Drive is the manual upload area: for files that did NOT come from WhatsApp or Telegram, but that you want inside the platform to consult later. PDF, spreadsheet, contract, slide, photo, video, audio, source code, e-book, certificate — any format works (up to ~100 MB per file). Drive keeps everything organized by folder, with unlimited subfolder hierarchy and free renaming. You can drag multiple files at once from your desktop into the browser — uploads happen in parallel with per-file progress indicators. Email attachments, browser downloads, exports from other tools: it all fits here.
Drive files automatically become Insight research material as soon as upload completes. You can ask questions in natural language and Insight returns the answer with a cited source from the right document, indicating the exact PDF page or spreadsheet cell. Scanned PDFs and images with text have content extracted via OCR (Tesseract with LLM post-processing), so even scanned contracts and whiteboard photos become searchable. Spreadsheets get structured extraction: Insight distinguishes header cells, rows, and formulas, so a question like 'what's the total of the March Revenue column' works without you opening Excel.
Share entire folders with a public link — anyone with the link accesses directly, no ZapRag login required. Useful for sending materials to clients, job candidates, or external collaborators. Configurable permissions: read-only or read+download. Search finds files by content, not just by file name. Look up by topic, specific phrase, or idea you remember — Drive understands semantic meaning, not just literal matching. 'That PDF about HIPAA compliance' finds it even if the file is named 'ComplianceGuide-v3.pdf'. Additional filters by type, upload date, and size refine the result.
Each uploaded file flows through an asynchronous pipeline: text extraction (PDF parser, OCR for images, Whisper for audio, FFmpeg+Whisper for video), splitting into 500-800 token chunks with overlap, embedding generation via OpenAI text-embedding-3, and indexing in Qdrant (vector store) with metadata (path, type, date, size). When you search, the system converts your query into the same vector space and returns the most relevant chunks, which become context for the LLM to answer. Files are encrypted at rest (AES-256-GCM with per-user key), and the indexing pipeline runs in isolated containers — your content is never used to train external models.
Legal teams maintain searchable contract libraries without opening 200 PDFs to find a clause. Consultants store reference materials (regulations, technical manuals) and query directly in natural language. Marketing teams archive past campaign briefs and recall by theme. Researchers index papers and run literature reviews via Insight. SMBs keep their 'cloud archive' (NDAs, certificates, receipts) organized and searchable. Educators store past lesson materials and recall exercises by subject. Anyone working with technical documentation stops spending 30 minutes hunting for the right passage. Tax/accounting professionals build a personal knowledge base of regulations and rulings.
Drive works alongside Insight (semantic search), Reporter (optional reference in briefings), Tasks (attach files to cards), and Sessions (send a Drive file straight to a WhatsApp conversation with 1 click). Unlike the Files directory (which holds media received from conversations), Drive is your personal library — controlled by you. Optional sync with Google Drive and Dropbox via OAuth: folders mirror automatically and indexing happens transparently. Version history kept for 30 days on all plans: you can revert an accidental upload. REST API available on Business plan for custom integrations with internal systems.
Q: How much storage do I get? — 50 GB on individual plans, 500 GB on Business, unlimited on Enterprise. Q: Do password-protected PDFs work? — No, you need to remove the password before upload (OCR limitation). Q: Can I edit files inside Drive? — No, it's repository + search only; to edit, download and re-upload the new version. Q: Does indexing slow down the upload? — No, the file is immediately available for download and viewing, indexing happens in background (~30s to 5min depending on size). Q: Is there automatic versioning? — Yes, uploading a file with the same name keeps the previous version in history.