How to Use Claude Cowork: A No-Code Guide for Professionals (2026)

What is Claude Cowork?

Claude Cowork is Anthropic's AI agent for knowledge work. You describe a task in plain English, and Claude plans it, works across your files and apps, and hands back a finished deliverable, a summary, a spreadsheet, a draft, a report. It launched on macOS on 12 January 2026 and on Windows on 10 February 2026. No coding required.

Here is the simplest way to understand it: Claude Chat answers your questions. Claude Cowork does the work. Chat is a conversation. Cowork is a working session, where Claude actually opens your files, builds the document, and gives it back done, while you steer.

Who Claude Cowork is for

If you can describe what you want to a capable assistant, you can use Claude Cowork. It is built for non-technical professionals, marketers, lawyers, CEOs, HR leaders, finance teams, consultants, not developers. You never touch code. You just explain the outcome you want.

What you need to get started

Here are the real requirements, so there are no surprises.

The Claude desktop app, for Mac or Windows. You need it whenever Cowork works with your local files, local connectors, browser use or computer use.

A paid Claude plan, Pro (from $17/month billed annually, or $20 monthly), Max, Team, or Enterprise. Cowork is included; there is no separate fee.

Your computer awake and the app open, for local work, when Claude works with files and apps on your actual machine, the app needs to stay open while a task runs. From 6 October 2026, new Cowork tasks on Pro and Max run in the cloud, so they keep going after you close the laptop, and scheduled tasks run with no device online.

On Windows, you need Windows 10 (version 22H2) or later, with the Virtual Machine Platform feature turned on. Cowork runs in a small isolated sandbox for safety.

How to set up Claude Cowork (step by step)

1. Open the Claude desktop app. On Pro and Max, chat and Cowork are now one Claude (rolling out in stages since 16 September 2026): you just ask, and Claude decides whether to answer quickly or take on the task. Team and Enterprise still show them separately for now, so find the mode selector at the top, you will see Chat and Cowork.

2. Click the Cowork tab to switch into task mode (on Team and Enterprise, or wherever you still see the switch).

3. Set your global instructions (optional but worth it). Go to Settings, then Cowork, then Edit, and add standing preferences, for example, "Always write in British English and a professional tone." Claude follows these every time.

4. Connect a folder. Point Cowork at a folder on your computer. That folder becomes its workspace, the files it can read and update for this project.

5. You are ready. That is the whole setup. Now you give it a task.

Your first task: a worked example

The best way to learn is to give Cowork one real, slightly tedious job. Here is a complete example you can copy.

Say you have a folder with three competitor PDFs. In the task box, type this: "Read every file in this folder. Write me a one-page summary of the three competitors, including a comparison table of their pricing and main features. Save it as a Word document called Competitor-Summary." Then it works like this.

Review the plan. Cowork shows you the steps it intends to take before it does anything.

Choose a permission mode. "Ask before acting" means Claude checks with you at each meaningful step, safer, and best while you are learning. "Act without asking" lets it run start to finish, faster.

Watch and steer. You see each step happen. If it goes the wrong way, you redirect it in plain English.

Get your file. The finished Word document appears in your folder.

That is the whole loop: describe, review, steer, done. Once a set of instructions works, save it, you now have a workflow you can run again next week in seconds.

What to actually use Claude Cowork for (by profession)

The sweet spot is repetitive, multi-step work involving files and information. Generic guides stop at "it can do tasks." Here is what that looks like in real jobs, these are the workflows we teach by profession.

Marketing teams, turn a campaign brief into a full content calendar, or compile a competitor monitoring report every Monday. See agentic AI for marketing teams →

Lawyers, summarise a contract and flag unusual clauses against your firm's standards. See AI for lawyers →

CEOs and executives, compile a board report from a folder of scattered documents and data. See agentic AI for CEOs →

HR professionals, draft job descriptions and screen applications against your criteria. See agentic AI for HR →

Finance professionals, build a spreadsheet with working formulas from raw monthly data. See agentic AI for finance →

Consultants, synthesise research notes into a client-ready deliverable. See agentic AI for consultants →

Five beginner mistakes to avoid

1. Being vague. "Help with my report" gives weak results. "Summarise these three files into one page with a pricing table" gives a great one. Describe the finished outcome.

2. Giving it your whole hard drive. Connect a specific project folder, not everything. Cowork works best with focused context.

3. Closing the app mid-task. When a task uses your local files or apps, Claude works on your real machine, if you close the app or your computer sleeps, the task stops. New cloud tasks on Pro and Max (from 6 October 2026) keep going.

4. Using "Act without asking" on day one. Start with "Ask before acting" until you trust how it works.

5. Re-typing the same instructions every time. When a prompt works, save it as a reusable workflow. That is where the real time savings come from.

Is Claude Cowork safe?

Yes, with sensible guardrails built in. Claude asks for explicit permission before permanently deleting any file. On Windows, its work runs inside an isolated virtual machine, separate from your operating system. You control which connected tools and internet access it has. And in "Ask before acting" mode, nothing happens without your say-so.

What Claude Cowork cannot do yet

Being honest about the limits saves you frustration.

It does not always remember across separate sessions. Memory now works across chat and Cowork, but it is off by default on Team and Enterprise (on by default for Pro and Max), and a saved project is still the most reliable place to keep a task's context.

It needs the app open and the computer awake to keep working on your local files and apps. From 6 October 2026, new tasks on Pro and Max run in the cloud instead.

You cannot share a live session with a colleague.

It uses more of your plan's usage than a normal chat, because it is doing far more work.

Keep learning: more Claude Cowork guides

Ready to go deeper? These guides build on this one.

Claude Cowork use cases by profession, real examples and copy-paste prompts for marketing, legal, HR, finance, executives, and consultants.

The best Claude Cowork prompts, a copy-paste prompt library organised by task.

How to automate your weekly busywork, a full worked example you can set up this week.

Claude Cowork vs ChatGPT, which tool to reach for, and when.

Is Claude Cowork safe?, security and privacy explained for professionals.

The fastest way to go from "I tried it" to confident

Most people open Cowork once, run a simple task, and never build the repeatable workflows where the real value is. Going from dabbling to genuinely building your own profession-specific workflows is exactly what the Eduk8agentic Claude Cowork course teaches, in plain English, for professionals, not developers.