Bilingual RAG مساعد المستندات

Loading the multilingual embedding model (about 120 MB, saved by your browser after the first visit)…

Ask your documents, in Arabic or English.

Search runs in this tab. Every answer cites the passages it came from and shows each step that produced it.

Using your own files

Open Documents and drop in a PDF, TXT or Markdown file. It is read, split into passages and embedded on your device, then you can ask about it here. Five sample documents of a fictional company, Falak Home Systems, are loaded to start with.

Write answers with an LLM

Retrieval stays the same. The top five passages and your question go to the model with the same prompt the Python service uses, and it writes a short answer with [n] citations. Without a key, the answer is the best one or two sentences quoted from the passages.

Provider

Sent only to the provider you chose, directly from this tab. Kept for this tab session unless you tick the box below.

Passage