Updated July 2026
Everything you need to know about Claude, Anthropic's AI assistant — what it actually is, how to prompt it well, every plan and price compared, the key settings explained, and the fastest way to get useful answers without wasting time scrolling docs.
Contents
- What is Claude?
- Getting Started
- Prompting Claude Well
- Prompt Types & What to Expect
- Plans Compared
- Settings — Explained
- Tips & Power Moves
- Common Mistakes
- Useful Claude Links
1. What is Claude?
Claude is Anthropic's family of AI models and the assistant built on top of them, available at claude.ai and through apps on web, iOS, Android, and desktop. Claude is built around being genuinely helpful, careful, and honest — Anthropic designs it with safety and reliability as first-class goals, not an afterthought.
The current model lineup spans four tiers: Haiku (fastest and most cost-efficient), Sonnet (the balanced, high-performance default for everyday work and coding), Opus (built for complex agentic and enterprise tasks), and Fable, the newest tier for long-running agent work. Above Opus sits Anthropic's newer Mythos tier, which is not yet broadly available to the public.
Quick orientation
- Multimodal by design — Claude can read text, images, PDFs, and code in the same conversation.
- Built for real work — Artifacts let you generate and edit documents, code, and slides right inside the chat; Projects keep related conversations and files organised.
- Agentic tools — Claude Code lets you delegate coding tasks from the terminal or desktop; Claude Cowork extends that to general knowledge work like spreadsheets, research, and documents.
- Connected to your tools — Remote MCP connectors let Claude read from and act on services like Google Workspace, Slack, and Microsoft 365 when you allow it to.
- Not always right — like every LLM, Claude can make mistakes or state things with more confidence than is warranted. Verify anything important.
2. Getting Started
Creating your account
- Go to claude.ai
- Sign up with an email address or Google account — no card required
- You land on the Free plan immediately
- To upgrade, open your profile menu and select Upgrade plan, or visit claude.com/pricing
Tip: Claude is also available through the Claude API and Claude Platform for developers, and through beta browser/office agents like Claude in Chrome, Claude in Excel, and Claude in PowerPoint.
The Claude interface — explained
| Element | Location | What it does |
|---|---|---|
| New chat | Sidebar top | Starts a fresh conversation with no prior context. |
| Projects | Sidebar | Groups related chats and files together with shared context and instructions. |
| Artifacts | Auto-opens when triggered | A side panel for viewing and iterating on documents, code, slides, or visuals Claude creates. |
| Research | Tools / message bar | Sends Claude to search across many sources and return a structured, cited report. |
| Claude Code | Terminal / desktop / mobile | An agentic coding tool for delegating development tasks directly from the command line. |
| Claude Cowork | Desktop app | An agentic workspace for non-developers — spreadsheets, documents, and multi-step knowledge work. |
| Model selector | Top bar | Switch between available models depending on your plan and the task at hand. |
| Message box | Bottom centre | Type your prompt. Attach files or images with the paperclip icon. |
| Share | Top right | Generates a link to share the conversation. |
3. Prompting Claude Well
Claude responds best to the same discipline as any capable assistant: a clear role, a clear task, real context, and a stated format. Because Claude can reason through multi-step problems and hold a lot of context, it also rewards you for thinking out loud — telling it how to approach a problem, not just what you want.
The 5-part prompt formula
- Role — Who should Claude be? ("Act as a senior product analyst")
- Task — What exactly do you want? ("Compare these two pricing models")
- Context — Give background or attach the source ("for an Indian D2C brand, using the attached spreadsheet")
- Format — How should the output look? ("table format, under 150 words")
- Constraint — What should it avoid? ("no jargon, keep it actionable")
Email writing
❌ Weak prompt:
Write an email about the delay.
✅ Strong prompt:
Act as a project manager. Write a polite client email explaining a 2-week delay caused by a vendor API issue. Apologise, give a new delivery date of July 15, and offer a 10% discount. Warm but professional tone, under 150 words.
Spreadsheets & data
❌ Weak prompt:
Formula for sales total
✅ Strong prompt:
In this sheet, column A has dates (DD/MM/YYYY), B has salesperson names, C has sale value. Write a formula that sums June 2026 sales only for "Zain." Explain each part of the formula.
Research
❌ Weak prompt:
Research my competitor
✅ Strong prompt:
Use Research to analyse Competitor X's pricing, positioning, and recent marketing campaigns over the last 6 months. Return a structured report with sources, organised by pricing, messaging, and weaknesses I can target.
Coding
❌ Weak prompt:
Fix my code
✅ Strong prompt:
Here's a Python function that's throwing an IndexError on empty lists (attached). Explain why it fails, then fix it and add a guard clause with a clear error message.
4. Prompt Types & What to Expect
Response time and quality depend on which model you're using, plan tier, and task complexity. As a rough guide:
| Prompt type | Example task | Typically fastest on | Available on |
|---|---|---|---|
| Simple Q&A | "What is inflation?" | Haiku | All plans |
| Short email | 100-word reply | Haiku / Sonnet | All plans |
| Long document draft | 400+ word formal letter | Sonnet | All plans |
| Article / blog post | 1000-word SEO post | Sonnet | All plans |
| PDF summarisation | 5-page document | Sonnet | All plans |
| Simple code | 20-line function | Sonnet | All plans |
| Complex code / agentic coding | Multi-file refactor | Opus, via Claude Code | Pro and above |
| Deep research report | Multi-source, cited research | Sonnet / Opus | Pro and above |
| Long-running agent tasks | Multi-step document or data work | Fable / Opus | Pro and above |
| Extended thinking | Complex multi-step reasoning | Any model, thinking mode on | All plans, higher limits on paid |
⚠️ Note: These are general guidelines, not guarantees. Actual speed and output depend on load, prompt length, and which model your plan gives you access to. Always check claude.com/pricing for the current model lineup.
5. Plans Compared
Important: Anthropic has adjusted Claude's pricing and included features more than once in 2026. The figures below reflect the latest published pricing at the time of writing. Always confirm current pricing and country availability at claude.com/pricing before subscribing.
Free — $0/mo
- Chat on web, iOS, Android, and desktop
- Generate code, visualize data, create files
- Web search and memory across conversations
- Connect Slack and Google Workspace via remote MCP
- Extended thinking for complex work
Best for casual, everyday use and trying Claude out.
Pro — from $17/mo (billed annually), $20/mo billed monthly
Everything in Free, plus:
- More usage
- Includes Claude Code and Claude Cowork
- Includes Claude Design
- Unlimited Projects
- Access to Research
- Access to more Claude models
- Claude for Microsoft 365 and Outlook
The most common upgrade path — the sweet spot for daily, professional use.
Max — from $100/mo
Everything in Pro, plus:
- Choose 5x or 20x more usage than Pro
- Higher output limits for all tasks
- Early access to advanced features
- Priority access during high-traffic times
Built for power users who regularly hit Pro's usage limits.
Team — from $20/seat/mo (annual), 5–150 seats
- Standard seat: $20/mo annual ($25 monthly) — more usage than Pro
- Premium seat: $100/mo annual ($125 monthly) — 5x the usage of a Standard seat
- Central billing, SSO, admin controls, enterprise search
- No model training on your content by default
- Mix and match seat types across your organisation
Enterprise — custom pricing
- Seat price plus usage billed at API rates
- SCIM, audit logs, compliance API, HIPAA-ready offering
- Role-based access, IP allowlisting, network-level controls
- Dedicated support for large-scale deployments
Which plan is right for you?
| Use case | Free OK? | Pro recommended? | Max/Team worth it? |
|---|---|---|---|
| Occasional Q&A or writing help | ✅ Yes | Not needed | No |
| Daily work: docs, code, research | ⚠️ Limited | ✅ Recommended | Only if heavy |
| Regular deep research / analysis | ❌ Runs out fast | ✅ Yes | For high volume |
| Serious agentic coding (Claude Code) | ❌ Not included | ✅ Yes | Max, for heavy use |
| Rolling out across a team | ❌ No | Individual only | ✅ Team or Enterprise |
Bottom line: start on Free. If you use Claude for work daily, or you're running out of usage or Research access, Pro is the sweet spot. Only move to Max if you're hitting Pro's limits regularly, and move to Team or Enterprise once you need shared access and admin controls across a group.
6. Settings — Explained
Open Settings from your profile menu (top left or top right, depending on device). Here's what the key sections actually do:
| Setting | What it does |
|---|---|
| Appearance | Light, dark, or system default theme for the interface. |
| Memory | Lets Claude reference facts and preferences from past conversations. Can be reviewed, edited, or turned off. |
| User preferences | Set your preferred tone, formatting style, and general instructions Claude should follow across chats. |
| Search and reference past chats | Allows Claude to look back at earlier conversations when relevant to a new one. |
| Connectors | Manage which external tools and services (Google Workspace, Slack, and more) Claude can access via MCP. |
| Web search | Toggle whether Claude can search the web for current information. |
| Code execution and file creation | Enables Claude to run code and generate downloadable files like spreadsheets, slides, and documents. |
| Privacy controls | Manage whether your conversations are used to help improve Claude's models, and delete chat history. |
Quick link: manage data and privacy settings any time from your account settings on claude.ai.
7. Tips & Power Moves
⚡ Power moves most people don't know
- Use Projects — group related chats and files under one space with shared instructions, so you're not re-explaining context every time.
- Lean on Artifacts — ask Claude to revise a specific section of a document or piece of code instead of regenerating the whole thing.
- Attach files directly — Claude reads PDFs, spreadsheets, and images natively; there's rarely a need to retype content.
- Use Research for anything multi-source — it's far more thorough than a single chat reply for competitive or market research.
- Chain your prompts — ask for an outline first, then "now expand section 2 in full detail," rather than one giant prompt.
- Set user preferences once — tone, formatting, and standing instructions saved in Settings apply across every new chat.
✨ Power prompt: turn Claude into your daily ops assistant
From now on in this chat, act as my operations assistant. I work in MIS and digital operations for an automotive group in India. Keep responses in concise bullet points, practical and India-context where relevant, and end with "Next step?" to keep momentum. Ready?
8. Common Mistakes to Avoid
| Mistake | Why it hurts | Fix |
|---|---|---|
| Vague one-line prompts | Generic, low-value output | Use the 5-part formula: Role, Task, Context, Format, Constraint |
| Starting a new chat for every follow-up | Loses all context | Keep the thread going in the same chat, or use a Project |
| Trusting facts blindly | Claude can still get things wrong | Verify anything important, especially numbers, dates, and quotes |
| Never using Projects for repeat work | You keep re-explaining the same context | Set up a Project once with instructions and files |
| Staying on Free for serious daily work | Usage and Research limits run out fast | Upgrade to Pro if you use Claude regularly for work |
| Ignoring the model selector | Wrong model for the task | Match the model to the task's complexity where your plan allows a choice |
| Pasting sensitive data carelessly | May be retained per your privacy settings | Review your privacy and memory settings for confidential work |
9. Useful Claude Links
- Claude App — Start chatting at claude.ai
- Plans & Pricing — Compare Free, Pro, Max, Team, Enterprise
- Developer Docs — Claude API documentation
- Claude Platform Console — Build with the Claude API
- Support Center — Official help and how-to docs
- Claude Code — Agentic coding from the terminal
- Claude Cowork — Agentic desktop app for knowledge work
- Claude Blog — Latest feature and model updates
