$ cat overview.txt
Overview
AiAssist is a Retrieval-Augmented Generation (RAG) chatbot that syncs with a company's knowledge base, creates vector embeddings, and generates AI-powered answers with source citations. It includes an expert review workflow where corrected answers become ground truth for future queries, making the system smarter over time.
$ cat challenge.txt
The problem
Support teams spend hours answering the same questions repeatedly, and knowledge base articles go unread. We wanted to build a chatbot that could actually answer questions accurately from existing documentation and get smarter over time as subject matter experts correct its responses.
$ cat solution.txt
The approach
We built a RAG pipeline that syncs knowledge base articles on a schedule, chunks them into ~400-token segments with semantic overlap, and generates vector embeddings using OpenAI's text-embedding-3-large model. When a user asks a question, the system performs cosine similarity search to find relevant context, then generates an answer with GPT-4o, always citing the source articles.
The key differentiator is the ground truth system. When an expert reviews a conversation and corrects an answer, that correction is embedded and stored as ground truth. Future queries that match closely (>0.95 similarity) return the expert-verified answer directly, creating a feedback loop that continuously improves accuracy.
$ ls features/
What shipped
- +Automatic knowledge base sync (every 3 hours)
- +Text chunking with semantic overlap (~400 tokens)
- +Vector embeddings via OpenAI text-embedding-3-large
- +Semantic similarity search for relevant context
- +GPT-4o answer generation with source citations
- +Ground truth system for expert-corrected answers
- +Admin dashboard with review queue and statistics
- +AI Training Mode for subject matter experts
- +Conversation logging and audit trail
- +Data export capabilities