CoCivil is an advanced AI-powered due-diligence and design platform tailored for land development in Ontario, Canada. It bridges the gap between plain-English development queries and complete planning submission packages (compliance matrices, rationales, precedent reports, massing models, and floor plans) in a fraction of the usual time.

• Search any Toronto address or parcel PIN.
• Instantly retrieve zoning information from Zoning By-law 569-2013, including:
  • Official Plan designations
  • Overlays
  • Setbacks
  • Height limits
  • Lot coverage
• Data is sourced directly from Toronto Open Data.
• Performs adjacency analysis, frontage detection, and spatial joins for full site context evaluation.

• Conversational chat interface with an AI agent trained on Ontario planning law.
• Grounded in key legislation and policy frameworks including:
  • Planning Act
  • Provincial Planning Statement (PPS 2024)
  • Ontario Regulation 462/24
• Capabilities include:
  • Answering complex zoning questions
  • Proposing variance strategies
  • Identifying optimal approval pathways.

Automatically generates development documentation including:

• Planning rationales
• Compliance matrices
• Shadow studies
• Planning submission packages

Outputs are:

• Auditable
• Grounded in cited planning policies
• Flagged for professional review.

• Describe a building in plain English.
• Generates a 3D massing model using Three.js.

• Interactive floor plan editing powered by Konva.js.
• Real-time Ontario Building Code (OBC) compliance checking.
• Violations are highlighted instantly.

Supports planning and analysis for municipal infrastructure systems.

Features include:

• Upload pipeline DXF files.
• 3D visualization of pipe networks.
• OPSD-compliant specifications.
• Infrastructure condition assessments.

Supported systems:

• Water
• Electrical
• Stormwater and drainage networks.

AI-powered precedent discovery across:

• Committee of Adjustment decisions
• Ontario Land Tribunal (OLT)
• CanLII legal databases

Capabilities:

• Identifies comparable approvals and refusals near the subject site
• Calculates entitlement risk scores based on:
  • Compliance gaps
  • Variance magnitudes
  • Policy conflicts.

Automatically evaluates development feasibility.

Features include:

• Unit mix and floor area allocation
• Layout optimization
• Construction cost estimation
• Revenue modeling
• Net Operating Income (NOI) projections
• Property valuation estimates

Also supports Variance Simulation, allowing users to fork policy scenarios and test development outcomes under different approval assumptions.

Fine-tuned RAG pipeline for policy and legal retrieval.

Capabilities:

• Vector search using ChromaDB
• Searches across:
  • Ontario planning documents
  • Zoning by-laws
  • Building code references
• Extracts and cites exact legal clauses and normalized regulatory rules.

CoCivil is structured as a modular monolith optimized for asynchronous analysis workflows and spatial data processing.

Framework

• React 19 + Vite

Mapping & Visualization

• MapLibre GL for GIS mapping
• Three.js for 3D building and infrastructure massing
• Konva.js for:
  • 2D floor plan editing
  • Pipe network editing

Authentication

• Auth0

Key UI Components

• ChatPanel
• PolicyPanel
• ModelViewer
• FloorPlanEditor
• InfrastructureViewer

Framework

• FastAPI (Python)

Functions as:

• API Gateway
• Orchestration layer

Database

• PostgreSQL 16
• PostGIS for geospatial operations
• SQLAlchemy 2.0 (async)
• Alembic for database migrations

Task Queue

• Celery 5.3+
• Redis message broker

Used for long-running tasks such as:

• OCR processing
• Massing model generation
• Financial modeling workflows.

AI Integration

• Claude (primary)
• OpenAI (fallback)

Design principle:

• Deterministic compliance engines first
• AI used for extraction, reasoning, and summarization.

Components include:

• Document ingestion pipeline
• Semantic chunking and segmentation
• Vector indexing

Supported inputs:

• PDF documents
• HTML planning policies
• GIS data layers

Vector storage is handled by ChromaDB.

Sequential AI research pipeline designed for Ontario site feasibility.

Workflow chain: Source Discovery
↓
Parcel Zoning
↓
Buildability Analysis
↓
Approval Pathway
↓
Precedent Research
↓
Constraints & Flags
↓
Report Generation

Each stage feeds structured outputs into the next stage to produce a final planning report.

The system strictly versions all data and outputs to ensure deterministic and reproducible results.

Manages:

• Parcel boundaries
• Addresses
• Current land uses
• Planning overlays (transit, heritage, floodplains).

Separates:

• Raw legislative text
• Normalized structured regulatory rules

Example normalized rule: max_height_m = 36

Generates buildable envelopes using:

• Parcel geometry
• Zoning and policy constraints.

Produces reproducible artifacts including:

• PDFs
• CSV files
• spreadsheets
• 3D models.

Rapid evaluation of site feasibility and financial viability.

Preparation of zoning research and planning submission packages.

Quick generation of massing options and floor plan layouts.

Infrastructure network design and gap analysis.

Application review with standardized compliance checks.

./scripts/bootstrap_data.sh 

for Adding data to DB