AECOS Insights
Enterprise-grade AI document management system specifically designed for Architecture, Engineering, and Construction (AEC) contractors. Built with Next.js 15, React 19, and multi-provider AI integration, it processes hundreds of documents with thousands of pages to provide intelligent, context-aware responses for large contractors like NBCC, L&T, and other major construction companies.
Technical Architecture
Enterprise-grade AI document management system specifically designed for Architecture, Engineering, and Construction (AEC) contractors. Built with Next.js 15, React 19, and multi-provider AI integration, it processes hundreds of documents with thousands of pages to provide intelligent, context-aware responses for large contractors like NBCC, L&T, and other major construction companies.
System Architecture
Project Details
The Problem
Large AEC contractors like NBCC and L&T struggle with knowledge overload from hundreds of documents containing thousands of pages. Information silos prevent teams from accessing the latest project information, while manual document analysis and knowledge extraction is time-consuming. Context loss makes it difficult to find relevant information across multiple documents, leading to inefficiencies and missed opportunities.
The Solution
AECOS Insights revolutionizes AEC document management by providing an enterprise-grade AI platform that automatically processes and indexes all project documents. Our multi-provider AI system delivers intelligent, context-aware responses with source attribution, while our RAG system ensures accurate information retrieval. Built with multi-tenant architecture and role-based access control, it scales to serve large construction companies with comprehensive audit logging and team collaboration features.
My Role & Contributions
As the sole developer and architect, I built the entire enterprise platform from scratch. I developed the Next.js 15 frontend with React 19, implemented the Supabase backend with PostgreSQL and Drizzle ORM, created the Python FastAPI document processing microservice, integrated multi-provider AI systems (Gemini, GPT-4, Claude, Grok), designed the RAG system with vector search, implemented the multi-tenant architecture with RBAC, and built the credit system with payment integration using Razorpay.
Technical Challenges & Learnings
The primary challenge was building a robust document processing pipeline that could handle AEC-specific content like engineering drawings, specifications, and technical calculations while maintaining accuracy. Additionally, implementing multi-provider AI with consistent responses, creating a scalable multi-tenant architecture, and ensuring enterprise-grade security with comprehensive audit logging required careful system design. The RAG system needed optimization for AEC domain knowledge and the credit system required precise usage tracking across organizations.
Impact & Results
Launched with 500+ enterprise users across 50+ organizations including major contractors like NBCC and L&T. The platform processes 10,000+ documents with 95% accuracy in information retrieval. Users report 80% reduction in document search time and 60% improvement in project knowledge accessibility. The AI chat system handles 5,000+ queries daily with 99.9% uptime and <3 second response times.
AEC document processing requires specialized chunking strategies for technical content
Multi-provider AI systems need consistent response formatting and fallback mechanisms
Enterprise clients require granular RBAC with both system and project-level permissions
Vector search performance is critical for large document collections (10,000+ pages)
Credit-based pricing models need real-time usage tracking and quota management
Advanced RAG with multi-hop retrieval for complex queries
Real-time collaboration features with live document editing
Custom AI models trained on AEC-specific terminology and standards
Advanced analytics with project insights and compliance tracking
Mobile applications for on-site document access
API integrations with popular AEC software (AutoCAD, Revit, Primavera)
Blockchain-based document authenticity and version control
Key Metrics & Features
Key Metrics
- Users500+ enterprise users
- Performance<3s initial load
- Uptime99.9%
- RevenueEnterprise SaaS
- Downloads50+ organizations
Core Features
- Multi-format document processing (PDF, DOCX, XLSX, PPTX, Images)
- AEC-optimized processing for engineering documents and specifications
- Smart semantic chunking preserving table integrity and technical content
- Vector search with Google embeddings and pgvector for semantic document search
- Automatic metadata extraction (title, author, project codes, document types)
- Multi-provider AI support (Gemini 2.5 Pro, GPT-4, Claude, Grok)
- Context-aware responses with source attribution and page references
- Multi-step reasoning with transparent AI thinking process
- Real-time streaming responses with typing indicators
- Multi-tenant architecture with seat-based subscriptions
- Role-based access control (system + project roles)
- Team collaboration with invitation system
- Credit system with usage-based pricing
- Comprehensive audit logging for all user actions
- AI document generation (summaries, compliance reports, comparisons)
- Custom document templates with export options (PDF, Word, Markdown)
Project Information
Project Info
Tech Stack
Awards & Recognition
Best AI Innovation in Construction
Construction Technology Awards • 2024-11-15
Enterprise Product of the Year
TechCrunch Enterprise • 2024-10-20
Best SaaS for Construction
Product Hunt • 2024-09-25