Gemini Can Now Generate, Preview and Edit Files Inside Chat: What Works Where

Google has turned the Gemini chat window into a place where files are made and fixed, not just described. In a post dated October 29, 2026, Google announced that you can now "generate, preview, and edit documents, PDFs, presentations, spreadsheets, and more directly in Gemini." Expand a file to full screen, type edits yourself, highlight text to ask questions, or tell Gemini to refine it. The company says the feature is available to all users in all languages where the Gemini app is offered.

This is a small announcement with a real workflow change behind it: the round trip of "ask the chatbot for a draft, copy it into another app, fix it, paste it back" gets shorter. Below is what Google actually said, the platform matrix, what we cannot confirm yet, and how it fits next to other recent moves in the Gemini app and rival assistants.

TL;DR: what changed in the Gemini app

What exactly can you do now?

Google's single worked example is a student workflow. Upload handwritten lecture notes, ask Gemini to create a PDF study guide, and then edit that PDF in place. You can change words by hand, highlight specific sections so Gemini rewrites the text or the visuals there, and ask further questions about the content, all without leaving the conversation.

That example tells you the shape of the feature:

• Generate. A prompt produces a real file, not only a block of text in chat.
• Preview. The file renders in the conversation so you can check layout before using it.
• Edit. A full-screen mode lets you type directly, the way you would in a word processor.
• Targeted AI edits. Highlight a passage and ask Gemini to change only that.
• Ask about a selection. Highlight and ask a question instead of requesting a change.

Google lists documents, PDFs, presentations, spreadsheets, code and more, and then names the web-only formats explicitly: Google Docs, Google Sheets, slide decks, Markdown files, TXT files and code.

What works on web versus mobile?

"Not listed" is our reading of Google's wording: the post says PDF is supported on web and mobile, while all other formats are supported on the web only. If you plan to edit a spreadsheet from your phone through Gemini, that is not promised here.

How does this compare with Claude in Google Docs?

The comparison readers will make is with the Claude for Google Workspace public beta, which puts a sidebar inside Docs, Sheets and Slides, reads the open file and edits it with approval per edit. The two approaches start from opposite ends.

Neither is strictly better. If your files live in Google Workspace and you want review-by-edit control, the sidebar model suits you. If you start from raw material such as lecture notes, a messy PDF or a code snippet and want a finished file quickly, the chat-first model removes a step. Because Gemini is Google's own product, it may also integrate with Docs and Drive more deeply over time, but the announcement does not claim that.

OpenAI has been pushing the same direction from another side. Its recent iCloud Mail and Calendar plugins for ChatGPT bring personal data into chat, and the pattern is the same everywhere: assistants are moving from answering questions to producing and modifying the artifacts you work with.

How does it fit with the rest of the Gemini app?

Google has shipped a steady run of Gemini app changes this month. The student hub gained immersive explorations, which turn topics into interactive lessons. Slash-command Skills replaced Gems in chat. On the developer side, the Connected Apps Platform lets builders publish to Search and the Gemini app through MCP or AppFunctions. And the underlying model line moved with the Gemini 4 Argon launch.

Read together, the study guide example in this announcement is no accident. Google is building a loop for students: upload notes, generate a guide, explore the topic interactively, then edit the output. The same loop applies to a report, a pitch deck or a spreadsheet at work.

Try it: prompts that use the new editing mode

Google's post does not publish prompt templates, so treat these as starting points to test, not documented behavior.

1. Study guide. Upload photos of handwritten notes and ask for "a two-page PDF study guide with headings and a summary table." Then switch to the full-screen editor and fix names or dates by hand.
2. Targeted rewrite. In a generated document, highlight a paragraph and ask Gemini to "make this more concise for a non-technical reader." Check that the rest of the file stays unchanged.
3. Ask, don't change. Highlight a clause in a contract-style PDF and ask what it means. Treat the answer as a pointer, not legal advice.
4. Spreadsheet draft (web only). Ask for a budget sheet with formulas, then inspect each formula before relying on totals.
5. Code or Markdown (web only). Generate a README, edit by hand, and ask Gemini to fix only a section you highlight.

A good habit with any of these is to diff the result mentally: after a targeted edit, scan the untouched parts to confirm nothing else moved.

What are the risks and open questions?

The announcement is short, and several practical questions are unanswered. We would verify these before depending on the feature for important work.

• Accuracy of generated files. Generated PDFs and sheets can contain confident errors, especially numbers and citations. Editing in place does not change that.
• Where edits are saved. The post does not say whether an edited file syncs to Google Drive, stays only in the chat, or can be exported, nor how versions are kept.
• Privacy. Uploading handwritten notes, contracts or financial sheets raises the usual questions about how Gemini handles uploaded content. Read the current Gemini Apps help pages and your privacy settings before uploading sensitive files.
• Limits. No file size, page count or rate limits are listed.
• Mobile gaps. Only PDF editing works on phones per Google, so mobile-first users get a narrower feature.
• Third-party plans. The post does not mention business or education admin controls. If your organization manages Gemini access, ask your admin how it applies.

If you are building agent workflows that touch documents, note that file-editing assistants widen the surface where a model can alter something silently. For teams worried about that, AgentBeam is the agent security platform from the explainx.ai team, built to stop AI agents before they take dangerous actions.

Does this matter beyond students?

Yes, in two ways. First, it lowers the cost of the first draft: a usable file appears in seconds, and you edit rather than compose. Second, it competes directly with the sidebar-assistant approach, which pushes every vendor to make editing feel native. For small teams without a Workspace admin, a chat that can output a clean PDF or slide deck is a convenient tool, as long as someone reviews the output.

For builders, the signal is the same as in the Connected Apps news: Google wants Gemini to be the surface where tasks complete. Files are the unit of work for most people, so generating and editing them in chat is a natural next step.

Bottom line

Google's October 29 update makes Gemini a place to create and edit files, with PDF support on web and mobile and the rest on the web. It is available to everyone, which makes it the most accessible version of this idea so far. What it leaves out, notably saving, limits and privacy details, is what you should check before using it on anything important. We will update this post if Google publishes more detail.

Related reading

• Claude in Google Docs, Sheets and Slides: public beta sidebar guide
• Gemini app Skills replace Gems
• Gemini Immersive Explorations in the student hub
• Google Connected Apps Platform
• Gemini 4 Argon launch, benchmarks and pricing
• ChatGPT iCloud Mail and Calendar plugins

Feature availability and platform support are accurate as of Google's October 29, 2026 announcement and may change.