RAG O Matic
You want an AI that can answer questions about your own documents.
The problem is that most language models do not know your data, and they cannot access it on their own. Their knowledge stops at what they were trained on, which means they miss anything new, private, or specific to your organization.
Retrieval Augmented Generation solves that. It connects an AI model to your documents so it can search, understand, and respond using real information instead of guesses.
Why I built this: We know RAG can solve the problem of connecting AI to your documents, but the question is: how does it actually work? Most explanations show you diagrams or tell you about chunking and embeddings, but they don't let you see it happening live. I wanted to create something that shows you exactly what happens under the hood, step by step, with real-time cost tracking so you understand not just what RAG does, but how.
What you'll learn: You'll see how we break documents into chunks, convert them to vectors that capture meaning, find the most relevant pieces for any question, and use AI to generate answers with proper citations.
How to use this: Scroll down to explore each step of the RAG pipeline. Each section explains what's happening and why it matters. Click the buttons to interact with each step. It's designed to be simple and educational - no technical jargon required!
Step 1: Input Your Documents
Choose from our sample document sets or add your own documents below. Each text box has a 10KB limit, and the total size is capped at 30KB.
📚 Choose Your Document Set
Pick one of the document sets below to get started! Each set contains different types of content to explore how RAG works.
Step 2: Chunk Your Documents
Break your documents into smaller chunks for better retrieval. Choose a chunk size below and click "Chunk It" to see how your text gets divided.
Choose Chunk Size
Smaller chunks = Less context, more precise matches
Larger chunks = More context, better understanding
We recommend larger chunks for better RAG performance!
Step 3: Generate Embeddings
Convert your text chunks into vector embeddings using OpenAI's text-embedding-3-small model. This allows us to find semantically similar content later.
text-embedding-3-small modelStep 4: Retrieve Similar Chunks
Ask a question and see which chunks are most similar to your query. The system uses cosine similarity to find the top 5 most relevant chunks.
💡 Try these questions:
Click any question below to automatically search and move to the next step
Step 5: Generate Answer
Ready to generate your answer! The AI will use the retrieved chunks to create a contextual response with citations.