Dynamite Docs : AI Data Extraction, PDF OCR, Image OCR, Browser to Excel

Capture pages, text & tables to Excel. Use hosted AI, your own models or local extraction with the Windows Ollama companion.

Capture what you need from a browser page, then review it in Dynamite Docs before extracting fields and spreadsheet rows. Connect your AI Models or use Hosted models. Run Document extractions on locally with the Windows Ollama companion. Choose a capture: - Capture visible area saves the part of the page or PDF currently on screen as a PNG. - Clip selected text saves the text you highlighted as a TXT file. - Copy a web table lets you choose a supported HTML table and save it as CSV. Preview the capture before sending it. Download it locally, or open it in your Dynamite Docs workspace. The workspace asks you to confirm import. Choose an extraction model, compare the proposed fields and rows with the source, correct values, then export to Excel, CSV or JSON. The extension is free to install. Dynamite Docs account, export and processing limits apply in the web app. A free account is available; paid plans offer higher limits. Hosted processing, your own provider key and eligible local Ollama processing use your existing workspace settings. The extension does not contain an AI provider key or bypass plan limits. The extension accesses the current page when you open it. It does not collect browsing history or run background page captures. Captures stay in browser memory until you choose to send or download them. A pending workspace transfer expires after 15 minutes and is removed after import or discard, when its receiving tab closes, or when Chrome restarts. Workspace imports check the file through our server, including browser-only imports. Cloud storage and AI processing follow your workspace choices. Visible-area capture does not download the original PDF or capture every page. Upload the original PDF in the workspace when you need the full document. Text and table access may be unavailable in protected pages or PDF viewers. Tables rendered in a canvas, cross-origin frames or unloaded virtual rows are not included. Supported table capture is limited to 1,000 rows and 1 MB of text per table, with up to 20 table choices. Review merged cells and column alignment after capture. Captures are limited to 4 MB.