Privacy

Privacy policy

Last updated 2026-09-23. This policy covers the apps at sheetsa.com. It is written to be read; the terms of use are the agreement.

The short version

• Split and Specs never send your files. They run in your browser. The drawing set or spec book you open is not sent to us and we never see it.
• Each run sends a short usage report: which tool ran, how long it took, how many pages, how it went, and counts of what the engine found. Never the drawing, its name, or anything read from it. Every field is listed below, with the switch to turn it off.
• Only if you tick a box on Split, your corrections are shared with us together with the titleblock crop they came from, so the engine can be fixed where it misread. Off by default. What is sent.
• Archive collects your email address to sign you in, and stores the files you choose to upload so your organisation can find and use them. Nothing else.
• No advertising, no third-party tracking, no selling data. There is no third-party analytics script or tracker on this site. The usage report is the only thing this site records about the free tools, and it goes to our own database.
• You can have it all deleted by writing to support@sheetsa.com.

1. What we collect, and why

Split and Specs (the free tools), and the Archive tools that run in the browser. The file you open is processed on your device and is not transmitted to us. Split keeps your current project in your browser's own storage on your device so you can resume it; that data never leaves your browser, and clearing this site's data removes it. After each run, the tool sends us a usage report, described next.

Usage reports

We collect them so we can see how the reading engine does on real sets it has never seen, and improve it. One report is sent when a run of Split, Specs, Optimize, Restore or the PDF/A check finishes or fails, and one more each time you correct a value in Split. A report contains:

• which tool ran, and whether it ran in your browser or on a local server you run yourself;
• how long it took, and the size of the file in pages and bytes;
• whether it finished or failed, and if it failed, the kind of failure (for example, out of memory or not a PDF), not the error text;
• the options you ticked, such as the quality setting or the PDF/A level;
• counts of what the engine produced: how many sheets were matched, which titleblock positions matched how often, how many pages were scans, how many read states of each kind, how many PDF/A findings of each rule, how many images were recompressed;
• for a correction in Split: which field you corrected (sheet number, title, date or revision) and what state the engine had it in;
• your browser's family and version, your operating system, and your machine's processor count and memory class, because "ran out of memory at this many pages" means nothing without them;
• a random id for the browser tab, discarded when you close it, so two runs in one sitting can be told from two visitors.

A report never contains your file, its name, any page image or text, any sheet number, title, date, revision or section number, any recognized line, or anything you typed. It carries no cookie, no account and no persistent identifier, and it is sent anonymously even when you are signed in to Archive. We do not keep the network address it came from with it.

Usage reports are on in this browser.

Turn usage reports off in this browser

The switch is kept in this browser only; set it again in another browser or after clearing this site's data. If your browser sends a Do Not Track or Global Privacy Control signal, no report is sent from it whatever the switch says.

Corrections you choose to share

This is the one case where part of a drawing leaves your browser, and it happens only if you ask for it. On the Split page there is a box, "Help improve the engine: share my corrections with Sheetsa", which is off unless you tick it. While it is ticked, each value you correct also sends us: the small image of the titleblock the engine read that sheet from (the same crop the page shows you), the value the engine had and the value you typed, which field it was, and what state and confidence the engine had it in. Nothing else from the set is sent: not the file, not its name, not the rest of the page, not the other sheets. We use these to see exactly where the engine misreads titleblocks and to fix it. They are stored with the usage reports, are anonymous in the same way, and are read only by us. Untick the box to stop; it is remembered in this browser. If you shared something by mistake, write to support@sheetsa.com with roughly when, and we will delete it.

Archive (early access). When you sign in we collect your email address, because that is how sign-in works: we email you a link, or you use a password you set. We keep the email address, the time you signed in, and the organisation you belong to. When you upload a set, we store the file, the per-sheet PDFs made from it, what Sheetsa read from it (sheet numbers, titles, dates, revisions and the evidence for each), and the corrections your team makes, with who made them and when. We store these because they are the product: an archive that remembers.

Email you send us. If you write to support@sheetsa.com we keep the email to answer it.

Server logs. Our hosting providers keep standard request logs (the address a request came from, the page requested, the time) for a short period to run and secure the service. We do not use them to profile anyone.

Questions you ask the archive

Archive can answer a question about your organisation's sheets ("Ask the archive"). To do that, Sheetsa reads the text printed on each sheet in your browser when a set is added and stores it for your organisation, next to the sheet's index. A question is a search over that text and index inside your organisation's own rows in our database, and the answer is composed in your browser from the sheets it found, each cited. No AI company receives your drawings or your questions. Today no language model is involved at all. When one is added it will run in your browser or on servers Sheetsa operates, it will receive only the index rows and text snippets that matched, never a drawing, and this policy will say so before it happens.

An AI of your own, connected to your archive

An administrator of your organisation can make a key that lets an AI your organisation already trusts (for example one running in your own Microsoft tenant, or a workspace under your own agreement with its provider) search your sheet index and the text printed on your sheets through Sheetsa's server. That AI receives only records: sheet numbers, titles, disciplines, revisions, dates, how each was read, and text snippets. It never receives a drawing or a link to one, and never another organisation's records. Nothing is connected until you make a key; revoking every key closes the door. Every request is recorded for your organisation to see. What that AI's provider does with what it receives is governed by your agreement with them, not by this policy.

Packages you send to others

A member of an organisation can package sheets from Archive and send them to someone by a link ("Package and send"). The package holds copies of the chosen sheets, their index, a manifest and a transmittal that records what was sent; it is stored for your organisation and opens for anyone who has the link, without an account, until the date you chose (7, 30 or 90 days) or until a member of your organisation closes it. The sender chooses whether the link allows printing and downloading as well as viewing; nothing can be edited from it. The link is emailed from your own mail program, so we never see the email. Before the sheets show, the recipient gives a name and an email address, and we record that, together with what they viewed, printed and downloaded and when, and show it to the sending organisation's members. Recipients' names and addresses are kept with that record and used for nothing else. If you received a package and want your name removed from its record, write to the address below.

2. What we do not do

• We do not sell, rent or share your data with anyone for their own purposes.
• We do not use your files, or what Sheetsa reads from them, to train any model or to improve the product for other customers, unless you have expressly agreed to that in writing.
• We do not run advertising or third-party analytics or tracking scripts on this site. The usage report described above is ours, and stays with us.
• We do not read your files except to run the service, to fix a problem you have reported, or where the law requires it.

3. Where data is stored, and who processes it

Archive's data and the usage reports live on servers run by Supabase (the database and file storage) in the us-east-2 region (Ohio, United States), on Amazon Web Services. The website itself is served by Vercel. Sign-in emails are sent through Supabase's email service. These providers process data on our behalf under their own terms and security programmes; they do not get to use your data for themselves. If we add or change a provider, this policy will say so.

4. Who can see your data

Files and records you put in Archive are visible to the members of your organisation and to nobody else. That is enforced in the database, not only in the app: every row is tied to an organisation, and the database refuses queries from anyone who is not a member. See Security and data handling for how. We, as the operator, can access data when needed to run the service or to fix a problem you report, and where the law requires.

5. How long we keep it

For as long as your organisation uses Archive, plus a short period afterwards to allow for mistakes. When you delete a set, it is removed from the live system; backups taken by our hosting provider expire on their own schedule (currently days, not months). When an organisation closes its account, we delete its files and records. During early access, the system may also be reset with notice; keep your own copies of anything you upload. Usage reports, and any corrections you chose to share, are kept for two years from the day we receive them and then deleted automatically. They contain nothing that identifies you, so there is nothing in them to delete on request; a shared correction you sent by mistake can still be removed, as described above.

6. Your choices and rights

• Access and export. Everything Sheetsa read from your sets can be exported from the app as JSON and CSV at any time, and any selection of sheets can be packaged with its index and sent by link.
• Correction. Every value Sheetsa reads can be corrected in the app; the correction is kept with the record.
• Deletion. Delete a set in the app, or write to support@sheetsa.com to have your account, your organisation's data, or a specific file removed. We will confirm when it is done.
• Where the law gives you more (for example under the GDPR or the California privacy laws), you have those rights too; write to the same address and we will act on them.

7. Children

Sheetsa is a tool for professional work and is not directed at children. We do not knowingly collect data from anyone under 16.

8. Changes

If this policy changes in a way that matters, we will note it on the What's new page and update the date at the top. Continuing to use Archive after a change means you accept the updated policy.

9. Contact

Questions about privacy, or a request to access or delete your data: support@sheetsa.com.