Rixoffice Documentation
PDF Editor

The AI Assistant in the PDF Editor

Published September 4, 2026Updated September 4, 20267 min read

Use the PDF Editor's AI assistant: 20+ tools that read pages, search, summarize, edit text, generate images, fill forms and edit outlines — processing locally.

On this page

The AI assistant in the PDF Editor is a set of more than 20 tools you drive in plain language: it reads pages, searches text, summarizes, marks up and edits text, generates images, fills forms, rotates and deletes pages, and edits the outline. The design principle is that the AI reads the PDF locally — content is processed on your machine by the local AI pipeline — so asking it to work through a 300-page contract is a private act, not an upload. You use it when the task is bigger than one click: the summary of a long report, the form filled from an attachment, the forty mentions of a product name that all need updating.

Ask in plain language

  1. Open the PDF and open the AI assistant panel.
  2. Type the task: "summarize this document," "find every mention of the old product name," "fill the form using the attached invoice."
  3. The AI works through its tools — reading pages, searching, parsing — and reports back.
  4. Review the result: proposed edits preview before applying; answers and summaries are quotable as they are.
  5. Apply what is right and Ctrl+S.

The assistant handles the whole document, not just the visible page. "Search the entire file for every occurrence of the termination clause and summarize each" is a normal request — the kind that takes the AI a minute and a human an afternoon.

The tool families

The 20+ tools group into jobs:

FamilyWhat the AI doesExample request
Reading and searchReads pages, searches text"What are the payment terms on page 14?"
SummarizingCondenses the document or sections"Summarize this in ten bullet points"
Markup and text editingMarks up and edits text in place"Update every 2025 date to 2026"
Image workGenerates images, applies image edits"Generate a diagram for this process"
FormsFills interactive fields"Fill this form from the attached W-9"
PagesRotates, deletes pages"Delete the blank pages"
OutlineEdits the bookmark tree"Rename the outline entries to match sections"

The families compose: a single request like "clean up this scanned report — fix the dates, delete the blank pages, add a summary page" walks several tool families in one pass.

File-parse: attachments as input

The file-parse capability handles attachments — files you hand the AI alongside the request:

  1. Provide the attachment: the invoice whose values fill the form, the spreadsheet whose figures belong in the summary, the old contract for comparison.
  2. The AI parses it and uses the content in the task.
  3. The form fills itself from the invoice; the summary incorporates the figures; the comparison lands as notes.

This is the bridge between "the PDF" and "the context around the PDF," and it is what makes form filling at scale practical: one form, one attachment, one request.

Local reading, concretely

"Reads the PDF locally" has a precise meaning here:

  • Document content is processed by the local AI pipeline on your machine.
  • With Ollama and a local model, the entire path — document, prompt, answer — stays offline.
  • With a cloud provider configured, calls follow that provider; the editor never uploads your PDF on its own initiative.

The practical consequence: confidential documents — contracts, medical records, financials — can be summarized and queried without the content leaving the building.

Where the AI fits and where it does not

The AI assistant is the multiplier for volume tasks and the reader for length problems. It is not the tool for judgment calls that are yours to make: which clause to accept, whether the summary's emphasis matches your audience's. A workable division of labor:

  • AI: find, summarize, fill, bulk-edit, restructure pages on command.
  • You: review the diff, decide, sign, send.

The best pattern is AI-first, human-final: let the assistant do the finding and the typing, then read what it changed before you save. Combined with the editor's manual tools — text editing, annotations, signatures — the result is a PDF workflow where the machine does the volume and you keep the pen for what matters.

What to explore next

Quick steps

  1. Open the PDF and open the AI assistant panel.
  2. Ask in plain language: summarize the document, find every mention of X, fix the outdated dates.
  3. Review the proposed changes or answers before applying them.
  4. Use file-parse when the AI needs values from an attachment.
  5. Apply, then save with Ctrl+S as usual.

Frequently asked questions

What can the AI assistant do in the PDF Editor?

More than 20 tools: read pages, search text, summarize, mark up and edit text, generate images, apply image edits, fill forms, rotate and delete pages, and edit the outline — driven by plain-language requests.

Does the AI read my PDF locally?

Yes. The AI reads the PDF locally — document content is processed on your machine by the local AI pipeline. If you run Ollama with a local model, the whole path stays offline; otherwise calls follow the provider you configured.

What is file-parse for attachments?

The AI's file-parse capability handles attachments: it reads supported files you provide — a source document whose values should fill a form, for example — and uses them in the task you asked for.

Related documentation