Claude Code for Business: Run 90% of Your Business with AI Agents

Terms worth knowing.

Claude Code

Anthropic's command-line coding agent that reads local files, runs commands, and executes multi-step tasks based on plain-English instructions rather than requiring written code.

Agent

An AI system that uses a language model to interact with its environment — files, tools, the web — to achieve a user-defined goal by reasoning, planning, and taking actions on its own.

MCP (Model Context Protocol)

An open standard that lets an AI agent connect to external tools and services like Notion, YouTube, or databases, so it can read and write real data instead of only producing text.

API key

A unique string that authenticates a program when it calls a third-party service, granting it permission to fetch data or perform actions on the account it belongs to.

LangGraph

A Python framework for building stateful, multi-step AI agent workflows in code. Powerful but requires engineering time most non-developers don't have.

Google ADK

Google's Agent Development Kit, a code-first toolkit for building and orchestrating AI agents on Google's stack.

CrewAI

An open-source Python framework for orchestrating teams of role-based AI agents that collaborate on tasks.

n8n / Zapier

Visual workflow automation tools where you wire together triggers and actions node by node. Reliable, but every step must be explicitly designed by the builder.

Slash command

A reusable shortcut in Claude Code, typed as /name, that runs a saved prompt or workflow on demand instead of retyping instructions each time.

CLAUDE.md

A markdown file at the root of a project that Claude Code reads as persistent memory — routing rules, folder layout, conventions — so every agent session starts with the same context.

Hooks

Configurable rules in Claude Code that run before or after tool calls, used as guardrails to block unsafe actions or trigger follow-up behavior like updating documentation automatically.

Agent skills

Packaged playbooks stored as folders of instructions and assets that an agent loads on demand when a task matches the skill's description, keeping the rest of the context clean.

Sub-agents

Secondary Claude agents spawned by a main agent to handle subtasks in parallel. Each gets a fresh context window, so they avoid degrading the orchestrator's performance.

Context window

The total amount of text — instructions, files, conversation — an AI model can consider at once. Performance degrades as the window fills up, even before the hard limit.

Compacting

When a chat session nears its context limit, the model summarizes earlier messages to free up space. Repeated compacting causes loss of detail from the original request.

Context engineering

The practice of deliberately structuring what information an AI agent sees — and when — to keep its working context relevant, lean, and accurate.

Orchestrator agent

A main agent whose job is to plan work, delegate tasks to sub-agents, and review their output rather than executing the work itself.

Claude Max plan

Anthropic's highest-tier consumer subscription, intended for users who run heavy parallel workloads like spawning many sub-agents without hitting weekly usage caps.

Claude Opus / Sonnet / Haiku

The three Claude model sizes from Anthropic. Opus is the most capable, Sonnet is the balanced default, and Haiku is the fastest and cheapest for lightweight tasks.

Gemini CLI

Google's command-line interface for using Gemini models as a coding and task agent — a direct alternative to Claude Code that follows the same patterns.