Veyneo · Privacy verification
What “no upload” actually means
Your document can stay on your device while the website still sends ordinary requests. Here is the distinction, and the evidence behind Veyneo’s current claim.
Production test record
2026-10-01T13:44:17.741Z · Chrome 154.0.8037.59 · win32 10.0.26200 x64
Three sampled tools · online and offline · synthetic PDFs
6/6 downloaded outputs passed page-count checks. No document content detected in captured request bodies.
| Tool | Mode | Pages | Bytes |
|---|---|---|---|
| merge-pdf | online | 6 / 6 | 2110 |
| split-pdf | online | 2 / 2 | 1149 |
| extract-pdf-pages | online | 2 / 2 | 1149 |
| merge-pdf | offline | 6 / 6 | 2110 |
| split-pdf | offline | 2 / 2 | 1149 |
| extract-pdf-pages | offline | 2 / 2 | 1150 |
Read the request / output record (JSON)
Tested build fingerprint: index-Ddsii6hH.js
The boundary is the document
For the current Local tools, Veyneo reads the selected file in the browser, runs the operation on your device, and creates a local result. The source document is not sent to a processing server.
Sending a PDF binary, a Base64 copy, extracted text or page images to a remote processing or storage service would cross that boundary. Loading public HTML, JavaScript and CSS does not.
Local processing still has telemetry
While online, Veyneo may send /api/analytics/events with an anonymous browser/session identifier, route, tool, browser/device class, acquisition source, size/count/page buckets, timings and outcome metadata.
Processing failures may also report exact selected-file count, total input bytes and PDF page count for diagnostics. This is metadata. The intended payload excludes document bytes, filenames, local paths and extracted text.
The live site also sends Cloudflare performance telemetry to /cdn-cgi/rum. In this capture it contained page URL/referrer, navigation timing and browser/OS version metadata, with no document content detected.
Feedback is a separate, deliberate submission. Send only what you intend to share.
What we tested in production
On 1 October 2026 we sampled Merge PDF, Split PDF and Extract Pages on the live veyneo.com site, using synthetic documents.
Each tool ran online and then in a fresh context disconnected before file selection. The downloaded PDFs were parsed again and checked against expected page counts. Request bodies were reviewed separately from request counts.
Repeat the check yourself
Load the tool completely, open DevTools → Network, clear the log, choose a non-sensitive PDF and run the operation. Inspect every new request body.
For a second check, reload online and wait until ready, then disconnect before selecting the document. Run the operation and download the result.
Limits of this evidence
These are dated desktop Chrome tests of three tools and small synthetic PDFs. They do not prove every tool, device, large-file case or future version.
A page must load before the offline test. Features that load additional code on demand may need that code loaded first. Future Cloud/AI features must disclose a separate upload boundary before file submission.
