🔬 Estimate construction costs from text with Telegram, OpenAI and DDC CWICR

⚡ 76 views · 🔬 Document Extraction & Analysis

💡 Pro Tip — HTTP Request scraping tends to break when sites update their markup. If you’re scraping a major platform, check if ScraperNode covers it — it has maintained scrapers for LinkedIn, Instagram, TikTok, YouTube, and 20+ other platforms that return structured data.

Description

A Telegram bot that converts natural-language work descriptions into detailed cost estimates using AI parsing, vector search, and the open-source DDC CWICR database with 55,000+ construction work items.

Who’s it for

Contractors & Estimators who need quick ballpark figures from verbal/text descriptions
Construction managers doing feasibility checks on-site via mobile
BIM/CAD professionals integrating text-based estimation into workflows
Developers building construction cost APIs or chatbots

What it does

Receives text messages in Telegram (work lists, specifications, notes)
Parses input with AI (OpenAI/Claude/Gemini) into structured work items
Searches DDC CWICR vector database via Qdrant for matching rates
Calculates costs with full breakdown (labor, materials, machines)
Exports results as HTML report, Excel, or PDF

Supports 9 languages: 🇩🇪 DE · 🇬🇧 EN · 🇷🇺 RU · 🇪🇸 ES · 🇫🇷 FR · 🇧🇷 PT · 🇨🇳 ZH · 🇦🇪 AR · 🇮🇳 HI

How it works

┌─────────────┐    ┌──────────────┐    ┌─────────────┐    ┌──────────────┐
│  Telegram   │ →  │  AI Parse    │ →  │  Embeddings │ →  │   Qdrant     │
│  Text Input │    │  (GPT/Claude)│    │  (OpenAI)   │    │   Search     │
└─────────────┘    └──────────────┘    └─────────────┘    └──────────────┘
                                                                  ↓
┌─────────────┐    ┌──────────────┐    ┌─────────────┐    ┌──────────────┐
│   Export    │ ←  │  Aggregate   │ ←  │  Calculate  │ ←  │  AI Rerank   │
│ HTML/XLS/PDF│    │   Results    │    │    Costs    │    │   Results    │
└─────────────┘    └──────────────┘    └─────────────┘    └──────────────┘

Step-by-step:

User sends /start → selects language → enters work description
AI Parse extracts work items: name, quantity, unit, room
Query Transform optimizes search terms for construction domain
Embeddings API converts query to vector (OpenAI text-embedding-3-small)
Qdrant Search finds top-10 matching rates from DDC CWICR
AI Rerank selects best match considering context and units
Calculate applies quantities, sums labor/materials/machines
Report sends Telegram message + optional Excel/PDF export

Prerequisites

Component	Requirement
n8n	v1.30+ (AI nodes support)
Telegram Bot	Token from @BotFather
OpenAI API	For embeddings + LLM parsing
Qdrant	Vector DB with DDC CWICR collections loaded
DDC CWICR Data	github.com/datadrivenconstruction/DDC-CWICR

Setup

1. Credentials (n8n Settings → Credentials)

OpenAI API — required for embeddings and text parsing
Anthropic API — optional, for Claude models
Google Gemini API — optional, for Gemini models

2. Configuration (🔑 TOKEN node)

bot_token     = YOUR_TELEGRAM_BOT_TOKEN
QDRANT_URL    = http://localhost:6333
QDRANT_API_KEY = (if using Qdrant Cloud)

3. Qdrant Setup

Load DDC CWICR collections for your target languages:

DE_construction_rates — German (STLB-Bau based)
EN_construction_rates — English
RU_construction_rates — Russian (GESN/FER based)
… (see DDC CWICR docs for all 9 languages)

4. Link AI Model Nodes

Open OpenAI Model nodes
Select your OpenAI credential
(Optional) Enable Claude/Gemini nodes for alternative models

5. Telegram Webhook

Activate workflow
Telegram Trigger auto-registers webhook
Test with /start in your bot

Features

Feature	Description
🤖 Multi-LLM	Swap between OpenAI, Claude, Gemini
🌍 9 Languages	Full UI + database localization
📝 Smart Parsing	Handles lists, tables, free-form text
🔍 Semantic Search	Vector similarity + AI reranking
📊 Cost Breakdown	Labor, materials, machines, hours
✏️ Inline Edit	Modify quantities, delete items
📤 Export	HTML report, Excel, PDF
💾 Session State	Multi-turn conversation support

Example Input/Output

Input (Telegram message):

Living room renovation:
- Laminate flooring 25 m²
- Wall painting 60 m²
- Ceiling plasterboard 25 m²
- 3 electrical outlets

Output:

✅ Estimate Ready — 4 items found

1. Laminate flooring ✓
   25 m² × €18.50 = €462.50
   └ Labor: €125 · Materials: €337.50

2. Wall painting ✓
   60 m² × €8.20 = €492.00
   └ Labor: €312 · Materials: €180

3. Ceiling plasterboard ✓
   25 m² × €32.00 = €800.00
   └ Labor: €425 · Materials: €375

4. Electrical outlets ✓
   3 pcs × €45.00 = €135.00
   └ Labor: €95 · Materials: €40

─────────────────────
Total: €1,889.50

[↓ Excel] [↓ PDF] [↻ Restart]

Notes & Tips

First run: Ensure Qdrant has DDC CWICR data loaded before testing
Rate accuracy: Results depend on query quality; AI reranking improves matching
Large lists: Bot handles 50+ items; progress shown per-item
Customization: Edit Config node for UI text, currencies, database mapping
Extend: Chain with your CRM, project management, or reporting tools

Author

DataDrivenConstruction.io
https://DataDrivenConstruction.io
info@datadrivenconstruction.io

Consulting & Training

We help construction, engineering, and technology firms implement:

Open data principles for construction
CAD/BIM processing automation
AI-powered estimation pipelines
ETL workflows for construction databases

Contact us to test with your data or adapt to your project requirements.

Resources

DDC CWICR Database: GitHub
Qdrant Setup Guide: qdrant.tech/documentation
n8n AI Nodes: docs.n8n.io/integrations/builtin/cluster-nodes/root-nodes/n8n-nodes-langchain

⭐ Star us on GitHub! github.com/datadrivenconstruction/DDC-CWICR

🔗 Nodes Used

HTTP Request, Telegram, Telegram Trigger, Basic LLM Chain, Anthropic Chat Model, OpenAI Chat Model

📥 Import

Download workflow.json and import into n8n: Workflow menu → Import from File

📖 Importing guide · 🔑 Credential setup