Upload any PDF. Set a page range. Get answers drawn exclusively from those pages — exact page numbers cited, follow-up questions supported, zero hallucinations from outside your upload.
Standard AI chat has a fundamental flaw for academic work — it answers from its entire training data, which often conflicts with your textbook's definitions, notation, or framing. Study Mode fixes this by pinning the AI exclusively to your uploaded document. Every answer is grounded in your material, not the internet. And unlike a basic chatbot, Study Mode carries your conversation forward — follow-up questions like 'explain that differently' or 'give me an example from page 40' work exactly as you'd expect, because the AI remembers what you just discussed.
Drag and drop any PDF — textbook chapter, lecture slides, research paper, past exam, or technical manual. LumiChats extracts every page and indexes the content for semantic retrieval. Free accounts cover up to ~50 pages with full semantic search; paid accounts up to ~150 pages. A notice appears if only part of your document is indexed so you always know your coverage.
Tell the AI which pages to focus on. 'Pages 30 to 60' means every answer is drawn exclusively from those pages — the AI cannot pull content from sections you haven't assigned. Perfect for chapter-by-chapter study without cross-contamination from other parts of the book.
Type questions in plain language. The AI retrieves the most relevant passages from your document and answers using only that content, citing the exact page number. Crucially, follow-up questions work: 'explain that with an example', 'how does this connect to what you said earlier', or 'go deeper on point 2' — the AI carries the last two exchanges into each new question, so study sessions flow like a conversation with a tutor, not a series of isolated lookups.
Once you've studied a section, switch to Quiz Hub. Generate a multiple-choice quiz on the same page range using the same document. Active recall on exactly your material — not a generic internet version of the topic. If you exit mid-quiz, the partial attempt is saved and labelled in your history so your score record stays accurate.
At the end of your session, ask LumiChats to generate structured notes from everything you covered. Get a clean, revision-ready summary — headings, key points, definitions, formulas — in seconds. Save it, share it, or use it as the basis for your next session.
Upload your course textbook, module notes, or lecture pack. Study unit by unit without the AI pulling content from sections you haven't covered. Works for any university, any subject, any country — from a first-year biology module in the UK to a final-year engineering paper in India or the US.
Upload a 60-page journal article. Ask LumiChats to explain the methodology, summarise the argument, identify the research gap, or compare two sections — all cited to the exact page. No hallucinated references, no invented citations.
Upload your official study material — whether it's an NCERT textbook for JEE/NEET, an A-Level syllabus document, an SAT prep guide, a GMAT official guide, or a Bar exam outline. Every answer is sourced to the exact page of your upload, not a generic internet summary of the topic.
Export your lecturer's slides or course pack as a PDF. Ask questions about specific slides or topics. The AI stays inside those slides — it cannot invent content that isn't in the presentation.
Upload API documentation, engineering specifications, legal texts, or medical guidelines. Study Mode works on any dense technical PDF — ask plain-language questions and get precise, page-cited answers from the source document.
Technical Note
Study Mode uses LumiChats' RAG (Retrieval Augmented Generation) pipeline. Uploaded documents are split into 4,000-character chunks with overlap, embedded using a text embedding model, and stored in pgvector. When you ask a question, your query is embedded and cosine similarity search retrieves the most semantically relevant chunks — grounding every answer in your document, not the model's training data. Conversation history: the last four messages (two exchanges) are passed alongside the retrieved context so follow-up questions have continuity without bloating the prompt. Chunk indexing is tier-aware: free accounts index up to 150 chunks (~50 pages); paid accounts up to 500 chunks (~150 pages). When a document exceeds the plan limit, a partial-index notice is displayed and keyword search covers the remaining sections.
5 free messages every day. Activate a ₹69 Day Pass whenever your workload picks up.
Create Free Account