Upload a PDF, ask questions, summarize, search by page, and get grounded answers with clickable citations.
Upload a text-based PDF and ask questions. The assistant indexes pages, answers from the document, and links back to sources.
Chat with any text-based PDF document. Upload contracts, reports, research papers, textbooks, school notes, or business documents, then ask questions in natural language. The assistant indexes extracted text by page, answers with citations, and lets you search, copy, or download a transcript.
Your PDF is processed in your browser using pdf.js. Selectable text is extracted and indexed locally by page. When the AI endpoint is available, the assistant can use it for richer answers; when it is unavailable or rate-limited, the local page-citation assistant still returns grounded extractive answers from your document.
Ask for summaries, extract key points, find dates and amounts, identify obligations or risks, search a term across pages, copy answers, download the transcript, or click a page citation to inspect the source page.
Your PDF file is parsed in your browser. If the AI endpoint is used, extracted text is sent for analysis, not the original file. The local fallback answers from extracted page text without needing the endpoint.
Text-based PDFs work best, including reports, contracts, research papers, and ebooks. Scanned or image-only PDFs may not extract text properly. For those, try our OCR tool first.
AI-backed responses may have daily limits. If the limit is reached, the local page-citation assistant remains available for summaries, search, details, and extractive answers.
PDFs up to 20 MB are supported. Large documents are indexed locally by page. The AI endpoint receives a capped context window, while local search can still work across all extracted pages.
Yes. The AI remembers your recent conversation, so you can ask follow-up questions naturally. The last 10 messages are included for context in each request.