Free RAG online: convert files without an account
Free RAG online means converting documents to chunks and embeddings in the browser without creating an account - a free hosted tier that still takes the file is a different product wearing the same words.
The phrase is doing two jobs at once
Some people searching 'free RAG online' want a chatbot over their files. Some want the conversion step: files in, a corpus out, so they can build the chatbot themselves. Those are not the same tool. A chatbot that is free still has to store your documents to answer later. A converter that is free can finish in the tab and forget you. RAG Converter is the second job. If you needed the first, you still need somewhere to put the JSONL afterwards.
No account is the proof it is not keeping the files
An account exists to attach a corpus to a person. If a site will not convert until you sign in, it is planning to remember the work. That can be the right product for a team. It is the wrong one for a confidential PDF you wanted searchable by tonight. The free path here starts without a login. Sign-in unlocks larger on-device models and the API, not the permission to convert a Word file.
What a free session is enough for
A folder of contracts, a handful of decks, a text-layer PDF library, notes and Markdown. That is most of what people mean by 'I just want to try RAG on my files'. It is also most of what a browser can do well. You get keyword or optional semantic vectors, overlapping chunks, and a manifest. You do not get a hosted index, a query endpoint, or speech off a two-hour recording. Those are paid, and they say so.
Free online still has to be honest about scans
A photograph of a page has no text layer. A free converter that indexes it as an empty template makes the corpus worse. The right free behaviour is to skip it and name it. RAG Converter does that on the open page; the signed-in app can OCR the same file still on your machine. 'Free RAG online' that silently swallows scans is how people decide RAG does not work.