ANTHROPIC GUIDE

Claude & Anthropic

The AI built for nuanced thinking, code, and design

What is Anthropic for business?

Anthropic is the AI safety company behind Claude — widely regarded as the best AI model for nuanced writing, careful reasoning, and long-document analysis. The current Claude 4 family (Opus, Sonnet, Haiku) leads independent benchmarks for reasoning, coding, and agentic task completion. Claude's ecosystem has grown well beyond the chat window: Claude Code turns the terminal into a full-stack engineering partner, Claude Design produces production-quality web prototypes from plain-English briefs, and the Agent SDK + MCP let teams build their own AI agents with access to internal systems.

Claude Opus 4 leads industry benchmarks for coding, reasoning, and long-horizon agentic work
Claude Code turns plain-English instructions into shipped software — used by engineering teams worldwide
Claude Design generates fully styled, interactive web prototypes in minutes — not days
Projects, MCP, and the Agent SDK let teams build custom AI tools grounded in their own data

ECOSYSTEM MAP

The Anthropic ecosystem, mapped.

Every tool in the Anthropic suite — what it does, which plan you need, and how the pieces fit together.

TOOL GUIDE

What each tool does.

Models

Claude Sonnet 4.5

Free tier

Best for: Writing, analysis, and everyday reasoning

Anthropic's flagship general-purpose model and the best all-round performer in the Claude family. Consistently ranks at the top of independent benchmarks for writing quality, instruction-following, and nuanced reasoning. Available free with limits and via Pro for higher usage — the right default for most professional tasks.

Claude Opus 4.5

Paid / Pro

Best for: Deepest reasoning, coding, and agentic work

Claude Opus 4.5 is Anthropic's most capable model — purpose-built for complex, long-running tasks including multi-file coding, multi-step research, and agentic workflows that require sustained reasoning. The model behind Claude Code's most demanding workloads and the default choice for tasks where quality and depth outweigh cost.

Claude Haiku 4.5

Paid / Pro

Best for: Fast, high-volume processing at scale

The fastest and most cost-efficient Claude model, designed for tasks that need scale rather than depth — bulk document tagging, content classification, first-pass drafting, and API-integrated workflows requiring sub-second response times. Available via the Pro plan and the Anthropic API.

Workspace

Projects

Paid / Pro

Best for: Persistent memory for ongoing work

Projects give Claude persistent memory across conversations — upload documents, style guides, templates, and prior outputs that Claude retains throughout the project. Ideal for ongoing client engagements, content programmes, and any work that requires consistent context across multiple sessions.

Claude Design

Paid / Pro

Best for: Rapid web prototyping and design mockups

Claude Design turns plain-English descriptions into fully styled, interactive web prototypes — complete with typography, layout, imagery, and responsive behaviour. Designers, founders, and marketers use it to mock up landing pages, pitch decks, and full marketing sites in minutes. Outputs can be exported as HTML or handed to developers as a visual specification.

Artifacts

Free tier

Best for: Shareable, interactive generated outputs

Artifacts renders Claude's outputs — documents, code, charts, and structured data — as standalone, shareable interactive objects within the conversation. Useful for producing polished outputs that can be handed directly to clients or colleagues without reformatting.

Analysis Tool

Paid / Pro

Best for: Interrogating uploaded data files

The Analysis Tool allows Claude to run Python code against uploaded files — spreadsheets, CSVs, and structured data — to produce visualisations, statistical summaries, and data-driven insights directly within the conversation. Replaces basic analyst work for structured datasets.

Developer

Claude Code

Paid / Pro

Best for: Agentic coding in the terminal and IDE

Claude Code is Anthropic's flagship agentic coding tool — a command-line companion that understands entire codebases, edits multiple files simultaneously, runs tests, commits to Git, and ships real software. Engineering teams use it to accelerate delivery on bug fixes, feature builds, refactors, and migrations while keeping a human in the loop. Integrates directly with GitHub, VS Code, and most modern development environments.

Claude Agent SDK

Enterprise

Best for: Building custom AI agents on Claude

The Claude Agent SDK is the same framework that powers Claude Code — now available to developers and product teams building their own agents. It handles tool calls, memory, subagents, and long-running task loops out of the box. Used by organisations building internal AI assistants that can read from their systems, take actions, and deliver end-to-end workflows.

Claude API

Enterprise

Best for: Integrating Claude into products and workflows

The Claude API provides production-grade access to all Anthropic models with pay-per-token pricing, fine-grained model selection, and enterprise SLAs. Used by development teams embedding Claude into customer-facing applications, internal tools, and automated processing pipelines.

Claude MCP

Enterprise

Best for: Connecting Claude to your data and tools

Model Context Protocol (MCP) is Anthropic's open standard for connecting AI models to external data sources, tools, and APIs. MCP integrations allow Claude to read from databases, query internal systems, and take actions in connected tools — moving beyond conversation into genuine workflow automation. Rapidly becoming the industry default, with major vendors shipping native MCP support.

Workbench

Paid / Pro

Best for: Prompt engineering and model testing

Anthropic's Workbench is a visual environment for building, testing, and refining prompts before deploying them to production. Teams use it to systematically improve prompt quality, compare model outputs, and benchmark performance across different Claude versions.

Plans

Claude Team

Paid / Pro

Best for: Shared workspace with centralised billing

Claude Team provides higher usage limits, Projects access for all members, and centralised admin billing for teams of up to 50 users. A straightforward step up from individual Pro accounts when a team needs shared context and consistent tooling without enterprise procurement.

Claude Enterprise

Enterprise

Best for: Secure enterprise deployment with full control

Claude Enterprise adds SSO, advanced admin controls, usage analytics, expanded context windows, and a guaranteed zero data retention policy. Designed for regulated industries and large organisations requiring audit trails, role-based access, and contractual data protection.

IN PRACTICE

Anthropic at work: real scenarios.

FINANCIAL SERVICES

The situation

A private equity firm needs board-level quarterly reports for 12 portfolio companies, each requiring consistent regulatory language and the firm's house formatting style. Writing each report takes a senior analyst four hours.

The result

A Project configured with the firm's style guide, regulatory requirements, and prior report examples generates first drafts for each portfolio company in minutes. Each report requires 45 minutes of editing rather than four hours of writing. Consistency across reports improves significantly, reducing compliance review time.

MARKETING & AGENCY

The situation

A boutique branding agency wins a pitch and needs to show a prospective client three distinct landing page concepts within 48 hours. Traditional mockup production would take their design team five days and push the pitch past the deadline.

The result

The strategist briefs Claude Design on the client's positioning and three creative directions. Within four hours, three fully styled, responsive landing pages are mocked up — typography, imagery placeholders, copy, and micro-interactions included. The client picks a direction on the call and signs the engagement on the same day. Pitch-to-concept time reduces by 80%.

TECHNOLOGY & SAAS

The situation

A mid-market SaaS company's 15-person engineering team has a backlog of 40+ small bugs, UI polish items, and dependency upgrades that never reach the top of sprint planning. The accumulated debt is slowing delivery on bigger features.

The result

Engineers pair with Claude Code in the terminal — scoping each backlog ticket, letting Claude implement the change across affected files, running the test suite, and opening a PR for human review. The 40-item backlog is cleared in three sprints instead of three quarters. Senior engineering time is redirected from small fixes to architecture and hard technical problems.

PROFESSIONAL SERVICES

The situation

Partners at a management consulting firm must review 600-page due diligence bundles before client meetings — a process that takes five hours and limits how many engagements can run simultaneously.

The result

The entire document bundle is uploaded in a single conversation. Claude identifies key risk areas, cross-references inconsistencies across documents, and produces a structured executive summary with specific page references. Partner preparation time drops from five hours to under one hour.

CREATIVE SERVICES

The situation

A creative agency manages 12 brand accounts, each with a distinct tone of voice. Copywriters spend 30–45 minutes re-reading brand guidelines before starting each new brief.

The result

Each brand has its own Claude Project loaded with guidelines, sample copy, and tone rules. Writers begin any task with the brand's full context immediately active. Output quality improves measurably and client revision requests drop by 40% in the quarter following implementation.

REAL ESTATE

The situation

A commercial property agency produces 30–40 property marketing reports monthly. Each takes three to four hours to write from scratch using raw sales data and location research.

The result

Property data, comparables, and location notes are pasted into Claude. A first draft — including investment thesis, location analysis, and comparable sales narrative — is generated in minutes. Artifacts renders a structured, shareable version ready for client delivery. Production time drops to under 45 minutes per report.

EXPLORE MORE PLATFORMS