About chinnamAI
A public agentic AI tool that turns any AI/ML topic into a grounded, sourced article. A multi-agent system does the work live in your browser — Research, Drafter, Critic, Verify — and hands you a downloadable PDF. Nothing is saved on either side.
What it does
You type a topic — "How transformer self-attention works", "Diffusion models for video", "Mixture-of-experts routing", anything in AI/ML. A multi-agent AI system goes to work: a Research agent searches a corpus of research papers, a Drafter writes an article from what was found (calling a verification tool mid-write to ground specific claims), a Critic audits the draft for unsupported claims, and a Verify step decides whether to accept the article or send it back for a rewrite. When the loop finishes, you get the article — to read in your browser or download as a PDF.
This is what an agentic AI system looks like in practice: agents that use tools, talk to each other through shared state, and self-correct when something isn't right — not a single language model giving a one-shot answer.
Every generation is independent. Nothing about your topic, your visit, or the article is stored. When you close the tab, it's gone.
How it feels in practice
A topic, a sentence, a question — anything in AI/ML.
It queries a corpus of indexed research papers, picks the most relevant chunks, captures notes for the drafter.
A full article with sections, inline citations, embedded figures where they help, and a References section at the end. You watch it stream in live.
Goes claim-by-claim against the sources. If too many things aren't grounded, the drafter rewrites with the critic's notes — up to a few times.
Markdown or PDF. Yours to keep, share, or build on.
Why it exists
Most AI article generators hallucinate confidently and don't show you where their claims come from. chinnamAI is the opposite: every substantive claim is traced back to a real research paper, the agents argue with each other when something is unsupported, and the final article either stands up to that scrutiny or is honest about where it didn't quite.
It's also a demonstration. The whole pipeline is a working agentic AI system — the kind of architecture that's becoming standard in production AI applications. Type a topic and you can watch it run.
Who built it
chinnamAI is built and maintained by Chinnam — an engineer with a background in distributed systems and AI engineering.
It uses Retrieval-Augmented Generation (RAG) to ground every article in real research papers, not in the model's training data. The agents search an indexed corpus of arxiv papers, retrieve the most relevant passages, and write articles that cite those sources directly. The stack is small and modern: Next.js for the web app, Postgres with vector search for retrieval, and a major large language model provider for the agents themselves.