Version 1.0 — An AI-powered Retrieval-Augmented Generation (RAG) system that enables businesses to chat with their internal documents using natural language.
Most companies store valuable business knowledge inside PDF documents.
These documents usually contain:
- Services
- Pricing
- Internal documentation
- Company policies
- FAQs
- Product documentation
- Employee knowledge
As businesses grow, searching through dozens of PDF files becomes slow and inefficient.
Support teams spend unnecessary time searching for information instead of helping customers.
This results in:
- Slow customer support
- Inconsistent answers
- Reduced productivity
- Repeated manual work
- Difficult employee onboarding
- Poor knowledge accessibility
Traditional keyword search is also limited because users rarely know the exact wording inside documents.
This solution transforms static company documents into an intelligent AI assistant capable of retrieving accurate business information in seconds.
Instead of searching through multiple files, employees simply ask questions in natural language and receive grounded answers directly from company documentation.
The result is:
- Faster customer support
- Better employee productivity
- Consistent answers
- Reduced manual searching
- Easier access to company knowledge
The system automatically creates an AI-powered knowledge base by:
- Searching company PDF files inside Google Drive.
- Downloading and extracting document content.
- Splitting documents into semantic chunks.
- Generating vector embeddings.
- Storing embeddings inside Supabase Vector Store.
- Retrieving relevant information using semantic similarity search.
- Allowing an AI Agent to answer questions based only on retrieved company knowledge.
Unlike a traditional chatbot, the assistant does not rely on general AI knowledge.
Instead, every answer is generated using the retrieved business documentation.
How much does SmartSoft ERP development cost?
According to the SmartSoft knowledge base,
ERP System Development starts from $5,000.
- AI-powered document search
- Retrieval-Augmented Generation (RAG)
- Semantic search using embeddings
- Google Drive integration
- Automatic PDF extraction
- Supabase Vector Database
- Hugging Face Embeddings
- OpenRouter AI Agent
- Multi-document retrieval
- Conversation memory
- Accurate document-grounded responses
| Technology | Purpose |
|---|---|
| n8n | Workflow Automation |
| Supabase | Vector Database |
| pgvector | Semantic Search |
| Hugging Face | Embedding Generation |
| Google Drive | Document Storage |
| OpenRouter | Language Model |
| AI Agent | Tool Calling |
| RAG | Knowledge Retrieval |
- Search PDF files
- Download documents
- Extract text
- Split into chunks
- Generate embeddings
- Store vectors
- Receive user question
- Generate query embedding
- Search vector database
- Retrieve relevant chunks
- Generate grounded answer
- Automatic document synchronization
- Detect newly uploaded documents
- Update changed documents automatically
- Duplicate detection
- Metadata filtering
- Multiple knowledge bases
- DOCX support
- Website indexing
- Production deployment
Adel Sheded
AI Automation Developer
Specialized in:
- AI Agents
- n8n Automation
- RAG Systems
- API Integrations
- Workflow Automation
This repository includes:
- Project documentation
- System architecture
- Workflow overview
- Demo screenshots
The complete n8n workflow files are intentionally not included. They are available upon request for clients or during project discussions.
MIT License
Note: This project is intended for portfolio and educational purposes. All API keys, credentials, and sensitive configuration have been removed. The included knowledge base documents are sample data created for demonstration only.