What exactly the new feature can do

Google officially announced that Gemini now supports native file generation directly in the chat interface. Instead of users having to copy text and manually format it in their own applications, it is enough to specify in the prompt what content and in what format they need.

Supported formats include:

• Microsoft Office — Word (.docx) and Excel (.xlsx)
• Google Workspace — Docs, Sheets, and Slides
• PDF — for printable documents and reports
• CSV and Plain Text (TXT) — for data and simple notes
• LaTeX, RTF, Markdown (MD), HTML, and XML — for developers, academics, and publishing
• ZIP archives — for bundling multiple files at once

The feature is available globally to all users of the Gemini app, as confirmed by the official Google The Keyword blog. After generating the file, it can be downloaded directly to the device or exported to Google Drive.

How it will change everyday work

This update significantly shortens the path from thought to output. Until now, it was necessary to have AI generate text, copy it, paste it into Word or Excel, and only then finish formatting there. Now a single prompt is enough.

Practical examples of use include:

• Financial planning — creating a budget in Excel based on natural language input
• Drafts and reports — generating structured documents in Word or PDF for team communication
• Code and development — exporting scripts in HTML, XML, or Markdown for direct use in the development environment
• Bulk bundling — creating multiple files at once and downloading them as a ZIP archive

Editors of the Android Authority server point out that this change is fundamental especially for so-called vibe coding — an approach where developers let AI generate code iteratively in natural language. Users on Reddit are already demonstrating advanced scenarios, such as searching for files in Google Drive and processing them into a ZIP folder with scripts.

Competition doesn't sleep

Although Google is catching up with this step, it still lags behind the competition in some areas. ChatGPT from OpenAI has for some time allowed through Code Interpreter (Data Analyst) to generate and run files including Excel, PDF, or Python scripts. Anthropic Claude, on the other hand, offers Artifacts — interactive panels where code, texts, and visualizations can be edited directly in the browser.

Gemini, however, offers one advantage: deeper integration with Google Workspace. Generating native Google Docs, Sheets, and Slides directly from chat makes work easier for companies and individuals who already use the Google ecosystem. Files can be immediately exported to Drive and shared in the team.

Prices and availability in the Czech Republic and Slovakia

The file generation feature is available for free within the basic version of Gemini. For more advanced models and extended limits, Google offers a Gemini Advanced subscription, which is part of the Google One AI Premium package for 19.99 USD per month (approximately 450 CZK). This plan includes access to the most powerful models, an extended context window, and priority access to new features.

For Czech and Slovak users, it is important that Gemini supports Czech both in the interface and in generated outputs. Articles, reports, and tables can therefore be created in the mother tongue. Availability is global, there is no regional restriction for the EU market.

From a regulatory perspective, Google must in Europe comply with the EU AI Act, which emphasizes transparency of generative models. Google therefore displays a label in Gemini that content may be generated by artificial intelligence, and provides mechanisms for checking the factual accuracy of outputs.

Limitations and early problems

Although this is a useful addition, the feature is not without shortcomings. Users report several problems:

• Transparent PNG — Gemini cannot yet generate images with a transparent background, which limits use for graphic design.
• Stability — for some users, the web application crashes when generating files, or the file is not created correctly.
• Editing — the model sometimes cannot edit an existing file, but must generate it entirely anew.

As pointed out by the PiunikaWeb server, Google has not yet published complete documentation of all supported formats and their limitations. It is therefore recommended to always check outputs before important use.

Where productive AI is heading

The trend is clear: chatbots are changing from passive conversational partners to active tools for content creation. Generating files directly in the conversation is another step towards artificial intelligence truly taking on the role of a personal assistant — not just for searching for information, but for complete document production.

With this step, Google is responding to growing pressure from OpenAI and Microsoft. The key question remains who can best connect content generation with existing corporate workflows — and whether Google can fully exploit its advantage in the form of the Workspace ecosystem.