Frequently Asked Questions

Cowork mode is a feature of the Claude desktop application (made by Anthropic) that turns Claude into a hands-on AI assistant for your computer. Unlike the browser version of Claude — which can only have conversations — Cowork mode can actually do things: create Word documents, Excel spreadsheets, PowerPoint presentations, search the web, connect to your apps, run code, and manage files. It's designed for non-developers who want to automate real tasks without any coding.

Cowork mode is part of the Claude desktop application. Claude offers a free tier, but Cowork mode may require a Claude Pro or Team subscription for full access. You can check the current pricing on claude.ai — Anthropic updates their plans periodically. Since Cowork mode is a Research Preview feature, availability may change as the product evolves.

No — they are different products. Claude Code is a command-line tool designed for developers to delegate coding tasks directly from their terminal. Cowork mode is designed for everyone — including non-technical users — and focuses on productivity tasks like creating documents, doing research, and automating workflows through a friendly desktop interface. You don't need to know any code to use Cowork mode.

The Claude desktop application supports macOS and Windows. You will need the desktop app installed on your computer to access Cowork mode — it is not available through the browser version of Claude.

Visit claude.ai in your web browser and sign in or create an account. From your dashboard, look for a "Download" or "Get the app" option. Download the installer for your operating system and run it. Once installed, sign in with your Claude account to get started.

You need a Claude account to use the desktop app. A free account gives you access to basic features, but Cowork mode — being a more advanced, resource-intensive feature — may require a paid plan (Pro or Team). Check the latest details on the Claude pricing page since plans and included features can change.

Within the Cowork mode interface, there is typically an option to "Select workspace folder" or "Choose output folder." Clicking this opens a file browser where you can pick any folder on your computer. All files Claude creates will be saved there. We recommend creating a dedicated folder like "CoworkOutputs" on your Desktop so everything is easy to find.

No — Claude can only access files in the specific workspace folder you select. It operates in a secure sandbox and does not have permission to browse your entire file system, access your Desktop (unless you select it as the workspace), or read files in other locations. This is intentional and is a core security feature of Cowork mode.

Cowork mode can create a wide range of file types including: Word documents (.docx), Excel spreadsheets (.xlsx), PowerPoint presentations (.pptx), PDFs (.pdf), Markdown files (.md), HTML files (.html), React components (.jsx), SVG images (.svg), and various code files (Python, JavaScript, etc.). The most polished outputs come from the document formats (docx, xlsx, pptx, pdf) as these have specialised Skills built for them.

Yes! You can upload files directly into the Cowork chat and ask Claude to work with them. For example, you can upload a rough draft Word document and ask Claude to "improve and expand this," upload a CSV file and ask Claude to "create an Excel dashboard with charts," or upload a PDF and ask Claude to "extract the key points." Image files (PNG, JPG) can also be viewed and analysed by Claude.

MCP (Model Context Protocol) connectors are integrations that let Claude connect to external apps like Slack, Notion, Google Drive, Asana, and more. Once you connect an app, Claude can read data from it — for example, searching your Slack messages, reading Notion pages, or pulling tasks from Asana — and incorporate that data into its work. You set up connectors in the Settings or Integrations section of the Claude desktop app. Each connection requires a one-time authorisation.

Skills are installable bundles of instructions that teach Claude how to perform specific tasks with expert-level quality. For example, the "pptx skill" contains detailed best practices for creating PowerPoint presentations, and the "docx skill" knows exactly how to format professional Word documents. Skills are triggered by specific phrases in your prompts. You can install community Skills or create custom ones using the Skill Creator built into Cowork mode. Plugins bundle multiple Skills together for specific use cases (e.g., a Sales plugin, Marketing plugin, etc.).

Claude can draft emails and messages, but it will always ask for your explicit confirmation before actually sending anything. It will never send messages, publish posts, or take any irreversible action without showing you what it plans to do first and waiting for you to say "yes." This is a core safety feature — you're always in control.

Cowork mode is designed with security in mind. It operates in a sandboxed environment, only accesses files you explicitly share, always asks for confirmation before taking irreversible actions, and never enters passwords or financial credentials into forms. That said, as with any AI tool, you should be mindful of what sensitive information you include in your prompts. Avoid including passwords, credit card numbers, or other sensitive credentials in your chat messages.

No — this is a hard rule that cannot be overridden. Claude will never enter passwords, bank account numbers, credit card details, social security numbers, or other sensitive credentials into any form or website, even if you explicitly ask it to. If a task requires entering sensitive financial or identity data, Claude will explain that you need to do this yourself. This protection is there to keep your most sensitive information safe.

Anthropic's data handling practices are governed by their Privacy Policy, which you can find at anthropic.com. For the most current and accurate information on how your data is used, stored, or shared, we recommend reading Anthropic's official Privacy Policy and Terms of Service directly. CoworkGuru is an independent site and cannot speak on Anthropic's behalf regarding data practices.

There are some things Cowork mode won't do by design: enter passwords or financial credentials, make purchases or financial transactions, permanently delete files, modify system files or security settings, create new accounts on your behalf, share documents with third parties, or generate harmful, illegal, or unsafe content. These restrictions exist to protect you and are intentional safety features, not bugs.

Like any AI, Claude can make mistakes. It might occasionally get facts wrong, misinterpret a vague prompt, or produce output that doesn't quite match your vision. This is why iteration is important — treat Claude's first output as a draft. Review it, then follow up with specific instructions to improve it: "Make the introduction shorter," "Add more data to section 2," "Fix the formatting." Claude gets better with each refinement.

Usage limits depend on your subscription plan. Free users typically have a limited number of messages per day. Pro and Team plans offer higher or unlimited usage. Since limits change with Anthropic's plans, check the current limits on claude.ai for the most up-to-date information. Complex Cowork tasks (like creating large documents) may count as multiple messages.

If Claude can't complete a task, it will tell you clearly and often explain why. Common reasons include: the task involves restricted actions (like entering passwords), the required information isn't available (e.g., trying to access an app that isn't connected), or the request falls outside what Claude is permitted to do. In many cases, Claude will suggest an alternative approach or ask for more information to proceed.