Claude Terms Decoded: What Beginners Actually Need to Know

You opened Claude. It asked what you needed. You typed something. It answered.

Then someone said "agentic" and you smiled and nodded and had no idea what they meant.

Fine. Here is the decoder ring.

Claude

Claude is Anthropic's AI assistant. You can use it in a web browser, desktop app, mobile app, and other Anthropic tools. When someone says, "I asked Claude," this is what they mean.

Anthropic

Anthropic is the company that builds Claude. It publishes the models, the Claude apps, Claude Code, and tools for developers and organizations.

Claude Chat, Cowork, and Claude Code

Claude Chat is the web, desktop, or mobile conversation. You ask questions, work through ideas, analyze files, and create drafts.

Claude Cowork runs through the desktop app and can carry out multi-step knowledge work across local files and connected applications. Anthropic says it is available on paid plans. You give it an outcome, review the work, and keep consequential decisions with a person.

Claude Code is built for software work. It can read and change code, run commands, and work across a project. It is useful for developers and for people building software with AI.

Read Anthropic's current Cowork overview.

Agentic

Agentic means the AI can take several steps toward a goal instead of answering one question and stopping.

It still needs limits. What can it access? What may it change? What needs approval? "Handle it" is not a safety plan.

Context window

The context window is the working material available to Claude for the current task. That may include your messages, Claude's responses, uploaded files, tool results, and system instructions.

It is large, not infinite. Long conversations can lose earlier detail or make the important information harder to find. A fresh conversation with a clean brief is often better than dragging one thread into next Thursday.

Tokens

Tokens are the small pieces of text an AI model reads and writes. Token limits affect how much material fits into a task and how some developer services are priced.

You do not need to count them during normal use. You do need to know that fifty files and a twelve-month conversation are not free storage.

Skills

A skill is a reusable set of instructions, examples, and sometimes tools for a specific kind of work. A voice skill can teach Claude how you write. A reporting skill can define the inputs, checks, and format for a weekly report.

A skill does not make the system infallible. It gives the work a repeatable starting point.

MCP and connectors

MCP stands for Model Context Protocol. It is an open protocol that lets AI applications connect to outside tools and data through MCP servers. In Claude, you may see these connections described as connectors or desktop extensions.

The protocol is the connection standard. Gmail, Slack, a file system, or another service is the thing being connected. Permissions still matter.

Read Anthropic's MCP documentation.

The part that matters

You do not need the full vocabulary before you begin. Know which Claude product you are using. Know what it can access. Know what it is allowed to do. Keep a person responsible for the final decision.

That is enough to start without nodding through the next conversation.