Home / Choosing the right way in
Chat, Cowork, Claude Code, three tools, three uses
The chat answers, Cowork acts on your files, and Claude Code works on code: three interfaces for the same family of models. What sets them apart is the tools the model can call and the context you give it. Code work goes to Claude Code, knowledge work to Cowork.
Anthropic ships three distinct interfaces for the same family of Claude models: the chat on claude.ai, the Cowork desktop agent, and Claude Code, the tool for working on code, which runs in the terminal (the window where you type commands) as well as in the desktop app, the code editor and the browser. One word comes back throughout this course and is worth defining right away: an agent is a program that chains several actions together on its own to reach a result you have described, instead of handing you an answer and stopping there.
The same model, different tools
The same family of models runs under all three surfaces: Opus, Sonnet and Haiku. What sets one surface apart from another is therefore not a different intelligence, but the tools the model can call and the context you give it.
The chat answers, Cowork acts
The chat on claude.ai remains the simplest entry point: you paste or describe some information, Claude answers in the conversation, and nothing else moves on your computer. Cowork changes that logic. It is Anthropic's desktop agent. As of the 2 September 2026 survey, it is available in the desktop app on all paid plans, and on the web and mobile on Pro, Max and Team plans, with an Enterprise account getting it wherever an administrator has switched it on. These commercial tiers shift. You give it access to specific folders and describe a result, such as a report written from scattered notes or a folder of invoices sorted by month, and Claude actually reads, edits, creates and organises files on your machine, in several steps, with no manual coordination from you.
Claude Code, reserved for repository work
Claude Code runs the same agent engine as Cowork, but Anthropic documents it for a different job: that of teams who write programs. It reads all the files of a software project, edits them, runs commands, and handles the actions specific to development work. The official distinction fits in one simple sentence: code work, fixing a bug, opening a merge request, running tests, goes to Claude Code, while knowledge work, writing a document, organising files, doing research, goes to Cowork. One nuance matters: Cowork runs its reasoning and its commands on Anthropic's servers, in an isolated environment, not on your computer; access to your local folders goes through the desktop app being left open.
Three ways of opening the same request show the difference:
Chat: "Summarise this meeting note in five points."
Cowork: "Sort the Invoices folder by month, without deleting anything."
Claude Code: fix the display bug in the invoice total, in the project repository.The last two ask an agent to act, on office files or on a code repository. The next lesson details how to open a first Cowork session, the recommended entry point for anyone who has never used an agent.
A merger under way since 16 September 2026
On an account that has moved to the new experience, chat and Cowork become a single conversation, and Claude Code is then the only separate tool. This lesson keeps its value for any account that the switch has not yet reached; the lesson on the merger explains what changes.
On 23 October 2026, the single conversation was extended to Enterprise accounts; it already existed for the Pro and Max plans since September. Another application, Claude Duet, where Claude acts as a proactive agent, is in research preview for Pro and Max subscribers in the United States, on a waitlist, on iOS, Android and claude.ai.
Chat, Cowork and Claude Code, four criteria
| Surface | File access | Technical level required | Typical use case | Approval mode available |
|---|---|---|---|---|
| Chat (claude.ai) | No direct access, a file has to be copied or attached into the conversation | None, opening a conversation is enough | Write, analyse, answer a one-off question | The basic chat acts on nothing; a connector it uses applies its own level, Always allow, Needs approval or Blocked |
| Cowork | Connected local folders, once access is approved | None; access to local folders goes through the desktop app | Organise a folder, fill in a spreadsheet, write a report from scattered notes | Three approval levels, changeable at any time, from a pause before every action to no control at all |
| Claude Code | An entire code repository, read and edited directly | Knowing how to write programs; the terminal stays optional, the desktop app, the editor or the browser are enough | Fix a bug, open a merge request, run tests | The starting mode depends on the plan and the surface, with Manual as the default label; six modes documented in all, default (Manual), acceptEdits, plan, auto, dontAsk and bypassPermissions, settable on the command line |
An employee opens a Claude conversation, pastes in her meeting notes and gets back a written summary. The next day, she wants the same work on the forty meetings of the year, whose notes are spread across as many files in a folder on her computer. She opens Cowork, gives it access to that folder, and asks for one summary per file.
Write in one sentence what this situation establishes, and in one sentence what it does not establish.
What this establishes: It establishes that the first job required her to bring the text into the conversation herself, and that the second assumes a tool able to go and read the files where they are already stored.
What this does not establish: It does not establish that the second job succeeded, nor what Cowork can reach outside the folder she gave it access to.
The three most common miscalibrations
- Too broad Cowork can now act on any file on the computer, connected folder or not, as soon as it is asked.
- Too narrow This situation shows nothing more than a personal preference for Cowork over the chat.
- Beside the point It establishes that the summary obtained on the first day was of better quality than those requested the next day.
- The three surfaces rely on the same family of Claude models. What changes from one surface to another is the tools available and the context provided.
- The chat processes the text you bring into the conversation. It does not access any file on your computer by itself.
- Cowork acts directly on the connected local folders, without ever opening a terminal, which makes it the right entry point for anyone who does not know the command line, as long as the account has not moved to the new experience that merges it with the chat.
- Claude Code reads and edits an entire code repository and handles commits and merge requests, a territory reserved for technical tasks.
- The choice of surface follows the nature of the expected deliverable, not habit or whichever surface is already open on screen.
Open the three interfaces on your machine, the chat, Cowork if your plan gives access to it, and Claude Code if you have it, then match each one to a concrete example of a task from your own job.
These points depend on an interface or a rule that may have changed since this was written. Check them on your own screen before relying on them.
- The short labels of the three Cowork approval levels shown in the interface, Manual, Auto and Skip, are read on screen as of the 2026-09-28 survey; a label translated or renamed since takes precedence over this course.
Every datable claim in this lesson links here to the public text behind it. A source that does not open proves nothing.
- Anthropic, get started with Claude Cowork, surfaces, plans and approval levels consultée le 2026-09-02
- Anthropic Help Center, Claude Cowork and chat are one Claude, release note of 16 September 2026 consultée le 2026-09-28
- Claude Code, documentation, permission modes consultée le 2026-09-28
- Anthropic Help Center, managing access to Claude's connectors consultée le 2026-09-28
- Claude.com, chat and Cowork come together for Enterprise consultée le 2026-10-24
- Claude.com, introducing Claude Duet consultée le 2026-10-24