All of Anthropic's products, explained

Anthropic is the company that develops Claude, the language model behind everything that follows. From there it has built a family of tools for using it in different contexts, from writing an email to programming an entire product. This guide goes through Anthropic’s products one by one: what each one is, who it’s for, and when we use them with clients at Galileo14.

What is Claude?

Claude is the model and the assistant behind everything else on this list. There are several versions, each trading capability for cost. Claude Fable 5 is the most capable, with a 1M-token context window and adaptive reasoning for long, complex tasks. Claude Opus 4.8 sits at the same top tier. Claude Sonnet 5 is the default model on the Free and Pro plans, built for everyday use. Claude Haiku 4.5 trades some capability for speed and low cost on simple, high-volume tasks.

You reach Claude through the web (claude.ai), desktop apps (Mac, Windows and, in beta, Linux) and mobile, all synced to one account. Plans range from Free to Pro, Max, Team and Enterprise, depending on usage volume and whether it’s for one person or an organization. At Galileo14 this is where every engagement starts: before automating anything, a team has to know how to use Claude well in a normal chat.

What is Claude Code?

It’s the question development teams ask us most often. Claude Code is an agentic coding tool that runs in the terminal (CLI), with integrations for editors, desktop and browser. It reads an entire codebase, edits across multiple files, runs tests, handles git and opens pull requests, without someone going file by file. It’s built for developers who want to hand off a full task (migrate a module, track down a bug) and review the result at the end, not every line along the way.

It’s the tool we implement most with technical teams: turning a repeated workflow (a deployment, a style review) into a reusable skill is usually one of the first improvements a team notices, and it’s exactly what we work on in the Claude Code training for development teams. This article explains what Claude Code is with more examples of installation and workflow.

What is Claude Cowork?

Claude Cowork is easier to understand by contrast with Claude Code: where Claude Code works on code, Cowork works on the rest of a company’s work. You hand it a task and it acts autonomously across files, calendar, email, messaging and other connected tools, splitting large jobs so they run in parallel. It reached desktop in January 2026 and is now rolling out in beta to web and mobile, starting with the Max plan. It’s used mostly in finance, HR, admin and content, not in programming.

For non-technical roles (operations, marketing, finance) it’s usually the entry point into Claude, since it doesn’t require writing code. In our experience, it works best when the first task is scoped tightly: people who delegate too much at once take longer to trust the result than people who start with one narrow, weekly task. This article on what Claude Cowork is goes into the detail.

What is Claude Design?

Claude Design came out of Anthropic Labs in April 2026 and turns a conversation into visual work: prototypes, mockups, presentations exportable to PPTX, one-pagers, marketing pieces. It reads existing code and design files to build a reusable design system, then hands that work to Claude Code through /design-sync to move it into production. It’s in research preview, included on the Pro, Max, Team and Enterprise plans, and runs on an Opus-tier vision model.

It’s built for designers, founders, product managers and marketers who need a visual piece fast without opening a full design tool. It’s still early to say whether it replaces established design tools in complex workflows; for quick prototypes and decks, it does shorten the path.

What is Claude Platform?

Claude Platform (the Claude Developer Platform) is the surface for anyone integrating Claude into their own product: the Messages API, the Console (keys, usage, workspaces), the Workbench for testing prompts, prompt caching, the Batch API, tool use and MCP support. It’s also where managed agents get deployed: agents that run in a managed sandbox on Anthropic’s servers, so the company doesn’t have to maintain that infrastructure itself.

It’s the piece development teams reach for when they want Claude built into their own application, rather than used through a chat window. In custom integration work (an internal endpoint, an automated flow with no human in the loop), that’s where the development work we do for clients ends up.

What is Claude Agent SDK?

Claude Agent SDK is the open-source framework for building production agents on the Claude API, in Python or TypeScript. It exposes the same agent loop Claude Code uses: tool management, context management and subagents. Instead of rebuilding that logic from scratch, a development team reuses it directly.

We use it when a client needs an agent with behavior that doesn’t fit either Claude Code or Cowork out of the box, for example an internal agent with access to specific company systems.

Model Context Protocol (MCP)

Model Context Protocol, published by Anthropic in November 2024, is an open standard for connecting an AI model to external tools and data through MCP servers and clients. It isn’t exclusive to Claude: the industry has adopted it as a common way to give an agent access to tools without writing a separate integration for each one.

In practice, this is what lets Claude Code or Cowork read a ticketing system, a database or a CRM without custom-coding that connection every time. We cover how to set it up on the MCPs page.

Agent Skills

Agent Skills are an open format: a folder with a SKILL.md file (name, description, instructions) plus optional scripts or templates. An agent loads only the skills a given task needs, instead of carrying all available context all the time.

It’s how we document a process a team repeats (a review, a report template) so that anyone in Claude Code or Cowork can call it up later without explaining it from scratch each time.

Claude in Chrome

Claude in Chrome is a browser extension: Claude reads pages, clicks around, navigates sites, fills forms, extracts data and runs multi-step workflows. It’s in beta for paid subscribers and only available on Chrome for now.

It fits tasks that live in web interfaces with no API available: filling in a form on a supplier portal, pulling data out of a tool that doesn’t expose it any other way. Where an API integration exists, that’s still the better option; this extension covers the gap when there isn’t one.

Claude in Slack and Claude Tag

Claude has been available inside Slack for a while, but June 2026 brought Claude Tag, the agentic version: it joins a channel as a teammate you @-tag to delegate tasks, instead of opening a separate chat.

For teams that already live inside Slack, moving part of the work with Claude there cuts down on switching tools constantly. It’s still early to say whether it replaces direct use of Claude Cowork for longer tasks.

Frequently asked questions about pricing

How much does Claude Code cost?

Claude Code doesn’t have its own price: you pay through a Claude plan (Pro, Max or a Team/Enterprise seat, with usage included in the plan’s limits) or per token if you go through the API. A team with good usage habits can stay in the tens of euros per person per month; without those habits, spend climbs fast.

How much is Claude for business?

It depends on the plan: Anthropic has per-user pricing for team and enterprise, plus token usage on top if you connect Claude to your own systems through the API. Our fee doesn’t replace that bill, it’s separate: we help pick the right plan and keep usage from running away.

Do I have to pay to use Claude Code?

Yes, for team use: Claude Code runs on a paid Claude plan (Pro, Max, or a Team/Enterprise seat) or on API usage. It’s the starting point for daily use, not an optional add-on.

Why is Claude so expensive?

It isn’t expensive on its own, it depends on which model you pick for each task. Using the most powerful model for something simple is what drives the bill up; matching the model to the task brings the cost down a lot. Most overspend comes from that choice, not from Claude’s price.

How the pieces fit together

The natural order for almost any company is: start with Claude in a normal chat, so the team understands what it can and can’t do. From there, depending on the type of work, it’s Claude Code for development or Claude Cowork for the rest of the operation.

MCP and Agent Skills are what connect those tools to each company’s own systems and processes. Claude Platform and the Agent SDK come in once the goal stops being “use Claude” and becomes “build something on top of it.”