Formaster opens a PDF in the browser, uses AI to suggest where answers belong, and exports the completed document in its original layout. It targets scanned pages and flat PDFs that lack convenient fillable fields, with tools for placing text, dates, checkboxes, signatures and images. The important privacy detail sits between opening the file and analyzing it: opening happens locally, while AI page analysis runs on Formaster’s server, according to its product page.
An October 28, 2026 WeSearch record identifies a Hacker News submission for the product. Formaster’s own page has no original publication timestamp, so that date documents its appearance there, not a verified launch date.
The useful question is what leaves the browser at each step. Formaster describes separate rules for page analysis, label translation, saved templates and entered answers. Those distinctions determine how cautiously to use it with sensitive paperwork.
Opening a PDF Locally Is Only the First Step
Formaster says selecting or dropping a PDF opens it directly in the browser without uploading it at that point. Users can fill PDFs without an account, and AI is optional.
AI-assisted field detection changes that workflow. A model reads the page and proposes fields for the user to review. The page’s account instructions say users must create a free account to use the AI conversion flow. Its privacy description adds:
“Page analysis runs on formaster’s server.”
That separates two actions that can look similar in a browser interface. Viewing a document locally is one operation; asking a remote service to interpret its page content is another. A file opening without an upload does not mean every later feature stays on the device. Formaster describes both stages on its product page.
The service offers two AI settings. Normal uses one credit per run; Smart uses two credits and is available on paid plans. New accounts receive five free credits, with no card required. Those are Formaster’s published usage terms, not evidence of comparative accuracy. Neither setting removes the need to check the proposed fields.
Redaction Protects the AI Request, Not Every Data Path
Formaster provides two ways to hide information before AI analysis: paint over regions of a page, or find email addresses and ID numbers to conceal. The company says AI receives only page content that the user has not hidden.
That is a useful control for a form containing personal details alongside ordinary labels. A user could conceal an existing identifier while leaving enough surrounding structure for the model to recognize where a field belongs.
I would treat that hiding step as preparation for a specific outbound request. Its documented purpose is to limit what the AI receives. The page does not describe it as permanent redaction of the PDF or promise that the exported document has had the underlying information removed.
Three distinctions matter:
- Information already on the PDF is page content. An address printed on a scanned application needs attention before analysis, even if the user has not entered anything into Formaster.
- Answers entered in Formaster have a separate rule. The company says those answers and signatures remain in the current session.
- Saved templates can carry the original PDF into account storage. Hiding content from AI is not described as sanitizing that original file for synchronization.
The last point is the consequential one. A reduced AI payload and a synced original document are different data paths. If a sensitive PDF becomes part of a saved, account-synced template, the original may leave the browser through that workflow even though selected regions were hidden from AI. That is an implication of Formaster’s documented sync and analysis rules, not an observed security flaw.
This analysis relies on the product page and discovery record, not hands-on testing or an independent privacy audit. The hiding, storage and session-handling behaviors are vendor claims; the available evidence does not independently verify their implementation or deletion practices.
Every Detected Field Still Needs Review
Formaster supports seven field types: text, dropdown, date, checkbox, signature, initials and image. AI proposes fields, while the user reviews them before they are added. Manual placement remains available.
That review is central to making a scanned or flat PDF reusable. A model can identify something that looks like an answer area without fully understanding the form’s instructions. The practical checks are therefore broader than whether a box appears in roughly the right place:
- Meaning: Does the field correspond to the correct question or label?
- Type: Should it accept text, a date, a selection or a checkbox?
- Position: Will the answer land on the intended line or inside the intended box?
- Space: Does a realistic answer fit without obscuring nearby text?
- Completeness: Are any questions or signature areas missing?
These are recommended checks derived from the workflow, not a list of observed Formaster errors.
The product provides arrow-key movement, direct position entry and zoom for adjusting placement. Those controls matter because a plausible field location is not always a usable one. A date shifted slightly into a neighboring box can make a completed document harder to interpret, even when the entered value is correct. Formaster’s page describes these placement tools.
My view is that the useful promise here is a faster first draft of the field layout. The review step turns that draft into something worth reusing. Saving an unchecked template could otherwise repeat the same placement or labeling mistake on every subsequent copy.
Translation Covers Labels, Not Answers
Formaster says users can view a form’s labels in their own language. Its stated data rule is specific: translation sends labels, never answers.
That can help someone understand a form without changing the original document layout. It also limits the translation payload compared with sending a fully completed page.
The scope deserves careful reading. Label translation is not a promise that typed answers will be translated, that every instruction will be interpreted correctly, or that the exported PDF will become a translated document. The published feature description covers labels, while the export workflow preserves the original PDF layout.
For a form that must be returned in a particular language, I would use translated labels as an aid to comprehension and separately check the required language for responses. Understanding the question and submitting an acceptable answer are related tasks, but the translation feature only explicitly addresses the first.
Templates Save the Layout, With a Sync Tradeoff
Formaster can save a form’s layout as a template, reopen it for another copy, or duplicate it to create a variant. That is useful when the same application or administrative form arrives repeatedly.
The product distinguishes reusable structure from completed responses:
| Item | Formaster’s stated handling |
|---|---|
| Saved templates without sign-in | Stay in the browser |
| Saved templates with sign-in | Layouts and original PDFs sync to the account |
| Entered answers and signatures | Remain in the current session |
Frequently Asked Questions
3 questions
1Does Formaster upload a PDF when I open it?
Formaster says opening a PDF happens locally in the browser without an upload. AI analysis is server-side, and saved templates and their original PDFs sync when signed in.
