You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
⭐ If you want to see new updates and database versions and if you find our tools useful please give our repositories a star to see more similar applications for the construction industry.
Star OpenConstructionERP on GitHub and be instantly notified of new releases.
What's New in v4.6.0 — Floating chat + Accommodation + Geo overlay + 10 dashboard widgets
The v4.6.0 wave adds a floating ERP-aware chat on every page, a new Accommodation module that unifies worker camps / rentals / hotels, 10 new dashboard widgets with server-synced layout, DWG/PDF raster overlay in the Geo Hub globe, and 6 per-module guided tours. 111 modules ship in the box; the /api/v1/* surface remains the public contract.
Floating chat — every page, talks to your ERP DB
Bottom-right FAB on every screen opens a panel backed by 17 database tools (projects, BOQ, schedule, validation, risks, CWICR search, BIM elements, semantic search across all modules). Streamed responses with tool-call cards rendered inline.
Construction cost estimation software is expensive, closed-source, and locked to specific regions. OpenConstructionERP changes that.
What you get
How it works
Free forever
AGPL-3.0 license. No subscriptions, no per-seat fees, no vendor lock-in.
Your data, your server
Self-hosted. Everything runs on your machine — nothing leaves your network.
27 languages
Full UI translation: English, German, French, Spanish, Portuguese, Russian, Chinese, Arabic, Hindi, Japanese, Korean, and 16 more.
30+ regional standards
DIN 276, NRM 1/2, CSI MasterFormat, GAEB, ГЭСН, DPGF, GB/T 50500, CPWD, ÖNORM, Birim Fiyat, Sekisan, SINAPI, and more.
AI-powered
Connect any LLM provider (Anthropic, OpenAI, Gemini, Mistral, Groq, DeepSeek) for smart estimation.
55,000+ cost items
CWICR database with 11 regional pricing databases (DACH, UK, US, France, Spain, Brazil, Russia, UAE, China, India, Canada).
How It Compares
Capability
OpenConstructionERP
Enterprise BIM
CAD Takeoff
Legacy Estimating
PDF Markup
License
AGPL-3.0 (free)
Proprietary
Proprietary
Proprietary
Proprietary
Self-hosted / offline
✔
✖
✖
⚠ partial
✖
Price
Free forever
~€500/mo
~€300/mo
~€200/mo
~€30/mo
AI estimation
✔ 7 LLM providers
✖
✖
✖
✖
UI languages
27
5
3
2
8
Regional standards
30+
4
3
2
—
BOQ editor
✔
✔
✔
✔
✖
CAD/BIM takeoff
✔ RVT IFC DWG DGN
✔
✔
✖
PDF only
4D/5D planning
✔
✔
✖
✖
✖
Cost database included
✔ 55K+ rates
✖ extra
✖ extra
✖ extra
✖
Resource catalog
✔ 7K+ priced
✖ extra
✖
✖
✖
Validation engine
✔ 42 rules
⚠ limited
✖
✖
✖
REST API
✔ full
⚠ limited
✖
✖
✖
Real-time collab
✔ soft locks
✔
✖
✖
✖
Open data export
✔ GAEB · XLSX · JSON
⚠ limited
⚠ limited
⚠ limited
PDF only
IDS / COBie requirements
✔ import + export
✖
✖
✖
✖
Property dev lifecycle
✔ Lead → SPA → Handover
✖
✖
✖
✖
3D globe / geo-anchor
✔ Cesium 3D Tiles
⚠ map only
✖
✖
✖
Comparison reflects typical category capabilities based on publicly available information as of Q1 2026. Pricing is approximate (per-seat, list price) and varies by vendor and region. OpenConstructionERP is an independent open-source project and is not affiliated with any commercial vendor in the categories above.
See It In Action
Each block below is a short GIF cut from the full walkthrough above — same order as the video, so you can jump to whichever workflow matters most. Prefer one continuous video? ▶ Watch the 12-minute walkthrough on YouTube.
1 · Role-Based Onboarding Sign in as Admin / Estimator / Manager — the wizard pre-selects the right 17 of 46 modules for your role
2 · New Project, Any Region Pick currency, classification standard, regional factor — live map & weather come along for free
3 · Build the Bill of Quantities Keyboard-first editor, 55K+ priced items, AI cost finder & Smart AI — quality score updates live
4 · BIM → BOQ Bulk Link Link 100 Revit walls → one BOQ line with aggregated area / volume / length — no IfcOpenShell
5 · DWG Drawings & Layers 636 wall entities across 10 DWG layers — every one linkable to the BOQ, measured in place
6 · PDF Takeoff Drop a floorplan, measure distance / area / count, push the numbers straight into the BOQ
CAD/BIM takeoff — Upload Revit (.rvt), IFC, AutoCAD (.dwg), or MicroStation (.dgn) files. DDC converters extract elements with volumes, areas, and lengths automatically
Interactive QTO — Choose how to group extracted data: by Category, Type, Level, Family. Format-specific presets for Revit and IFC
Linked geometry preview — Click the BIM link badge on any BOQ position to see a 3D preview of linked elements with interactive rotate/zoom/pan controls
BIM Quantity Picker — Select quantities (area, volume, length) directly from linked BIM elements and apply them to BOQ positions. The source parameter name is shown next to the unit
DWG polyline measurement — Click any polyline in the DWG viewer to instantly see area, perimeter, and individual segment lengths with on-canvas labels
PDF measurement — Open construction drawings directly in the browser. Measure distances, areas, and count elements with calibrated scale
AI estimation — Describe your project in plain text, upload a building photo, or paste a PDF — AI generates a complete BOQ with quantities and market rates
AI Cost Advisor — Ask questions about pricing, materials, or estimation methodology. AI answers using your cost database as context
Cost matching — After AI generates an estimate, match each item against your CWICR database to replace AI-guessed rates with real market prices
Geo Hub (3D Globe)
Anchor every project on a real spherical earth — Cesium 3D Tiles 1.1 with live HUD and pin layers:
Pin layers — HSE incidents, Punchlist items, Daily Diary entries plotted on the globe with category icons
"View on map" CTAs — One click from BIM viewer, PropDev plot, Daily Diary entry, Project card → globe with that asset selected
DWG / PDF raster overlay — Upload a site plan or floorplan, drag four corner pins onto the globe → the raster drapes over real terrain; polygon crop with vertex drag for cookie-cutter trimming
Canonical pipeline — POST /api/v1/geo-hub/from-canonical/{cad_import_id} turns any DDC cad2data conversion into glTF 3D Tiles via pure-Python pygltflib (no commercial toolkit needed)
Example: open a Berlin masterplan in Geo Hub, switch to Development mode, see all 12 plots colored by sale status, click one → reservation pipeline opens with that buyer pre-filtered.
Property Development
End-to-end real-estate developer workflow — from first lead to handover snags to warranty close-out:
Lead → Reservation → SPA → Handover → Warranty — Full lifecycle FSM with auto-creation of ContractParty on SPA conversion, Payment Schedule state machine, idempotent stage transitions
Sub-entity tabs — Phases · Blocks · Brokers · Price Matrix · Escrow — one screen for every dev operation, no page hops
House Type catalogue — ISO 3166-1 picker covering 180+ countries plus Custom region; CountryCombobox + HouseTypeEditModal share the same backend taxonomy as catalog & costs
SnagsBlock per handover — Photo upload, status (Open / Resolved / Disputed), one-click promote-to-warranty when a snag survives the defects-liability period
Contacts ↔ PropDev bridge — Every Lead and Buyer is idempotently tagged as a Contact via Contact.module_tags, so CRM and PropDev stay in sync without duplicates
Price Matrix — Per-phase × house-type × view-premium grid with currency-aware totals and bulk apply
Escrow — Per-buyer payment schedule with milestone receipts and outstanding balance roll-up
Bootstrap to Accommodation — One click on a development block creates a worker-camp / rental inventory in the Accommodation module (1:1 plots → rooms, idempotent)
Example: import a 240-unit residential masterplan, generate price matrix from house-type × view, push to globe, accept 18 reservations across 3 brokers, convert 11 to SPA, hand over 4, track 7 open snags in the warranty period — all in one app.
Accommodation
One module for three lodging kinds — worker camps for site crews, rentals for staff, hotels for visiting consultants — with rooms, bookings and charges in a unified data model:
Three kinds, one module — worker_camp · rental · hotel, with tab filter and per-kind capacity counters on every card
Rooms with status — available · occupied · maintenance · blocked; 409 prevents booking into a blocked or maintenance room
Booking state machine — reserved → checked_in → checked_out with cancelled from any non-final state; idempotent same-state updates; final states locked
Charges with Decimal precision — Base rent, extras, deposits, refunds, all in the room's inherited currency (no hardcoded EUR)
PropDev bootstrap — One click on a development block iterates its plots and creates rooms 1:1, idempotent (running twice creates nothing extra)
HR autobook (suggest-confirm) — Pick an employee Contact → suggest the lowest-labelled available worker_camp room → human confirms with a real booking POST
BIM + Geo aware — bim_element_id carries through from PropDev plots; cards with geo_lat/geo_lon get a "Geo" deeplink to the globe
IDOR-hardened — Every helper returns 404 (never 403) on cross-tenant access; tested in backend/tests/modules/accommodation/
Example: 240-plot worker camp on a remote site — bootstrap from the PropDev block, HR autobooks 187 crew members from the Contacts directory over three weeks, base-rent charges roll up to the project P&L automatically.
Floating Chat with the ERP Database
Bottom-right floating chat on every page — talks to the entire ERP database through 17 typed tools (projects, BOQ items, schedule, validation, risks, CWICR search, BIM elements, full semantic search):
That's it. Installs backend + pre-built React frontend in one wheel (~7.4 MB), opens your browser at http://localhost:8080, creates a SQLite database, and seeds the three demo accounts on first boot. No Docker, no Node.js, no extra services. PyPI package.
Ubuntu / Debian users: on Ubuntu 23.04+ (including Ubuntu 26) and Debian 12+, pip install directly to the system Python fails with error: externally-managed-environment (PEP 668). Use a venv:
Alternative 3: Local development (clone + npm + uvicorn)
git clone https://github.com/datadrivenconstruction/OpenConstructionERP.git
cd OpenConstructionERP
# Install dependencies
pip install -e ./backend[server]
cd frontend && npm install &&cd ..
# Start (Linux/macOS)
make dev
# Start (Windows — two terminals)# Terminal 1: cd backend && uvicorn app.main:create_app --factory --reload --port 8000# Terminal 2: cd frontend && npm run dev
Open http://localhost:5173 — for hacking on the codebase. Requires Python 3.12+ and Node.js 20+.
Demo Accounts
Three demo accounts are created automatically on first start. Each
password is generated per installation (via secrets.token_urlsafe)
and printed to the backend startup log so you see it immediately, e.g.:
[seed] Demo user created: demo@openestimator.io / xK7p_Q2nR8sT4uV6wX9yZ
[seed] Pre-set DEMO_USER_PASSWORD env to skip random generation
The same passwords are also persisted to
~/.openestimator/.demo_credentials.json (chmod 600) so you can recover
them later. To pin known passwords (e.g. for a team demo or CI), set the
env vars before the first boot:
Demo accounts include 5 pre-loaded projects from Berlin, London, New York, Paris, and Dubai with complete BOQs, schedules, and cost models.
Security note. For any internet-exposed deployment, set the three
DEMO_*_PASSWORD variables to strong, unique secrets, or disable demo
accounts entirely with DISABLE_DEMO_ACCOUNTS=1. Do not reuse
passwords from examples, screenshots, or earlier versions of this README.
Full RTL support (Arabic), locale-aware formatting
Architecture
How the platform turns raw CAD/BIM into structured ERP data
OpenConstructionERP is built around seven cooperating pipelines that turn closed CAD/BIM files (RVT, IFC, DWG, DGN, PLN, TSK) into structured, queryable ERP data — without locking you into a proprietary stack. Every module in the platform plugs into one or more of these stages:
Mining — collect existing project data (CAD/BIM models, CDE drops, COBie deliverables) into a semi-structured pool. Handled by cad, documents, bim_hub, file_search modules + the DDC cad2data converters (RVT/IFC/DWG/DGN/PLN → canonical JSON).
QTO Check & Quantity Take-off rules — apply rule sets per discipline / family / type to extract quantities deterministically. Handled by takeoff, requirements, validation, and the rule editor in match-elements.
BlackBox (company standard) — codify your firm's classifications, formulas, unit factors and assembly recipes as a single canonical rule book (COBie / CFIHOS / SQL / Excel / Access). Handled by costs, assemblies, catalog, cost_intelligence.
New project modeling — apply the BlackBox to a fresh model: geometry → filtering → grouping → verification → BOQ. Handled by boq, projects, match-elements, clash.
BlackBox mapping — map a new project's raw element list onto the canonical rule book via vector + lexical + rule-based matchers. Handled by match-elements (7-stage pipeline) + AI Estimate.
Project-specific data (4D/5D/6D) — derive scheduling, cost, hours, ordering and environmental footprint per group. Handled by scheduling, advanced_schedule, 5d_planner, risk, carbon, hse.
Saving data & Machine Learning — persist project history into the database, data lake and ML models so each new project starts further ahead. Handled by analytics, bi_dashboards, ai_chat, cost_intelligence, and the rule-learning loop in clash (Wave A4) and match-elements.
The right-hand side of the diagram is the automatic data retrieval layer that every UI surface (Dashboard, BOQ editor, /clash, /match-elements, /files, /scheduling) consumes: dashboards, calculations, reports, tables, charts, geometries, and ERP-ready exports — all driven from the same canonical store. The Machine-Learning column on the far right is where the platform progressively automates classification, parameterisation, recognition and marking, replacing manual steps () with automatic steps () over time.
This pipeline is the reason OpenConstructionERP can replace several commercial point-solutions with a single self-hosted stack — every module reads from and writes to the same canonical data layer.
Technical stack
flowchart TB
UI["Frontend SPA<br>React 18, TypeScript, Vite<br>AG Grid, Tailwind, PDF.js"]
subgraph Backend ["FastAPI Backend, 111 modules"]
CORE["Core<br>Module loader, Event bus, Hooks, RBAC<br>Validation, FSM + audit log"]
ESTIM["Estimating<br>BOQ, Costs, Catalog, Assemblies<br>Takeoff, BIM Hub, Match-Elements, 5D"]
FIELD["Field Operations<br>Service, Equipment, Daily Diary<br>Portal, Resources & Crew"]
COMM["Commercial<br>CRM, Contracts, Subcontractors<br>Bid Management, Variations, Suppliers, Property Dev"]
QSAFE["Schedule & Quality<br>Advanced Schedule (LPS/CPM/EVM)<br>QMS, HSE, Carbon & ESG, BI Dashboards"]
AIS["AI<br>AI Chat SSE, AI Estimate, Cost Intelligence<br>7 LLM providers"]
CORE --> ESTIM
CORE --> FIELD
CORE --> COMM
CORE --> QSAFE
CORE --> AIS
end
subgraph Data ["Data layer"]
PG[("PostgreSQL 16<br>SQLite in dev")]
VEC[("Vector DB<br>LanceDB / Qdrant")]
S3[("MinIO / S3<br>files, CAD, PDFs")]
end
subgraph Pipelines ["Pipelines"]
CAD["DDC cad2data<br>RVT, IFC, DWG, DGN to canonical JSON"]
CV["CV / OCR<br>PaddleOCR 3.0, YOLOv11<br>PDF takeoff, symbol detection"]
end
UI -- "REST, SSE" --> CORE
BIZ --> PG
BIZ --> VEC
BIZ --> S3
CAD --> BIZ
CV --> BIZ
classDef fe fill:#1f6feb,stroke:#0d3885,color:#fff;
classDef be fill:#238636,stroke:#104822,color:#fff;
classDef ai fill:#db6d28,stroke:#7a3c14,color:#fff;
classDef data fill:#8250df,stroke:#4a2c7d,color:#fff;
class UI fe
class CORE,BIZ be
class AIS ai
class PG,VEC,S3,CAD,CV data
LoadingPlain-text version (for screen readers or non-Mermaid renderers)
OpenConstructionERP includes security hardening for production deployments:
Path traversal protection on all file download endpoints
CORS wildcard blocking in production mode
Bounded input validation on bulk price operations
Generic error responses to prevent account enumeration
Production startup checks for secrets, credentials, and database configuration
Do not open public GitHub issues for security vulnerabilities.
Report privately via
GitHub Security Advisories
or by email to
info@datadrivenconstruction.io.
See SECURITY.md for the full coordinated-disclosure
policy, response timelines, supported versions, scope, and
regulatory-reporting obligations under EU Regulation 2024/2847
(Cyber Resilience Act).
AI disclaimer
AI suggestions produced by this software are preliminary estimates. A
qualified construction-estimation professional must verify all
quantities, classifications, and costs before any contractual or
tender-submission use. See NOTICE and TERMS.md §4.
Trademarks
All product names, logos, and trademarks referenced in this repository
are property of their respective owners. Any comparative references to
commercial categories or products are made for fair comparative purposes
based on publicly available information. OpenConstructionERP is an
independent project and is not affiliated with, endorsed by, or sponsored
by any third-party trademark owners. Full attributions in NOTICE.
Export control
This software contains cryptographic functionality (bcrypt password
hashing, JWT signing). Export is classified under US EAR 740.17 (TSU
mass-market exemption) and EU Regulation 2021/821 (dual-use). The
Software is not authorised for download, use, or re-export to
jurisdictions subject to comprehensive OFAC sanctions. See NOTICE
for the full notice.
License
AGPL-3.0 — see LICENSE. Third-party attributions in
NOTICE.
You can freely use, modify, and distribute this software. If you modify
and deploy it as a service, AGPL §13 requires you to make the
corresponding source code available under the same licence.
Unlock the Power of Data in Construction Move to full-cycle data management where only unified structured data & processes remain and where your data is yours