Menu
Products

Claude

5 min readbeginnerUpdated 28 Sept 2026
1 · In one line

Claude is Anthropic’s AI product for conversation, analysis and completing tasks with files, connected tools and editable outputs.

1 · What it is

Claude is Anthropic’s app for talking through problems and handing off work. You describe the result you want, and Claude maps out a plan before it starts. If you turn on permission settings, it shows the plan and waits for your go-ahead.

What Claude can reach is up to you. Connectors link it to apps such as Google Drive, Slack or Linear, but a connector only sees what your own account can see. For current facts, Claude can search the web and cite the pages it used.

Files are made by code, in a private computing space where Claude writes and runs programs. Those programs build the documents, spreadsheets and slides you get back. A person still reads the result and decides what happens next.

2 · Why it exists

Useful work needs more than a clever answer.

Context is scatteredClaude can work across files, projects, web pages and connected applications when those sources are available.
Answers need formClaude can produce editable documents, spreadsheets, slides and other artifacts rather than only a chat reply.
Review still mattersAnthropic presents the user as the person who reviews, edits and signs off on Claude’s output.
3 · How it works

Follow one task from brief to reviewed result.

How a task moves through Claude A person gives Claude a goal and files. Claude plans the steps, then reaches only the project files, connectors and web search the person allows. It writes and runs code in a private workspace to build the file, and the person reviews it and redirects if needed. ONE HANDED-OFF TASK ONLY WHAT THE PERSON ALLOWS Your goal Plan the steps Allowed access Build the file You review Project files Connectors Web search plain-language brief + files pauses to ask on your decisions your own account permissions apply runs code in a private workspace doc · sheet · deck you have final say shared across the project's chats Drive · Slack Linear · MCP current pages, with citations redirect
  1. 1 · assignYou describe a goal in plain language and provide the files or project context Claude should use.
  2. 2 · planClaude turns your goal into a plan and pauses to ask you when a decision is yours to make.
  3. 3 · accessClaude reaches only the files and tools you allow, and a connector carries over your own permissions.
  4. 4 · produceTo make a document, spreadsheet or slide deck, Claude writes and runs code in a private workspace.
  5. 5 · reviewYou inspect the work, redirect Claude and decide what is ready to use or send.

Access is part of the task: Claude can only work with the files and tools it is allowed to use.

4 · Where it's used
WhoWhat they askWhat it works with
Product manager“Group this month’s feedback into themes and show the strongest evidence for each.”Shared feedback files and authorised connectors
Finance team“Compare these two workbooks and flag numbers that do not reconcile.”Uploaded spreadsheets and the team’s instructions
Lawyer“Check this draft against our playbook and mark the clauses that need a decision.”The draft, the playbook and permitted source material
Designer“Turn this brief into an interactive prototype I can revise.”The brief, project context and artifact tools
5 · What it solves, and what it doesn't
solves
  • It connects to apps and services through connectors and can act inside them.
  • It can search the web for current facts and cite the pages it used.
  • It can create artifacts that open beside the conversation.
  • It can keep project context, instructions and files together across related work.
doesn't solve
  • The user, not Claude, has the final say.
  • A connector cannot reach a file or record that you cannot open yourself.
  • Finished files are presented for review, not as automatically approved work.
  • Consumer data settings and commercial data terms are not the same thing.
6 · Go deeper
LLMAI agentsMCPComputer usePrompt engineering

Sources used

This explainer is written in original language. The links below support its factual claims.

  1. officialThe AI for Problem Solvers | Claude by Anthropic, Anthropic · read 27 Sept 2026
  2. docsIs my data used for model training?, Anthropic · read 27 Sept 2026
  3. docsWhat are artifacts and how do I use them?, Anthropic · read 27 Sept 2026
  4. docsHow can I create and manage projects?, Anthropic · read 27 Sept 2026
  5. docsUse connectors to extend Claude's capabilities, Anthropic · read 27 Sept 2026
  6. docsEnable and use web search, Anthropic · read 27 Sept 2026
  7. docsCreate and edit files with Claude, Anthropic · read 27 Sept 2026
Suggest a correction