The usual PDF editor workflow is click the spot, pick a tool, type, drag to resize, repeat. AI edit flips that around. You open a PDF, type a request like "put John Smith on the Name line and March 14 2026 in the date box at the bottom right", and the AI reads the page, finds those spots, and drops the overlays in place. You review the result and export. For AcroForm PDFs the AI writes directly into the named fields; for flat or scanned PDFs it places text or image overlays at the positions it detects.
Requests that work well are specific and tied to something on the page. "Add a checkmark in the Single box in Part I", "write 2026-04-19 in the Date field at the top", "place this signature image above the Signature of Payee line", "stamp PAID in red across the top of page 2". You can chain a few edits in one request or do them one by one and watch each land. If you paste a block of info and ask the AI to fill what matches, that goes through the AI fill path instead, which is a separate feature aimed at form fields.
When the AI puts something in the wrong place, undo it and either rephrase the request with a clearer landmark (the line above, the box labeled X, the column with the dollar signs) or drop into manual mode and move the overlay yourself. The free manual editor has the same text, image, and checkmark tools the AI uses under the hood, so nothing is locked behind the AI path. The editor keeps the same text, image and checkmark tools available when you prefer to make an edit yourself.
Placement edits on top of the existing PDF: text overlays, image overlays (signatures, logos, stamps), checkmarks, and updates to AcroForm field values when the PDF has real form fields. It can also apply a few of those in one request, like signing and dating at the same time. It does not do page-level layout changes, does not reflow paragraphs, and does not generate new pages.
No. A PDF is not a Word document, and rewriting text inside the original content stream would re-flow the whole page and usually break fonts, spacing, and any forms embedded in it. What the AI does instead is overlay new content on top, or, for the specific case where there is a blank-out needed, draw a white rectangle and place the replacement text over it. If you need the source text itself rewritten, you want the original document file (the .docx the PDF was exported from), not a PDF editor.
Hit undo and the overlay is removed. You can retry the same prompt (sometimes the AI picks a different spot on the second attempt), rephrase with a more specific landmark, or drag the overlay into place manually. Manual adjustments do not consume AI credits. If the page is a low quality scan and the AI is not finding fields reliably, manual editing is usually faster than fighting with prompts.