GeneGIS

_{North-star prompt 名古屋市の人口密度を表示 · real pipeline output · DuckDB verified}

AI-native · Cloud-native · GPU-native open geospatial workbench

If GIS were invented in 2026, it would not look like a 2000s desktop app.

GeneGIS is not a QGIS clone. It is a next-generation GIS platform built around workflow graphs, AI agents, cloud-optimized formats, and GPU rendering — designed for spatial intelligence in the GeoAI era.

Why GeneGIS exists

Traditional GIS asks you to find data, fix CRS, wire geoprocessing by hand, validate results yourself, and export maps elsewhere. GeneGIS inverts that:

Intent → Data Discovery → Workflow Graph → Verified Execution → Map / Insight / Report

Example north-star prompt:

名古屋市の人口密度を表示

GeneGIS resolves the place, discovers datasets, normalizes CRS, computes density, renders a choropleth, and shows sources + workflow graph + verification — not just a chat reply.

Five differentiators

Pillar	What it means
AI Agent Native	Agents plan and verify spatial workflows, not just chat
Cloud Native Data First	GeoParquet, COG, COPC, PMTiles, STAC as first-class citizens
GPU First	LOD, tiles, range reads — never load billions of features wholesale
Figma for GIS	Collaboration, comments, branches, style systems at the center
VSCode for GIS	WASM / TS / Rust / Python SDK + marketplace extensibility

Quick start (Phase 0–1)

# Build the workspace
cargo build --workspace

# Print the MVP workflow graph (IR only)
cargo run -p genegis-cli -- workflow run nagoya-density

# North-star one-liner (Intent → Workflow → Map)
cargo run -p genegis-cli -- ask "名古屋市の人口密度を表示"

# Plan only (human-in-the-loop / Strict mode preview)
cargo run -p genegis-cli -- ask "名古屋市の人口密度を表示" --plan-only

# Execute analysis + DuckDB verification + summary JSON
cargo run -p genegis-cli -- workflow run nagoya-density --execute

# Execute and export choropleth HTML map
cargo run -p genegis-cli -- workflow run nagoya-density -x --html -o nagoya-density.html

# Execute and export choropleth PNG map
cargo run -p genegis-cli -- ask "名古屋市の人口密度を表示" --png --no-html

# Rebuild ward boundaries from 国土数値情報 N03 (optional)
python3 scripts/build-nagoya-wards.py

# Full example (writes examples/nagoya-population-density/output/)
cargo run -p nagoya-population-density

# COPC metadata smoke (local PDAL fixture)
cargo run -p copc-metadata

# Plugin discovery smoke
genegis plugin list

# Collaboration smoke
genegis collab comment list
genegis collab export -o .genegis/collab.json

# Agent orchestration smoke (offline rule planner + DuckDB verify)
genegis agent run "名古屋市の人口密度を表示"
genegis agent run "名古屋市の人口密度を表示" --json -o .genegis/agent-run.json
genegis agent plan "名古屋市の人口密度を表示" && genegis agent execute
genegis agent export-audit -o .genegis/audit-bundle.json

# Second verified workflow (remote COG metadata via HTTP range-read)
genegis agent run "リモートCOGデモのメタデータを表示"

# Third verified workflow (local bundled COG — offline)
genegis agent run "ローカルCOGデモのメタデータを表示"
genegis catalog stac list

# Multi-client collab demo (terminal 1: server, terminal 2: workbench)
cargo run -p genegis-server
cargo run -p genegis-workbench
# CLI sync against server
genegis collab pull
genegis collab comment add "Check 中区 density" --author reviewer
genegis collab push

# Desktop workbench (Tauri — requires extra patches on Rust 1.94)
cd apps/desktop && npm install && npm run dev

# Local web workbench (recommended MVP launcher; no Tauri deps)
cargo run -p genegis-workbench

# Collab sync server (GET/PUT session JSON)
cargo run -p genegis-server

# WebGPU canvas prototype (requires GPU + display)
cargo run -p genegis-render --example canvas_prototype

# WebGPU choropleth — Nagoya population density (Phase 2 alpha)
cargo run -p genegis-render --example choropleth_nagoya

Architecture at a glance

Intent → GeoWorkflow IR → Verified Execution → Map
              ↑
         AI + CLI + UI (all emit Commands)
              ↓
    GIS Core (Rust) + DuckDB + wgpu + Cloud formats

See docs/architecture/overview.md and docs/rfcs/0001-master-architecture.md.

Repository layout

crates/     Rust engines (core, render, workflow, ai, …)
apps/       Desktop (Tauri), web, server, CLI shells
plugins/    Official and community extensions
sdk/        Rust, TypeScript, Python SDK
docs/       Architecture, ADRs, RFCs, roadmap
examples/   Reproducible demos (Nagoya density, COG, COPC, …)

Roadmap → GitHub Stars

Phase	Theme	Star target
0	Foundation / Manifesto	0 → 300
1	MVP: Nagoya density demo	300 → 1,000
2	Alpha: GPU choropleth, GeoParquet, catalog	1,000 → 2,500
3	Beta: STAC, COG, GPU workbench integration	2,500 → 5,000
4	Plugins & COPC — SDK, WASM host, point cloud alpha	5,000 → 7,500
5	Figma for GIS — comments, branches, collab sync	7,500 → 10,000
6	Autonomous GIS platform — multi-agent orchestration	10,000 → 15,000
7	Audit trail & release workbench — run history + provenance UI	15,000 → 20,000
8	Intent expansion — multi-workflow agent verify beyond Nagoya	20,000 → 30,000

Tech stack (decisions)

• Core: Rust
• Rendering: wgpu / WebGPU
• Desktop: Tauri + TypeScript UI
• Local analytics: DuckDB Spatial
• Cloud vectors: GeoParquet
• Enterprise DB: PostGIS

Contributing

See CONTRIBUTING.md. We use RFC culture for major design changes.

License

Licensed under Apache-2.0 OR MIT at your option.

---

GeneGIS is not a GIS with AI. GeneGIS is a GIS designed for AI agents and humans to collaborate.