Do we really need a vector database for every AI documentation chatbot?

Been seeing a lot of discussions around RAG, embeddings, and vector databases lately.
That made me wonder, do we really need a vector database for every AI documentation chatbot?

So I explored Google's Open Knowledge Format (OKF) and built Ask-Your-Markdown to better understand where this approach fits. I also wrote a blog sharing what I learned. Both the article and the GitHub repository are mentioned below.

If you've been curious about OKF, I hope this helps.

imelanthirayan / Ask-Your-Markdown

Ask questions about a folder of Markdown files using AI or offline search. Built on Google's Open Knowledge Format (OKF).

📚 Ask Your Markdown

Ask questions about a folder of markdown files - and get answers grounded in your documents.

No database. No embeddings. No vendor lock-in. Just your Markdown files, organized with OKF and ready to browse - with or without an LLM.

Built on the Open Knowledge Format (OKF) - an open, vendor-neutral standard from Google Cloud.

Works with OpenAI · Azure AI Foundry · or fully offline.

---

🧭 Table of contents

• Why this exists
• How it works
• How this compares to classic RAG
• The two chapters: Build → Ask
• What's in the box
• Quick start
• Chapter 1 - Build a bundle
• Chapter 2 - Ask questions
• Labels: the fast path
• Big documents
• Try it with no API key
• Why people like this
• What's OKF vs. this sample
• How well does it work - honestly

---

💡 Why this exists

Imagine a drawer full of notes -…

View on GitHub