Frequently asked questions

Schema and extraction

Can I edit a schema after I’ve already used it?

Yes, at any time. Editing a schema doesn’t change documents you’ve already processed — those reflect the old schema. To get the new shape on previously processed documents, you’d re-run them against the updated schema. Re-runs are billed (see below).

Do re-runs cost me again?

Yes. Each pass through the engine deducts pages at the same per-page rate. Most teams finalize their schema on a small batch (5–10 documents) before processing volume, so the production run is the only paid run.

The good news: correcting a wrong value in the side-by-side viewer does not cost a re-run. Corrections store an override on top of the existing extraction — they’re free.

What counts as a “failed” extraction?

A page failed if the engine couldn’t return a result for it — typical causes are a corrupt file, a password-protected PDF, an unsupported sub-format (a TIFF with unusual compression, for example), or an internal error.

Failed pages are refunded to your pack automatically. The dashboard shows the failure reason so you can decide whether to re-upload the source after fixing it.

Does the engine know which page a value came from?

Yes. Every extracted value has a bounding-box coordinate tied to its source page. The side-by-side viewer uses that to highlight the region when you click a field (and vice versa). Multi-page documents work the same way — clicking a field on page 6 of a 7-page contract scrolls the viewer to page 6 with the value visible.

What’s the “Schema updated” badge in the documents list?

When you edit the project’s schema (rename a field, add or remove one, change types), documents that were extracted under the older version are tagged Schema updated. The existing data is still fine to view and export, but its shape lags the current schema. To re-cast it into the new shape, use Re-extract on the row or Re-extract stale at the top of the documents list — both pass the document through the engine again and deduct pages accordingly.

Files and formats

Are there file size or page limits?

• Per file: 500 MB max.
• No total quota on uploads in a session.

If your file is larger, split it in your PDF tool of choice and upload the chunks as separate documents.

Which file formats does Ztract support?

PDF, OFD, Word (.doc / .docx), Excel (.xls / .xlsx), PowerPoint (.ppt / .pptx), HTML (including .mhtml), TXT, CSV, RTF, plus images (JPG, PNG, WebP, TIFF, BMP). HEIC, ZIP archives, and email (.eml) are not supported in the current release.

Does Ztract handle non-English documents?

Yes. Latin, CJK (Chinese, Japanese, Korean), Cyrillic, and Arabic scripts are all supported, including pages that mix two or three on the same document. The schema is the same regardless of source language; field values come back in their original script. Numeric formats (dates, amounts) are normalized to a canonical form for downstream filtering.

What languages does the dashboard itself speak?

The interface is available in English, 中文 (Simplified Chinese), 日本語, 한국어, Français, Deutsch, Español, and Português. Switch languages from the avatar menu in the top-right; your choice is remembered per browser. Schema templates are localized to your UI language too — the catalog updates when you switch.

Can the engine read handwriting?

Block-printed handwriting on forms is read alongside typed text. Cursive handwriting is best-effort — the engine will try, but confidence on cursive fields is usually lower and worth a manual review pass.

Team and account

How do I sign in?

Enter your email on app.ztract.com and we’ll send a one-time verification code; there’s no password to remember. The same screen also supports Continue with Google if you’d rather use a Google account. The flow is the same for first- time signups and returning users — no separate “create account” step.

Can multiple people on my team share an account?

Not in a multi-seat way yet. The current release is single-account — each Ztract login is one user. Multi-seat team workspaces with per-user permissions are on the roadmap. Smaller teams sometimes share login credentials in the meantime; that’s at your own risk and we don’t recommend it for production use.

How do I close my account?

Email support@ztract.com from the account’s registered address. We delete account data and any remaining documents within 30 days, except for tax records we must retain.

What happens to my documents if I close my account?

They’re deleted — immediately from active storage and within 14 days from routine backups. Same as if you’d deleted them one by one through the dashboard.

Privacy and data

Does Ztract train models on my documents?

No. Not our engine, not the third-party LLMs we route through. The commercial APIs we use (OpenAI, Anthropic, Google) prohibit training on API-submitted data; we rely on those commitments and don’t opt into any optional training programs.

How long are documents stored?

Indefinitely while they exist in your account; immediately deleted when you remove them. Routine backups overwrite within 14 days. See Privacy and data handling for the full picture.

Where is my data physically stored?

In the United States — Fly.io’s Los Angeles region for compute, Supabase on AWS us-east-2 (Ohio) for the database, and Cloudflare R2’s Western North America region for documents. For EU customers, the transfer is covered by Standard Contractual Clauses incorporated in our DPA.

Is Ztract HIPAA-compliant?

We are not HIPAA-certified in the current release. We follow strong security practices (TLS in transit, encryption at rest, isolated processing, no training on customer data) but we do not offer a Business Associate Agreement (BAA). If you have a HIPAA requirement, talk to us — we track demand and may be able to accommodate enterprise arrangements.

What about SOC 2?

A SOC 2 audit is on our roadmap. We don’t have an attestation today. Our Privacy Policy and DPA describe our security and processing practices in detail.

API and integrations

Does Ztract have an HTTP API?

Not yet. The current release is dashboard-first — you upload, review, and export through the browser. An HTTP API, webhooks, and platform integrations (Zapier and similar) are on the roadmap. Email support@ztract.com to get on the early-access list.

Can I integrate Ztract with Zapier / n8n / Make?

Not directly today (no API yet). The closest workflow today is to export from the dashboard as JSON / CSV / Excel and feed those files into your automation tool’s file-watch trigger. We’ll have native connectors once the API ships.

Billing

Is there a free plan?

There’s a 30 free pages credit on signup, with no time limit, no credit card required. Beyond that, Ztract is purchase-based — buy a pack when you need more pages. Packs are valid for 12 months. No subscription, no recurring charge.

Can I get a refund?

Yes, for any reason, within 14 days of the pack’s purchase date. Email support@ztract.com from the registered account address.

Can I get an invoice with my company’s tax details?

Yes. Every purchase generates a Stripe invoice automatically, with VAT or GST details where applicable. EU customers can add their VAT number at checkout for the B2B reverse charge. Past invoices live in the Stripe billing portal linked from your account settings.