01
Infer the structure
The model separates document fields, repeating rows, totals, and notes based on the file in front of it.
How does the PDF table extractor work?
Dynamite Docs finds tables in digital or scanned PDFs, rebuilds their rows and columns, and lets you review each table before exporting it to Excel, CSV, or JSON.
What to expect
Multi-page tables still need checks for repeated headers, wrapped descriptions, merged cells, blank values, and rows split across page boundaries.
The full workflow
The output stays beside the evidence.
Check the proposed fields against the source, correct a value, then export the reviewed structured PDF tables.
Dynamite Docs- 01Select the source file
- 02Review fields and rows
- 03Download structured PDF tables
02
Correct against the source
The source remains visible while you edit cells, rename columns, inspect warnings, and mark the result reviewed.
03
Send useful data onward with Dynamite Docs
Download structured PDF tables, or move a recurring workflow into Google Sheets, Drive, webhooks, or the API.
Dynamite Docs- Table title
- Transaction detail
- Column headers
- Date, description, debit, credit
- Row values
- 28 Aug 2026, Transfer, 420.00
Choose the AI route
Bring a free provider key. Keep the review workflow.
Open AI & Processing, choose Connect Provider, and pick a provider. The setup screen links to that provider's key page. Create a free key where the provider offers one, paste it into Dynamite Docs, verify it, then select a model for the run.
Your provider key is encrypted before it is stored.
Switch between hosted, own-key, privacy-restricted, and local models.
The same source-first review screen works whichever route you choose.
Choose the page for the job using Dynamite Docs
Choose the PDF Table Extractor when the main challenge is recovering tabular structure: identifying where a table begins, which headers govern each column, how wrapped descriptions stay within a single row, and whether adjacent visual grids represent separate datasets. Before uploading, note the target table, its expected columns, and any printed control totals.
During review, check for merged section labels, repeated headers across page breaks, genuinely blank cells that should not shift subsequent columns, and footnotes mistakenly absorbed into data rows. Confirm that subtotals and brought-forward sums are explicitly tagged so they do not distort downstream calculations.
Use PDF to Excel when you need an immediate multi-sheet workbook with typed numeric cells. Choose PDF to CSV for single flat tables destined for database ingestion, or PDF to JSON when an application expects row arrays and schema definitions. If the same report structure arrives on a recurring schedule, convert the layout into a saved workspace workflow.
The 28-day Free trial includes 25 page-equivalents per month shared by hosted and own-key processing, 5 hosted pages per document, and 5 documents per day. Anonymous trials have separate daily limits.
Setup, review and export questions
Can it extract tables without visible borders?
Often, yes. The model can use alignment, spacing, headings, and repeated value patterns to infer a table even when the PDF has no drawn grid lines.
What happens to tables that span several pages?
The extractor can join repeated table structures across pages. Review page headers and carried totals because they can be mistaken for data rows.
Can I edit a cell before export?
Yes. The review workspace lets you correct extracted cells while comparing them with the source PDF.
How should I check a table that continues across pages?
Inspect the last detail row on one page and the first detail row on the next. Remove repeated headers, preserve carried-forward context, and decide whether page totals are controls or data. Then compare the final row count and total with the source before export.
Process the whole folder in one run.
Starter and above add stored files, batch jobs and connected exports. Email intake requires Pro or Scale. File and processing limits depend on your plan.
Compare batch plans