Construction Project Monitor · AI Reporting Suite

Every Report.
Every Discipline.
Written Automatically.

Drop in a weekly PDF or a monthly spreadsheet — an n8n workflow reads it, an AI agent classifies it by discipline and document type, and a backend engine turns it into EVM metrics, cross-discipline rankings, and a formatted report that emails and Telegrams itself out before the coffee's cold. This hub tracks the four moving pieces: extraction, reporting, mobile, and the automation gluing it all together.

24
n8n workflow nodes, one automated pipeline
10
Construction disciplines classified
3
Specialized extractors — EVM, Manpower, Cost
2
Notification channels — Email & Telegram

— Available Modules

One Pipeline, Four Views

A document goes in one end of the n8n workflow and a formatted report, a live dashboard update, and a push notification come out the other. Each card below is a different angle on that same pipeline — pick whichever piece you're working on.

🧠
n8n · GPT-4 · OCR
Data Extraction Engine
Document → Classification → EVM metrics
OCR pulls text out of PDFs, an AI classification agent sorts it by discipline and document type, then one of three specialized extractors turns raw text into real numbers.
  • Three-level classification: discipline, document type, sub-category
  • Full EVM extractor — PV, EV, AC, CPI, SPI, EAC, VAC
  • Manpower & cost extractors with trend detection
  • Generic fallback so nothing silently drops on the floor
Explore Data Extraction →
📨
HTML · PDF · Alerts
Reports · Notifications
Extracted data → distributed report
Every processed batch becomes a styled HTML report, a PDF attachment, an email to stakeholders, and a Telegram message with the document riding along.
  • Executive summary, rankings, and recommendations, auto-written
  • PDF conversion via html2pdf, saved to local storage
  • Email (Nodemailer) + Telegram bot, both with the PDF attached
  • Every send logged to MySQL for an audit trail
Explore Reports →
📱
React Native · Push
Mobile App
Reports and alerts, on site
The same dashboard data reaches a cross-platform app — overview metrics, discipline rankings, and alerts, with push notifications the moment a report lands.
  • Overview, Disciplines, and Alerts tabs
  • WebSocket live updates with offline-cache fallback
  • Push notifications for critical SPI/CPI breaches
  • Pull-to-refresh and configurable backend URL
Explore Mobile App →
🔄
n8n · Webhooks
Automation Workflow
The wiring that connects everything above
A 19-node n8n workflow — trigger, retrieve, OCR, classify, extract, analyze, hand off to the backend — deployable via Docker in minutes.
  • Webhook trigger accepts PDF or Excel on POST
  • Cross-discipline intelligence & ranking before handoff
  • Authenticated webhook call into the backend report engine
  • Docker Compose deployment, credentials, and test payloads included
Explore Automation →

— Shared Architecture

Why These Four Live Together

Nothing here is a standalone script. The n8n workflow, the backend's report/notification engine, and the two front ends (web dashboard and mobile app) all read and write the same webhook contract — a document uploaded once reaches a phone in someone's site trailer within about thirty seconds.

Input
PDF / Excel Upload
via n8n webhook trigger
⇄
Processing
OCR → Classify → Extract
GPT-4 classification + specialized extractors
⇄
Backend
Report + Notify
HTML/PDF report, email, Telegram, WebSocket broadcast
⇄
Output
Web + Mobile
Live dashboard and push notification

— Under the Hood

One Stack, Documented End to End

Every module below ships with its own setup guide, environment template, and quick-reference card — nothing here is a black box.

Backend Node.js, Express, WebSocket server, MySQL — authentication, RBAC, and rate limiting included.
Extraction Engine OpenAI GPT-4o-mini for classification, OCR Space for PDF text, three specialized code-based extractors for EVM, manpower, and cost.
Reports & Notifications HTML report generator, html2pdf conversion, Nodemailer for email, Telegram Bot API — all logged to a dedicated notifications schema.
Mobile React Native + Expo, AsyncStorage caching, WebSocket client, push notifications via Expo Notifications.
Automation n8n, Docker Compose deployment, webhook-authenticated handoff into the backend report engine.

Start Here

Follow the pipeline from the source

Start at Data Extraction to see how a raw PDF becomes a discipline score, or jump straight to Automation to see the workflow that ties it together.

Data Extraction → Automation Workflow →