What Is Claude AI? The Assistant, the Apps and the Models

Claude AI is the AI assistant made by Anthropic. You talk to it in plain language at claude.ai, in the Claude desktop app for macOS and Windows, or in the mobile apps for iOS and Android, and it writes, analyzes files, searches the web, runs code and makes finished documents. Around the chat sit projects for ongoing work, artifacts for things you can share, connectors to your other apps, and Cowork for longer jobs. The same Claude models power Claude Code, Anthropic’s coding agent, and the Claude API that developers build on. There is a Free plan, and the paid plans are Pro, Max, Team and Enterprise. Everything below reflects Anthropic’s own pages as of September 30, 2026.

People search for it as “Claude AI”, “Anthropic Claude” and “the Claude chatbot”; they all mean the same product. The company is covered in what is Anthropic, and how Claude stacks up against OpenAI’s assistant is in Claude vs ChatGPT.

Claude in one paragraph

Anthropic’s product overview↗ (opens in a new tab) calls Claude “more than an AI assistant: it’s a helpful, intuitive, and powerful collaborator you can put to work on real tasks, not just questions.” In practice that means two modes of use. You can ask and get an answer, the way you would with any chatbot: explain this contract clause, rewrite this email, summarize these meeting notes. Or you can hand it a task with an outcome, such as “build a spreadsheet comparing these three vendors from the attached quotes”, and it plans the steps, uses tools, and comes back with a file. The second mode is where most of the recent product work has gone.

Where you use Claude: the apps

• Web: claude.ai in any modern browser. Nothing to install.
• Desktop: the Claude app for macOS and Windows, including Windows on Arm, with a Linux version in beta. It adds local files, desktop extensions and the Code tab; Claude desktop app vs web covers what else.
• Mobile: Claude for iOS and Android, with the same account, chats and projects.
• Inside other tools: Claude in Chrome, a browser extension that can read and act on pages (Claude in Chrome), and Claude for Microsoft 365 in Office apps (Claude and Microsoft 365). Both are on paid plans.

Your chats, projects and settings follow your account, so you can start something on the laptop and check it from your phone.

What turns the chatbot into a workspace

• Projects. Anthropic’s help page What are projects?↗ (opens in a new tab) describes them as “self-contained workspaces with their own chat histories and knowledge bases.” You upload the files and write the instructions once, and every chat in the project uses them. Free accounts can create up to five; paid plans get retrieval over larger knowledge. See Claude Projects best practices.
• Artifacts. According to Anthropic’s artifacts guide↗ (opens in a new tab), an artifact is “anything Claude makes for you that you’d put in front of someone: a design, a deck, a document, a dashboard, or a small interactive tool.” They are on every plan, start private, and can be shared. Examples are in Claude artifacts examples.
• Connectors. Anthropic’s connectors page↗ (opens in a new tab) says they “let Claude access your apps and services, retrieve your data, and take actions within connected services.” Pick from a directory, or add any remote MCP server as a custom connector; Free accounts are limited to one custom connector. Claude connectors explained goes through what each kind can reach.
• Memory. Claude can remember context from past chats. It is on by default on Free, Pro and Max, off by default on Team and Enterprise until an owner turns it on, and each project keeps its own memory. Claude project memory explains the details.
• Skills. Folders of instructions Claude loads when a request matches, for procedures you repeat. See agent skills.
• File creation and code. Claude can run code in a sandbox and produce spreadsheets, slides, documents and PDFs you download. This is on the Free plan too.

Cowork: longer jobs, and a name in transition

Cowork is how you hand Claude a multi-step job and walk away. It uses the same agentic architecture as Claude Code, without a terminal: Claude plans the work, runs the steps in the cloud, reaches your local files through the desktop app when a task needs them, and hands back finished files. It is on paid plans only. The name is changing under you: Anthropic’s page Claude Cowork and chat are one Claude↗ (opens in a new tab) describes a new experience, rolling out gradually from Pro and Max, in which chat and Cowork merge into a single conversation and Claude decides whether your request needs an answer or a task. Once your account has it, you cannot switch back. What is Claude Cowork has the full picture.

Claude Code, in one line

Claude Code is the same assistant with hands inside a code repository: it reads your project, edits files and runs commands in a terminal, an IDE, the desktop app’s Code tab or the browser. It needs a paid plan or a Claude Console account. Everything else about it is in what is Claude Code.

The Claude API, for developers

If you want Claude inside your own product, you use the Claude API rather than the apps. You sign up for a Claude Console account, create an API key, and send requests to the Messages API. The API also offers batch processing, file uploads, skills and, in beta, managed agents that run in Anthropic’s sandboxes. The same models are sold through Amazon Bedrock, Google Cloud and Microsoft Foundry for teams that want their cloud provider’s billing. The API is billed by usage, separately from any app subscription. Claude API: a plain guide walks through the first request, and Claude API key covers keys.

The models behind Claude

“Claude” is also the name of the model family. Anthropic’s models overview↗ (opens in a new tab) lists four current models and suggests starting with Opus 5.5 when you are unsure:

• Claude Fable 5.1: “for demanding reasoning and long-horizon agentic work”, and the slowest. In the apps it is on every paid plan, included on Max and premium Team and Enterprise seats and used with usage credits on Pro and standard Team seats.
• Claude Opus 5.5: “for long-running agentic coding and knowledge work.”
• Claude Sonnet 5.5: “the best combination of speed and intelligence.” Anthropic released it on September 28, 2026, so Sonnet 5 is now a legacy model.
• Claude Haiku 4.5: “the fastest model with near-frontier intelligence.”

A fifth name, Claude Mythos 5.1, is invitation-only for organizations in Anthropic’s Project Glasswing, and is not something you can pick in the app. Which model suits which job is covered in Claude Sonnet vs Opus.

Is Claude AI free? The plans, by name

• Free: chat on web, desktop and mobile, web search, file creation and code execution, memory, artifacts, connectors, and up to five projects.
• Pro: more usage, projects without the cap, more models, the ability to hand off and schedule tasks, Claude Code, and Claude in Chrome and Microsoft 365. Anthropic’s plan page also lists Claude Design and Claude Science here.
• Max: everything in Pro with more usage, in two sizes, plus early access to new features.
• Team: for teams of 2 to 150, with standard and premium seats, project sharing, Claude Code and Cowork, and admin controls.
• Enterprise: Team plus advanced security controls such as SCIM, audit logs and custom retention.

Anthropic’s pricing page has the current prices; they change, so they are not repeated here. On Pro and Max, one subscription covers the apps and Claude Code, and they draw on the same usage limits, as Claude usage limits explains. Team versus Enterprise is in Claude Enterprise vs Team.

Which part of Claude for which job

The product has grown enough that the first question is often which surface to open. A quick map:

• A quick answer, a rewrite or a summary: an ordinary chat, on any device.
• Work you will come back to for weeks, such as a client account or a grant application: a project with the key files and a few lines of instructions.
• Something to show other people, like a one-page dashboard or a small calculator: ask for an artifact.
• A job with several steps that ends in a file, such as “turn these receipts into an expense report”: Cowork, or any conversation once your account has the merged experience.
• Data that lives in Google Drive, Gmail, Slack or another tool: connect it once as a connector, then refer to it in any chat.
• Changes to a codebase: Claude Code, not the chat.
• Claude inside your own app or workflow: the Claude API.

A first-week setup checklist

1. Write your Instructions for Claude in Settings: who you are, the format you like, what to avoid. Keep it under ten lines.
2. Create one project for your biggest ongoing piece of work and upload the three files you reference most.
3. Connect the one app you search every day, such as your email or your drive.
4. Check the memory setting, and use an incognito chat for anything you do not want remembered.
5. Ask for one artifact and one file, so you know what each looks like before you need it.
6. If you write code, install Claude Code and try it on a small change.

Where the list of work lives

Claude remembers conversations; it does not keep a shared list of who is doing what. fenbs is a small task board that Claude joins as a custom connector at https://fenbs.ai/api/mcp, with a browser sign-in and the scopes you tick, so it fits in the one custom connector the Free plan allows. Claude can then read the board’s rules and AI context, add a feature, enhancement or bug with a note and a plan, move it through To Do, Next Up, In Progress and Completed, and comment on what it did, and History records every change under its name. fenbs has no due dates, sprints or assignee field; it keeps the list and who changed it. How to connect Claude to your project board with MCP has the steps.