Anthropic, with its recent Claude update, turns the chat window into a full document factory. You can now ask Claude to create and edit files in PDF, slide deck and spreadsheet formats right inside your conversation — gone are the cut-and-paste handoffs between AI and office apps.
The rollout begins with power users, as Max, Team, Enterprise plans and Pro members receive access over time. It’s a deceptively simple tweak, but one that eliminates one of the longest running friction points in AI-driven work: getting polished files out of a chat and into a format you can share.
What’s new: launching the ability to create files in a chat
In a similarly Claude conversation you would tell Claude what kind of output you want — an analysis-ready spreadsheet or presentation that you can send to a client or pdf delivered in the format needed, and Claude will generate the file for download. It can also easily be converted between formats, such as turning a PDF into slides or raw data into a spreadsheet with formulas, charts and written summary.
Instructions of formatting are fair game: imagine “double-spaced Times New Roman 12 point with running header” or “slides with a title, three bullets and a chart on each page.” The goal is to bridge this final mile between draft content to a sharable asset.
How does it work, and who gets it first
Anthropic explains Claude uses a private compute environment to write code and run the software to construct and edit files—similarly to the ‘sandboxed code interpreter’ system which has become a mainstay of advanced AI tools. This allows the model to generate formulae, plot charts and ensure the structure of the document without ever having to navigate out of the chat.
The feature is available now for Max, Team, and Enterprise users, and is coming soon for Pro subscribers. To turn it on, switch “Upgraded file creation and analysis” in experimental features under Settings. You can upload source content for editing or have Claude generate from the ground up, then download the result, or save it in Google Drive.
Why this is important in the real world
For a lot of teams, the problem isn’t coming up with ideas — it’s shipping them. Where time vanishes is in transitioning from chat history to neat spreadsheet or presentation. By creating files in place, Claude eliminates the wasteful swivel-chair work that bogs projects down and causes errors.
It also edges AI into agentic behavior. Claude doesn’t just recommend content to read, he performs more complex processes such as parsing data, executing calculations, charting the output, and saving the work in a shareable file. Gartner has identified AI agents as an emerging category, and this is on track for that.
The update takes the form of one small part of an overall push to plug AI into everyday tools. Microsoft’s Copilot generates certain types of Word, Excel and PowerPoint files within Office; Google’s Gemini accomplishes similar tasks within Docs, Sheets and Slides. Anthropic’s model they are building centers the chat as the hub, and adds recent integrations which automate tasks inside other services such as Canva, Asana and Figma. The shared objective: decrease loss of context among apps.
Adoption tailwinds are strong. In recent study by McKinsey, over 50% of companies have implemented generative AI in at least one business function. The fastest adoption often comes in features that pull the process from idea to artifact smaller, because it’s the daily pain point.
Security, privacy, and governance
Anthropic warns that allowing this feature provides Claude internet access to generate and analyze files, which could offer a data security risk. The company recommends keeping a close eye on chats. For this sensitive work, organizations should put in place the standard controls: classify data, restrict uploads, route outputs through review.
Enterprises will need clarity about data handling, retention and auditability before they roll out the toggle more broadly. Advice from groups like the Cloud Security Alliance and the NIST AI Risk Management Framework advocates for guardrails like DLP policies, redaction of PII and staged rollouts with human-in-the-loop checks.
Use cases — and what to watch
There are a few day-to-day victories that stand out: Making sense of a messy CSV for a revenue model, creating pivot tables and charts, or taking a research brief and putting it into a 10-slide deck with speaker notes; Reformatting the manuscript into a submission-ready PDF; Structuring and turning a PDF syllabus into Slides for a training.
And because the work is confined to a single thread, context is preserved through edits and iterations.
Limitations remain. For very complex spreadsheets : Some Special macro, edge case formulas or having heavy datasets might need manual clean up. Charts and diagrams should also be evaluated for correctness and ease of use. Like any generative system, trust comes from checkathons, especially when you’re dealing in a numbers game.
The bottom line: Claude’s generation of new files brings together AI output and usable products in a gap as old as AI itself. For your crew that’s spent hours exporting, reformatting, rechecking, it’s a change that could bring hours back to the task of actual analysis and storytelling — all within the chat where the work started.
