection: Question is compared to vector descriptions of relevant datasets, ensuring efficient and accurate queries.
- Few-Shot Prompt Generation: Builds system prompts with cached examples to improve generation and reduce errors.
- Interactive Frontend: React + Tailwind interface with:
- Prompt bar for natural language input
- Table view of query results
- Interactive map layer displaying queried geometries
- Dynamic data dictionary to see available search criteria
- ETL Pipeline: Automated ingestion, cleaning, and transformation of multiple datasets, with an easy path to add more.
- Extensible Dataset Support: Current version supports 10 datasets; architecture allows seamless addition of new sources.
| Dynamic data dictionary |  |
| Explorable results |  |
Note: Geoff can only answer questions using data from its current datasets. Query capabilities are expanding with each release.
Supported Query Types:
- Basic Queries: Single and multi-layer selection with filtering
- "Show fire stations built after 1980"
- "Show parks and schools"
- Spatial Operations: Distance-based filters, intersections, spatial joins
- "Show bike lanes within 100m of schools"
- "Show each school and its nearest fire station"
- Attribute Joins: Join datasets based on shared attributes
- "Show the ward that Allan Gardens is in"
- Aggregations & Analytics: GROUP BY with aggregation functions
- "Count schools per neighbourhood"
- "Show total parking lot area within each ward"
- Advanced SQL: CTEs, subqueries, UNION operations, computed columns
- "Show the 5 longest bike lanes"
- "Merge all emergency service locations into one layer"
Coming Soon:
create: Geometry generation (buffers, centroids)route: Custom pathfinding and routing
For detailed specifications, see docs/specs/query_builder.md.
Application Architecture
flowchart LR
subgraph Data["Database: Postgres/PostGIS"]
ETL["ETL Pipeline<br>(ingest + transform)"] --> DB[("PostGIS Database")]
Vector_DB[("Vector Database")]
end
subgraph Backend["Backend: Python + FastAPI"]
Schema["Schema & Example selection (vector embed)"]
Prompt["System Prompt Builder"]
Query_Builder["Query Builder<br>(JSON → SQL)"]
Exec["Execute SQL, Parse GeoJSON Layers"]
end
subgraph Ollama["LLM"]
LLM["Generate Plan<br>(Natural Language → JSON)"]
end
subgraph Frontend["Frontend: React + Tailwind"]
UserPrompt(["User Question"])
Leaflet(["Layers on Leaflet Web Map"])
end
Frontend ~~~ Backend
Backend ~~~ DB
UserPrompt --Embed--> Schema
Vector_DB --> Schema
Schema --> Prompt
Prompt --> LLM
LLM --> Query_Builder
Query_Builder --> Exec
DB --> Exec
Exec --> Leaflet
Tech Stack
- Database
Docker,PostGIS,PostgreSQL,pgvector,Python,SQL
- Backend
FastAPI,Ollama(local LLMs),OpenAI,Python,geoalchemy,sqlalchemy
- Frontend
React,Tailwind CSS,Vite,Leaflet.js,OpenStreetMap,Node.js
- Deployment & Infrastructure
Docker,Nginx(server & reverse proxy), Cloud VPS, VPN Tunnel
Currently includes 10 Toronto datasets with spatial and attribute data. The in-app data dictionary provides complete, up-to-date schema information.
Available Datasets:
- Ambulance/EMS Stations
- Attractions / Points of Interest
- Bike Lanes
- Fire Stations
- Neighbourhoods
- Parking Lots
- Parks
- Police Stations
- Schools
- Wards
See docs/tech/data_schema.md for detailed schema information.
Documentation is available in the docs/ directory:
- Current State: What's implemented, in progress, and planned
- Vision: Long-term goals and aspirational capabilities
- Roadmap: Prioritized feature development plan
- Changelog: Detailed version history
Technical Documentation:
- Architecture: System design and data flow
- Modules: Detailed module specifications
- Data Schema: Database structure and datasets
- Query Builder Spec: SQL generation capabilities
- JSON Plan Spec: LLM output format
Use Cases:
🔜 Next:
- Layer History: Previous query results panel with visibility controls
- Geometry Creation: Buffers, centroids, and synthetic geometry generation
- Layer References: Reference previous layers in new queries
🎯 Future:
- Place Resolution: Better handling of place name ambiguity and hierarchies
- Transparency: Query explanations, clarification questions, SQL visibility
- User Feedback: Thumb up/down with automated improvement loop
- Workspaces: Multi-layer management and export capabilities
See docs/roadmap.md for detailed feature descriptions.
Suggestions & feedback are currently welcome. Open contribution is not currently available. If Geoff could help you or your organization, please reach out.
Instructions on setting up Geoff locally for custom use or additional data access will be available in the future. Reach out if this is a priority for you.
Datasets currently used by Geoff are sourced from: