Claude
Claude is Anthropic's general-purpose AI assistant for writing, coding, data analysis and research. It runs on the web, on mobile, on the desktop and in the terminal, connects to your own tools through MCP, and keeps a permanent free plan.
What is Claude?
Claude is the AI assistant built by Anthropic PBC, an American public benefit corporation. It is a general-purpose assistant rather than a single-task tool: the same conversation can draft a document, work through a spreadsheet, write and run code, or read a long PDF and pull the argument out of it. Anthropic trains it with a method it calls Constitutional AI and presents it as an assistant meant to be safe, accurate and secure.
You reach it in several places. The web app at claude.ai is the main entrance, with native apps for iOS and Android, desktop applications for macOS, Windows (x64 and arm64) and Linux, an extension for Chrome, and integrations for Slack and Microsoft 365. Developers get a separate platform with a documented API, managed agents, web search and code execution. Behind all of it sits Anthropic's own model family: Mythos, Fable, Opus, Sonnet and Haiku.
The free plan is unusually complete. It already covers chat across web, iOS, Android and desktop, code generation, data visualisation, web search, memory that carries across conversations, file creation and code execution, desktop extensions, Slack and Google Workspace connections, remote MCP connectors, and extended thinking for harder problems. Paid plans add more usage and unlock Claude Code for the terminal, Claude Cowork for delegated and scheduled tasks, plus Claude Design and Claude Science, unlimited projects, the Research mode and access to more models.
Usage is metered rather than counted in messages. Limits reset on a rolling five-hour window, paid plans add weekly caps, and everything you do across chat, desktop, mobile and Claude Code draws from one pool. There is no fixed message quota, and Anthropic reserves the right to adjust the limits.
For organisations, the Team and Enterprise tiers add single sign-on, SCIM provisioning, audit logs, role-based access, spend controls, custom data retention and a HIPAA-ready offering, and they turn off training on customer content by default. Availability is not worldwide: Anthropic publishes a list of 184 countries and territories where Claude.ai is offered.
What it does
- Write, rewrite and edit text, from a short email to a long report
- Generate, explain and debug code, in chat or directly in the terminal with Claude Code
- Read long documents and files, then summarise, extract or restructure what is in them
- Analyse data, run code in a sandbox and produce charts
- Search the web and run deeper multi-step research
- Create finished files: documents, slide decks and spreadsheets
- Delegate longer jobs to Claude Cowork and schedule them daily, weekly or monthly
When to use Claude / When not to
A quick filter to help you decide if Claude is the right fit.
When to use Claude
- Software engineers and technical teams, who get Claude Code in the terminal with every paid plan
- Knowledge workers who write, edit and summarise all day and want one assistant across web, mobile and desktop
- Analysts and researchers who need to read long documents, run code and turn numbers into charts
- Students, teachers and academics, covered by dedicated education plans and university-wide licensing
- IT and security teams at companies, thanks to SSO, SCIM, audit logs, role-based access and a HIPAA-ready offering
When not to use Claude
- Anyone under 18, since the consumer terms set that as the minimum age
- Users in countries outside the 184 territories where Claude.ai is offered
- Organisations that require self-hosting or an on-premise deployment, which Anthropic does not offer
- Consumers who want their content excluded from model training without touching a setting, as training is on by default and opt-out is manual
- Teams that need a guaranteed, contractually fixed message volume, because usage limits are elastic and set at the vendor's discretion
How to use Claude
A typical end-to-end flow, from setup to results.
- Open claude.ai and create an account with Google, an email address or your company SSO
- Start on the free plan, or pick Pro, Max, Team or Enterprise depending on how much you expect to use it
- Install the desktop app for macOS, Windows or Linux, and the mobile app from the App Store or Google Play
- Type your question, or drop a document, image or spreadsheet straight into the conversation
- Turn on web search or extended thinking when the task needs current information or deeper reasoning
- Connect your own sources: Slack, Google Workspace, Microsoft 365 or any remote MCP connector
- Group related chats and files into projects to keep long-running work together (paid plans)
- Run Claude Code in your terminal for anything that touches a real codebase
- Hand longer jobs to Claude Cowork, ask for the output format you want, and schedule it to repeat
- Watch your consumption under Settings > Usage, and switch on usage credits at API rates if you hit a limit
Pros & Cons
Pros
- A permanent free plan that already includes code generation, web search, memory and code execution
- Genuinely broad reach: web, mobile, desktop, terminal, browser, Slack and Microsoft 365, plus a public API
- Strong compliance posture on paper: SOC 2, ISO 27001, GDPR and CCPA, with a HIPAA-ready offering
- Training opt-out is available to everyone, and Team and Enterprise plans exclude training by default
- Subprocessors are published, and the privacy policy is specific about controllers, transfers and rights
- Anthropic assigns you its rights in the outputs, so what you generate is yours to reuse
- A very well-capitalised vendor, with USD 65 billion raised in its Series H round
Cons
- Usage limits are never expressed as a number of messages and can be tightened at the vendor's discretion
- On consumer plans, conversations feed model training unless you find and switch off the setting yourself
- Claude.ai is offered in 184 countries and territories only, so availability is not universal
- The website interface is published in six languages, which is narrow for a product sold worldwide
- Enterprise pricing mixes a seat fee with usage billed at API rates, so the monthly bill is hard to forecast
- Payments are generally non-refundable outside the cases set by the terms or by local law
- There is no self-hosted or on-premise option: everything runs as a hosted service
Pricing & Plans
A permanent free plan is available at no cost. The cheapest paid tier is Claude Pro at USD 17 per month when billed annually, which is invoiced as USD 200 up front, or USD 20 per month when billed monthly. Max starts at USD 100 per month. For organisations, a Team standard seat is USD 20 per seat per month billed annually, and Enterprise is USD 20 per seat per month plus usage charged at API rates. Prices exclude applicable tax and may be changed at Anthropic's discretion.
- Free — USD 0 — chat on web
- iOS
- Android and desktop
- code generation
- data visualisation
- web search
- memory across conversations
- file creation and code execution
- Pro — USD 17 per month billed annually (USD 200 up front) or USD 20 monthly — everything in Free plus more usage
- Claude Code
- Claude Cowork
- Claude Design
- Claude Science
- unlimited projects
- Research
- more models
- Max — from USD 100 per month — everything in Pro plus 5x or 20x the usage
- higher output limits
- early access to new features and priority at busy times
- Team — USD 20 per standard seat per month billed annually (USD 25 monthly)
- or USD 100 per premium seat per month billed annually (USD 125 monthly)
- for teams of 2 to 150 — SSO
- central billing and administration
- enterprise search
- connector controls
- desktop deployment
- no training on your content by default
- Enterprise — USD 20 per seat per month plus usage at API rates
- billed annually — everything in Team plus role-based access
- SCIM
- audit logs
- a compliance API
- custom data retention
- network access control
- IP allowlisting
- Education — a university-wide plan for students
- faculty and staff at discounted rates
- with dedicated API credits and a learning and academic research mode
- API — Opus 5 at USD 2 per million input tokens and USD 10 per million output tokens (introductory rate until 31 August 2026
- then USD 3 and USD 15)
- Haiku 4.5 at USD 1 and USD 5 per million tokens
Data, GDPR & hosting
A consolidated view of how Claude handles your data.
GDPR overview
GDPR compliance is claimed in writing. The enterprise page lists SOC 2, ISO 27001, GDPR and CCPA compliance, and the privacy policy, effective 8 July 2026, works through the regime in detail. Anthropic Ireland, Limited is the controller for the EEA, the UK and Switzerland; Anthropic PBC is the controller elsewhere. A Data Protection Officer is reachable at dpo@anthropic.com and rights requests go to privacy@anthropic.com. Access, rectification, erasure, objection and portability are all described, with a one-month response deadline under the EU and UK GDPR, extendable by two months for complex requests. Transfers outside the EEA are covered by legal derogations, subprocessors are published in the trust centre, and the right to complain to a supervisory authority is set out. No Article 27 representative is named, which is consistent with an Irish establishment.
Who owns the data?
The terms group your prompts and Claude's responses under a single label, Materials. You keep whatever rights you already hold in the inputs you submit, and Anthropic assigns to you all the right, title and interest it may have in the outputs, provided you comply with the terms. The service itself stays the property of Anthropic and its providers. Anthropic may use Materials to run and improve the service and to train its models unless you opt out in your account settings. Two exceptions survive an opt-out: content you submit as feedback, and content flagged for safety review. The controller is Anthropic Ireland, Limited in the EEA, the UK and Switzerland, and Anthropic PBC everywhere else.
Reuse rights
Yes. Because Anthropic assigns you its rights in the outputs, you can reuse what Claude produces without asking for further permission, including commercially. Two limits apply. First, the Usage Policy still governs what you may do with the service, and you remain responsible for holding the rights to whatever you feed in. Second, the terms are explicit that outputs may be inaccurate even when they look precise, and ask you to confirm them independently before relying on them. Feedback you send is treated separately: Anthropic may use it freely and owes you nothing for it.
Data retention & training
Hosting summary
Anthropic hosts in the United States. The privacy policy states that when you use the service your personal data is transferred to Anthropic's servers in the US, or to other countries outside the EEA and the UK, in order to deliver the service, train models or conduct research. Responsibility is split by region: Anthropic Ireland, Limited is the controller for the EEA, the UK and Switzerland, and Anthropic PBC is the controller everywhere else. Transfers to countries without an adequacy decision rely on derogations provided for in applicable law, and the list of third parties involved is published in the trust centre at trust.anthropic.com/subprocessors. On the API side, US-only inference is offered as a data residency option at 1.1x the standard token price. Custom retention controls are limited to Claude Enterprise. One caveat on the network data: the domain resolves to an anycast address geolocated in the United States and announced by Anthropic's own network, which identifies an edge node rather than a storage location.
Where Claude works
Country-level availability.
Available in
Not available in
Things to keep in mind
Risks and trade-offs to weigh before adopting Claude.
- Anthropic itself warns that outputs may contain material inaccuracies even when they read as precise and detailed; treat anything factual as a draft to verify
- Delegating your reasoning day after day is a real risk to your own skills: the assistant is at its best when it argues with you, not when it replaces you
- On consumer plans, training on your conversations is the default, and turning it off is a manual step buried in the settings
- Even after opting out, content you submit as feedback or content flagged for safety review can still be used for training
- If you sign up with a work email, your account may be linked to your employer's organisation, whose administrator can then monitor and control it
- Your data travels to servers in the United States and to countries without an adequacy decision, which matters if you handle regulated material
- Cancelling a subscription does not delete anything: your chats, projects and files stay attached to the account until you remove them
Setup & Integrations
Technical difficulty
Very low for ordinary use. You sign up with Google, an email address or SSO and can start typing immediately; the desktop and mobile apps are standard installs. Difficulty rises with ambition: connectors and remote MCP need configuration, Claude Code assumes you are comfortable in a terminal, and the API requires a separate Console account. Rolling it out across a company is a genuine IT project, since SSO, SCIM provisioning, role-based access and IP allowlisting all have to be set up and maintained.
Deployment
Apps stores
Integrations
Supported languages
Behind Claude
Fundraising
Social
Resources
All the official URLs gathered for verification and reference.
Frequently asked questions
What is Claude?
Is there a free plan?
How much does Claude cost?
How do the usage limits work?
Is Claude Code included?
Are my conversations used to train the models?
Where is my data stored, and how long is it kept?
Which apps and integrations are available?
Is Claude available everywhere, and is there an age limit?
Can I cancel, and will I get a refund?
Should you pick Claude?
Claude is one of the few AI assistants that is genuinely usable end to end without paying: the free plan already carries web search, memory, file creation and sandboxed code execution, which is more than most competitors give away. Paying mainly buys headroom and the surrounding products, chiefly Claude Code in the terminal and Claude Cowork for delegated, scheduled work. At USD 17 per month on an annual plan, the entry point is in line with the rest of the market.
The governance story is better documented than average. Anthropic names its controllers, publishes its subprocessors, claims SOC 2, ISO 27001, GDPR and CCPA compliance, offers a HIPAA-ready configuration, and assigns you its rights in whatever Claude produces. The training position deserves attention: on consumer plans your conversations feed the models unless you go and switch that off, and even then feedback and safety-flagged content remain in scope. Team and Enterprise plans reverse that default.
The real friction is elsewhere. Usage is elastic rather than contractual, expressed as rolling five-hour windows and weekly caps rather than a number you can plan around, and Anthropic can adjust it. Availability stops at a published list of 184 countries. There is no self-hosted option. Enterprise billing mixes seats with usage at API rates, which makes forecasting awkward.
For an individual who writes, codes or analyses for a living, it is an easy tool to adopt and a cheap one to trial. For an organisation, the admin controls are serious and the compliance claims are explicit, but the usage model and the geographic restrictions are worth checking against your own constraints before committing.
- Choosing a selection results in a full page refresh.
- Opens in a new window.