Any region you drag over in Chrome — a PDF page, an image, a video frame, a dashboard
Chrome extension — Manifest V3
You can see the table. You just can't have it.
PDF reports, screenshots, legacy dashboards — the Table Snap Chrome extension turns any table visible in your browser into clean CSV. One drag, flagged cells to review, done.
You're on the list. We'll email you at launch.
First access in signup order · founding price locked for your first year
No spam. One email at launch, one if we kill it. That's it. Your address is used only for these updates — privacy policy.
At a glance
Problem
Data you can see but can't select: PDFs, screenshots, old dashboards. Retyping wastes hours; blind OCR slips silent errors into your spreadsheet.
Input
Output
Clean CSV or clipboard table · column types preserved · low-confidence cells flagged for one-pass review
Data handling
Only the dragged region is captured · sent over TLS, deleted after processing · never used for training
Time to value
Under 60 seconds per table
Price
Free: 10 tables / month
Pro: $7/mo
Status
Pre-launch — waitlist open
What it handles
Real tables are messy. These are the cases blind OCR gets wrong — and the ones Table Snap is built around.
Structure
Layout that breaks naive parsers.
- Merged cells
- Spanned headers and grouped rows are un-merged into rectangular, analysis-ready data
- Multi-line cells
- Wrapped text stays in one cell instead of spilling into phantom rows
- Header detection
- Header rows recognized and kept as CSV header; footnote markers (*, †) stripped from values, kept as notes
Values
Numbers that mean what they say.
- Number formats
- 1,234.56 vs 1.234,56 · ¥ and full-width digits · % · negative-in-parentheses (123) → -123, all normalized per column
- Column types
- Numbers, dates, currency, and text are typed per column so Excel and Sheets don't mangle them on paste
- Confidence flags
- Blurry or ambiguous cells (0 vs O, 1 vs l) are highlighted for you to confirm — never silently guessed
Sources
If it renders in a tab, it can be snapped.
- PDFs
- Any PDF open in Chrome; Pro batches the same layout across pages (bank statements, monthly reports) into one CSV
- Anything visible
- Screenshots, dashboards without export buttons, docs sites, video frames
What extraction looks like
Sample output — illustrative, not a real customer's data.
SOURCE quarterly-report.pdf · page 14 · dragged region 412×260px DETECTED 6 columns × 14 rows · header row found Region | Q1 Rev | Q2 Rev | Growth | Units | Note Tokyo | ¥4,120,000| ¥4,890,500| +18.7% | 1,204 | Osaka | ¥2,310,800| ¥2,84O,100| +22.9% | 887 | † ... ⚠ 2 cells flagged for review (of 84) D7 "¥2,84O,100" — letter O or zero? [click to fix] F12 footnote symbol detached from value [confirm] → EXPORT report-p14.csv · copy to clipboard · open in Sheets
How it works
-
Drag over the table
Click the extension icon and select the region — in a PDF, a screenshot, any tab. Only that region is captured.
-
Review flagged cells
The parsed table appears with low-confidence cells highlighted. Fix the two or three that need eyes; everything else is already clean.
-
Export
Download CSV, copy as a spreadsheet-ready table, or pipe straight into Google Sheets. Pro keeps your extraction history.
Pricing
Start free. Upgrade when tables are part of your week, not your month.
Free
$0 /mo
- 10 extractions / month
- Flagged-cell review UI
- Export: CSV · clipboard
- Multi-page PDF batch mode
- Direct Google Sheets export
- Extraction history
FOUNDING PRICE — LOCKED 1 YEAR
Pro
$7 /mo
- Unlimited extractions
- Flagged-cell review UI
- Export: CSV · clipboard · direct to Google Sheets
- Multi-page PDF batch: same layout across pages → one CSV
- Extraction history, re-downloadable
Questions
How accurate is the extraction?
Instead of promising a percentage, Table Snap shows its work: every cell gets a confidence level, and low-confidence cells are highlighted for you to confirm or correct before export. You review three flagged cells, not three hundred — and errors never slip through silently.
What permissions does the extension ask for?
Capture works through Chrome's active-tab permission: it can only see a tab when you click the Table Snap icon on that tab, and only the region you drag over is captured. It cannot read your tabs in the background, and it requests no access to browsing history.
Does my data leave my machine?
The selected image region is sent over TLS to our extraction API, processed, and deleted — it is not stored after the response and not used for training. Nothing else on your screen is captured, only the region you drag over.
Does it work on PDFs?
Yes. Any PDF you can open in Chrome works — drag over the table on the page. Pro adds batch mode: extract the same table layout across a multi-page PDF (think monthly statements) into one CSV.
What about Japanese and other non-Latin tables?
Japanese is fully supported, including full-width digits, ¥ amounts, and mixed JP/EN headers. CJK, Latin, and common European number formats (1.234,56 vs 1,234.56) are handled at launch.
Why not just paste the screenshot into ChatGPT?
You can — and sometimes it works. But a chat reply gives you no per-cell confidence, silently invents values when the image is blurry, loses column types, and takes a prompt-copy-paste round trip every time. Table Snap is one drag, a review pass over flagged cells only, and a clean CSV — every time, the same way.
Which browsers does it support?
Chrome at launch (Manifest V3). It installs and runs on other Chromium browsers like Edge and Brave, and full support for them is on the roadmap after launch.
When does it launch?
Early access ships within a few weeks of this waitlist validating demand, followed by Chrome Web Store review. Signups are invited in order, and founding pricing is locked for your first year. If we decide not to build it, you get one honest email saying so.