ConstructAI

AI-powered construction project control assistant combining Civil Engineering project controls, Model Context Protocol (MCP), SQLAlchemy, and a local LLM.

🚧 Overview

ConstructAI is an AI assistant designed for construction project control.

It allows users to ask project-control questions in natural language and uses MCP tools to retrieve project data, analyze schedules, identify risks, evaluate delays, and recommend recovery actions.

Key Capabilities

• Analyze project schedule status

• Identify critical activities

• Find activities by name

• Analyze construction delays

• Determine delay impact on project completion

• Identify delayed activities

• Analyze activity dependencies

• Identify resource bottlenecks

• Analyze project risks

• Generate schedule recovery strategies

• Provide project-control summaries

Related MCP server: Unanet GovCon ERP MCP Server

🏗️ Architecture

User
  │
  ▼
Natural Language Query
  │
  ▼
Qwen3 8B (Local via Ollama)
  │
  ▼
MCP Tool Calling
  │
  ▼
ConstructAI MCP Server
  │
  ├── Schedule & CPM
  ├── Delay Analysis
  ├── Risk Analysis
  ├── Resource Analysis
  └── Schedule Recovery
  │
  ▼
SQLAlchemy
  │
  ▼
Project Database
  │
  ▼
Engineering Analysis
  │
  ▼
AI Recommendation

🤖 Technology Stack

• Python

• MCP 2.1.1

• Qwen3 8B

• Ollama

• SQLAlchemy

• SQLite

• Natural-language tool calling

The LLM runs locally through Ollama, so the current system does not require a paid AI API.

🔧 MCP Tools

ConstructAI currently provides 11 MCP tools:

📊 Project Control Capabilities

Schedule Management

ConstructAI can analyze:

• Planned start and finish dates

• Activity durations

• Early and late dates

• Total float

• Critical activities

• Activity dependencies

• Project completion dates

Delay Analysis

The system can determine:

• Delayed activity

• Delay duration

• Original project finish

• New project finish

• Project completion impact

• Affected downstream activities

• Delay priority

• Recommended corrective action

Schedule Recovery

ConstructAI can recommend strategies such as:

• Expediting material procurement

• Increasing manpower

• Increasing working hours

• Prioritizing critical activities

• Coordinating successor activities

• Controlled activity overlap

🧠 Example AI Workflow

User Question

Electrical Rough-in is delayed by 3 days because of a material shortage. What is the impact on Project 3 and how can we recover the schedule?

Tool Execution

User Question
     │
     ▼
find_activity_tool
     │
     ▼
Find Electrical Rough-in → Activity ID = 33
     │
     ▼
analyze_delay
     │
     ▼
Calculate Delay Impact
     │
     ▼
schedule_recovery
     │
     ▼
Generate Recovery Strategy
     │
     ▼
Engineering Recommendation

Result

Original Project Finish : 2026-11-06
New Project Finish      : 2026-11-09
Project Impact          : 3 days
Priority                : CRITICAL

Recovery Actions:
1. Expedite material procurement
2. Increase manpower
3. Increase working hours
4. Prioritize critical-path activity
5. Coordinate successor activities

🏢 Demo Project

The project includes a sample construction schedule: ConstructAI Building Control Demo

The schedule contains activities covering:

Site Mobilization · Excavation · PCC · Footing Reinforcement · Footing Concrete · Pedestal Reinforcement · Pedestal Concrete · Column Reinforcement · Column Concrete · Slab Reinforcement · Slab Concrete · Masonry · Electrical Rough-in · Plumbing Rough-in · Plastering · Flooring · False Ceiling · Painting · Final MEP Installation · Testing & Commissioning · Final Inspection & Handover

🧪 Testing

The project contains test files for major project-control modules and MCP functionality, including:

• test_cpm.py

• test_delay.py

• test_delay_impact.py

• test_project_status.py

• test_project_risk.py

• test_schedule_recovery.py

• test_resource_bottleneck.py

• test_project_control_summary.py

📁 Project Structure

ConstructAI/
│
├── ai_client.py
├── server.py
│
├── database.py
├── models.py
│
├── cpm.py
├── schedule.py
├── schedule_variance.py
├── forecast.py
│
├── activity_details.py
├── activity_dependencies.py
├── critical_activities.py
├── delayed_activities.py
├── find_activity.py
│
├── delay_analysis.py
├── delay_impact.py
│
├── recovery.py
├── recovery_impact.py
├── schedule_recovery.py
│
├── resource_bottleneck.py
├── project_risk_analysis.py
├── project_control.py
├── project_control_summary.py
├── project_status.py
├── recommendations.py
│
├── requirements.txt
├── README.md
└── .gitignore

🚀 Installation

1. Clone the repository

git clone https://github.com/bilalgit/ConstructAI.git
cd ConstructAI

2. Create a virtual environment

python -m venv venv

Activate it on Windows:

venv\Scripts\activate

3. Install dependencies

pip install -r requirements.txt

4. Install Ollama

Install Ollama and download the Qwen3 model:

ollama pull qwen3:8b

5. Run ConstructAI

Start the MCP server and AI client according to the project configuration.

💰 Cost

No paid AI inference API is required for the current setup.

🎯 Project Objective

The objective of ConstructAI is to explore how Artificial Intelligence and Model Context Protocol can be applied to construction project management and project controls.

The project combines:

• Civil Engineering

• Construction Planning

• Critical Path Method

• Schedule Management

• Delay Analysis

• Risk Management

• Resource Management

• Artificial Intelligence

• MCP Tool Calling

• Local LLMs

• Database-driven project controls

🔮 Future Development

Potential future improvements include:

• Construction cost control

• Quantity tracking

• Earned Value Management

• Resource leveling

• Progress forecasting

• Automated daily progress reports

• AI-generated weekly and monthly reports

• BIM integration

• Primavera P6 integration

• Power BI dashboards

• Construction document intelligence

• Multi-project portfolio monitoring

👨‍💻 Author

Bilal

Civil Engineering | Construction Project Controls | AI Engineering

⭐ If you find this project interesting, feel free to explore the repository.