Chat with the Articles — Grounded RAG Assistant
A live, gated demo: ask questions about Vishal’s published articles and get answers grounded strictly in them, with citations linking back to each post. The same RAG pattern the blog explains, running live. LLM-agnostic; this instance uses Gemini.
Overview
Chat with the Articles is a working retrieval-augmented generation (RAG) assistant over Vishal’s own published writing. When a question is asked, the app retrieves the most relevant articles, passes those excerpts to the model as grounding context, and answers strictly from them — citing each source with a link back to the full post. If the corpus does not cover a question, it says so rather than inventing an answer. This is the exact pattern described in the blog (retrieve → ground → cite), demonstrated end-to-end. The architecture is LLM-agnostic — retrieval and grounding work the same against a local model (Ollama / Llama) or any cloud LLM; this deployment uses Google Gemini.
Problem Solved
Generic chatbots hallucinate and cannot point to a trustworthy source. This demo shows grounded, citable answers drawn only from a known body of content — the difference between a plausible-sounding guess and an answer you can verify against the original article.
Capabilities
- Retrieval over the published article corpus
- Answers grounded strictly in retrieved content
- Inline source citations linking to each article
- Refuses to answer when the corpus does not cover the question
- Conversational multi-question thread
- Pluggable LLM backend — Ollama / Llama (local) or cloud LLM
- Gated access and rate limiting