ConsulAI Swarm 🏗️🤖

An Advanced Multi-Agent System (Agent Swarm) for Automated Architectural and Engineering Project Review.

ConsulAI Swarm is an intelligent orchestration platform designed to automate Quality Assurance (QA), compliance checking, and technical review of architectural drawings, Building Information Models (BIM), and BOQs. By leveraging a multi-agent AI architecture, the system simulates a team of specialized senior engineers to detect violations, ensure building code compliance (e.g., ADIBC 2026), and provide a unified, critic-reviewed final decision.

---

🌟 Key Features

• Multi-Agent Orchestration: A central Manager Agent analyzes project context and dynamically delegates tasks to specialized engineering agents (Architectural, Structural, Electrical, Civil Defense, BOQ).
• Automated Code Compliance: Agents review plans against strict building codes and safety regulations, flagging violations instantly.
• RAG & Knowledge Base Integration: Employs Retrieval-Augmented Generation (RAG) using ChromaDB to ground AI decisions in actual project specifications, contracts, and engineering standards.
• Critic & Consensus Engine: A master Critic Agent reviews the outputs from all sub-agents to resolve contradictions, summarize safety flags, and output a final approval score.
• BIM & BOQ Analytics: Integrated modules (bim_manager.py, analytics_engine.py) to process and analyze complex data formats (like .dwg, .xlsx, and PDFs).

🧠 System Architecture

1. Manager Agent: Acts as the Lead Architect. It reads the initial document/drawing context, determines the project type and drawing stage, and outputs a JSON routing plan.
2. Specialized Agents:
   • 🏛️ Architectural Agent
   • 🏗️ Structural Agent
   • ⚡ Electrical Agent
   • 🚒 Civil Defense Agent
   • 💰 BOQ Agent
3. Critic Agent: Acts as the QA Director. It consumes all specialized agent reports, evaluates them for contradictions, and generates a final JSON report containing a unified ProbabilityScore, Violations list, and approved boolean.

🛠️ Tech Stack

• Python 3.x
• LLM Engine: Google Gemini (via google-genai), orchestrated for structured JSON outputs.
• RAG: ChromaDB for vector storage and semantic retrieval.
• Data Processing: Custom pipelines for PDF parsing, Excel (BOQ) extraction, and CAD metadata handling.

🚀 Getting Started

Prerequisites

1. Python 3.10+
2. A valid Gemini API Key (or other supported LLM API keys).

Installation

1. Clone the repository:

   git clone https://github.com/Roduan25/ConsulAI-Swarm.git
   cd ConsulAI-Swarm

2. Install dependencies:

   pip install -r requirements.txt

3. Environment Setup: Create a .env file in the root directory and add your API keys:

   GEMINI_API_KEY=your_api_key_here

Usage

Run the main application:

python app.py

📂 Project Structure

• app.py: Main application entry point.
• orchestrator.py: Contains the AgentSwarm, Manager Agent, and Critic Agent logic.
• llm_router.py: Handles dynamic routing of prompts to the appropriate LLM models.
• rag_engine.py & build_rag.py: Scripts for ingesting documents into ChromaDB and running semantic search.
• bim_manager.py: Utilities for interacting with Building Information Models.
• analytics_engine.py: Processes and analyzes project metrics and BOQ files.
• knowledge_base/: Directory containing source documents for RAG.

🤝 Contribution

Contributions, issues, and feature requests are welcome! Feel free to check the issues page.

---

Built to reflect advanced expertise in building robust, production-ready AI Agents for the Architecture and Engineering sector.