Extract tables from PDF documents as CSV data.
Extract tables from your PDF using AI vision. Pages are rendered locally in your browser and sent securely to Google Gemini for table detection. No data is stored. Your CSV is ready to download instantly.
Free users limited to 10MB files. Upgrade for 100MB, batch & no wait
Convert PDF tables to CSV online for free with our AI-powered PDF to Excel tool. Upload a PDF containing tabular data — invoices, financial statements, reports, or spreadsheets — and our tool extracts all tables into clean CSV format using Google Gemini vision AI. Unlike traditional PDF table extractors that rely on text parsing, our approach uses vision AI to understand the layout and structure of every table on each page. This means it works even with scanned PDFs, image-based tables, and complex multi-column layouts. Each table is separated by a blank line in the output, and headers are preserved exactly as they appear. You can copy the CSV directly from the textarea or download it as a .csv file for use in Excel, Google Sheets, or any spreadsheet application. Your data is processed securely — pages are rendered locally in your browser and only the anonymized page images are sent to Gemini for extraction. No files are stored on our servers.
Tables locked inside a PDF are easy to read and nearly impossible to work with. Every total you want to check, column you want to sort, or chart you want to build has to wait while someone re-types the numbers. This tool reads the tables on your pages and hands you the data as a spreadsheet file you can open, copy, or edit anywhere. It is free to use, works right in the browser, and relies on AI vision to figure out where each table begins and what belongs in every cell. No plugins and no formulas to write — the data is ready the moment the spreadsheet opens.
The extractor looks at every page, finds the tabular regions, and returns their contents as clean CSV. Each table is separated by a blank line in the output, and column headers are kept exactly as they appear on the page, so nothing gets renamed along the way.
The CSV format keeps things simple on purpose. Every cell is a real value, so the moment the file is open in a spreadsheet, SUM, filters, and charts behave the way you expect without any extra formatting steps.
Pages are rendered locally in your browser, then the anonymized page images are sent to a vision AI service that reads the layout and returns the table structure. That is what allows the tool to recognize real tables wherever they appear, including on scans where there is no text layer to lean on. The PDF file itself is never stored, and you keep full ownership of the result.
You can also select everything in the preview and paste it straight into an existing sheet. That works well when a table from the PDF is just one block of a larger workbook you are already building.
Tables with clear rows and columns extract the most reliably. If a table is rotated on the page, or a page is split down the middle by a break, fixing the PDF first saves cleanup time later. After extraction, glance at a few cells — especially totals and dates — before building anything on top of the data.
When a table stretches across several pages, expect the pages to be handled in order. If a row is cut by a page break, the cleanest fix is to adjust the source document first and then extract again.
Pro tip
Spot-check the numbers that matter most: totals, dates, and the first and last rows. AI extraction is fast and accurate on clean tables, but thirty seconds of verification beats rebuilding a bad spreadsheet later.
The tool is free for everyday use, with a daily allowance and a 10 MB file ceiling. Premium removes the wait, raises the limit to 100 MB, and adds batch processing for people who regularly work through piles of reports. Either way, the pipeline is the same: render locally, extract with vision AI, download your spreadsheet. And on whichever tier you are on, the files you work with remain your own — the tool is a convenience, not a vault.