Preparing the workspace…
Index exactly one text-based PDF locally and ask extractive questions answered with exact page-cited quotes.
This method deterministically matches your question to embedded, selectable PDF text. It does not use a generative model or independently verify the document.
The source PDF is discarded after local indexing. Session tokens are kept only in this tab.
The controls stay out of the way until you need them.
Choose from your device or drag files straight into the secure workspace.
Use the recommended defaults or tailor the result to your exact job.
Review the result and save it before the private download expires.
The tool currently requires no account or payment. Each file and the combined upload can be up to 90 MB; availability and limits may change.
Files are encrypted during transfer, processed in an isolated job and scheduled for deletion at the displayed expiry. We do not use document content to train AI models.
Chat with PDF accepts PDF files. Each file and the combined upload can be up to 90 MB.
The tool is configured to preserve quality wherever the format allows. If a setting can affect fidelity or file size, you can choose the balance before processing.