Welcome to Claude

When you ask Claude to create substantial content — a document, code, a chart, a diagram — it appears in a dedicated panel alongside the conversation. This is an Artifact.

What Artifacts support:

• Documents and reports (rendered Markdown)
• Code (syntax-highlighted, with a copy button)
• React components and HTML pages (live preview!)
• SVG graphics and diagrams
• Mermaid flowcharts and sequence diagrams

Key advantage: You can iterate on artifacts conversationally. Say "make the table wider" or "add a section on risk factors" and Claude updates the artifact in place. You're editing collaboratively without switching tools.

Claude can search the web to find current information. This is useful for research, fact-checking, looking up recent events or documentation, and finding competitors' latest moves.

Claude will automatically search when it determines the question needs current information, or you can explicitly ask it to "search for" something.

Upload files directly into any conversation. Claude can read and reason about:

• PDFs — contracts, reports, regulatory documents, invoices
• Images — screenshots, diagrams, photos of whiteboards, charts
• Spreadsheets — CSV, Excel files for data analysis
• Code files — any programming language
• Documents — Word docs, text files, markdown

You can upload multiple files at once and ask Claude to compare, summarize, extract data, or answer questions across all of them.

Styles let you control how Claude writes. Instead of re-explaining your preferences every time, set a style and Claude will consistently match it.

Built-in styles: Concise, Explanatory, Formal — quick presets for common needs.

Custom styles: You can create your own. Paste a sample of writing you like and Claude will learn from it, or describe the tone, structure, and conventions you want. Useful for maintaining a consistent brand voice across the team.

Your Teams plan includes connectors for the tools you already use:

• Google Drive — search and reference docs, sheets, and slides
• Gmail — search emails for context
• Google Calendar — check schedules and meeting context
• Slack — reference channel discussions and messages
• Microsoft 365 — OneDrive, Outlook, Teams integration
• GitHub — pull in code repositories and issues

These connectors let Claude pull in context from your existing work without you needing to copy-paste or upload manually.

Teams plan includes a dedicated search project that provides unified knowledge access across your connected tools. Think of it as an internal search engine powered by Claude — ask a question and it searches across your Google Drive, emails, Slack messages, and other connected sources to find the answer.

For complex problems, Claude can use "extended thinking" — a mode where it reasons through the problem step-by-step before responding. This is especially useful for:

• Complex analysis or multi-step problems
• Code debugging and architecture decisions
• Legal or regulatory interpretation
• Mathematical or financial calculations

You can see Claude's thinking process, which also helps you verify its reasoning.

For complex problems, ask Claude to reason through it before giving an answer:

"Think through this step by step before giving me your recommendation."

"Walk me through your reasoning, then provide the final answer."

This dramatically improves accuracy on multi-step problems, financial calculations, and strategy questions.

Tell Claude exactly how to structure its response:

"Format your response as: Summary (2-3 sentences), Key Findings (bullet points), Recommended Actions (numbered list with owner and deadline for each), Risks (brief paragraph)."

Claude is excellent at following format specifications. If you need JSON, CSV, a table, or a specific template — just describe it or paste an example.

Claude is trained to be helpful, which sometimes means it'll give you what you asked for even if the ask isn't quite right. Give it permission to challenge you:

"If you think my approach has flaws, tell me directly before proceeding."

"If this question is based on a wrong assumption, correct me."

This is especially valuable for strategy, technical architecture, and financial decisions.

Search for "Claude Code" in the VS Code Extensions view (Cmd+Shift+X). Install, then open Command Palette (Cmd+Shift+P) → "Claude Code: Open in New Tab".

The extension provides inline diffs, @-mentions for files and symbols, plan review, and full conversation history — all without leaving your editor. Also works in Cursor.

The VS Code extension is one of the most popular agentic AI tools in the marketplace.

Install the Claude Code plugin from the JetBrains Marketplace. Provides interactive diff viewing and selection context sharing. Restart your IDE after installing.

A standalone app for running Claude Code outside your IDE. Review diffs visually, run multiple sessions side by side, schedule recurring tasks, and kick off cloud sessions. Available for macOS (Intel + Apple Silicon) and Windows.

Download from claude.ai, sign in, and click the Code tab to start.

Run Claude Code entirely in your browser at claude.ai/code. Great for kicking off long-running tasks and checking back when they're done, working on repos you don't have locally, or running multiple tasks in parallel. Also available on the Claude iOS app.

Got a spreadsheet you share around the team for tracking something? Give it to Claude Code and it'll turn it into a proper web app with search, sorting, and filtering — something that's easier to use and share than passing around an Excel file. Great for operations, finance, or anyone managing data in spreadsheets.

Need a polished landing page quickly? Describe the sections, tone, and content — Claude Code produces a complete, responsive page ready to deploy. Marketing and product teams can prototype campaign pages, feature announcements, or partner-facing content without touching a design tool or waiting for a developer.

Claude Code reads the module, understands the business logic, generates comprehensive unit and integration tests, runs them, fixes any failures, and commits the result. It'll cover edge cases like idempotency, retry logic, and error handling — things that are tedious to write manually.

Pipe your logs directly into Claude Code. It'll trace through the codebase, identify the root cause (e.g., a missing await on a database call causing the webhook to timeout), explain why it's happening, implement the fix, add a test for the regression, and open a PR.

Describe what you need and Claude Code handles the full vertical: Prisma schema update, migration, service layer, controller, route registration, validation middleware, tests, and API documentation. It follows the patterns already established in your codebase because it reads your CLAUDE.md and existing code.

Point Claude Code at a PR and it'll do a thorough security review — checking for injection vulnerabilities, authentication bypasses, PCI compliance issues, improper error handling that could leak sensitive data, and more.

One of those tasks everyone puts off. Claude Code updates your package.json, resolves conflicts, reads changelogs for breaking changes, updates your code to match the new APIs, runs your full test suite, and opens a PR with a detailed summary of what changed.

Ask Cowork to research what a competitor payment platform has launched recently. It browses their website, reads press releases, checks their changelog, and delivers a structured summary — what they launched, how it compares to your offering, and where the gaps are. Great for pre-meeting prep or quarterly competitive reviews.

Give Cowork the customer details and your onboarding template. It fills in the customer-specific information, assembles the welcome documentation, generates the setup checklist, and packages everything into a folder ready to send. What used to take 45 minutes takes 5.

Point Cowork at a settlement CSV and your transaction records. It reads both files, matches entries, flags discrepancies, calculates variance, and produces a reconciliation summary with the exceptions highlighted. The Finance team's new favourite trick.

After a strategy meeting, dump your raw notes into Cowork. It organises the content into a logical flow, writes clear slide titles and bullet points, adds structure (agenda, key decisions, next steps, owners), and delivers a presentation-ready document. You decide the format — Google Slides, PowerPoint, or a formatted PDF.

What it does: Claude can search Slack channels, read message threads, and post messages. Great for pulling context from team discussions into your work.

Example use cases:

• "What did the team discuss about the settlement feature in #engineering last week?"
• "Post a summary of today's deployment to #releases"
• In Claude Code: mention @Claude in Slack with a bug report and get a pull request back

What it does: Search and read files from your connected Google Drive. Claude can reference specs, reports, spreadsheets, and documents without you uploading them manually.

Example use cases:

• "Find the Q1 customer report in Drive and summarise the key metrics"
• "Pull the API spec from our shared drive and check if the implementation matches"

What it does: Read issues, PRs, code, and repository metadata. Create branches, commit code, and open pull requests. Automate code review.

Example use cases:

• "Review PR #247 and leave comments on any security concerns"
• "Read issue #89, implement the feature, and open a PR"
• Automated PR review on every push via GitHub Actions

What it does: Connect Claude to your database for read access. Claude can explore schemas, run queries, and analyse data directly.

Example use cases:

• "How many new customers onboarded last month? Break down by category."
• "Show me the schema for the transactions table and suggest indexes for our most common queries"

Security note: Configure with read-only credentials. Never give write access to production databases through MCP.

What it does: Query product analytics, view dashboards, check experiments, review session replays, and analyse feature flag usage.

Example use cases:

• "What's the checkout conversion rate this week vs last week?"
• "Show me session replays where users dropped off at the payment step"
• "How is the new settlement UI experiment performing?"

What it does: Read design files, inspect components, extract styles, and generate code from designs. Great for bridging the design-to-code gap.

Example use cases:

• "Look at the new checkout flow in Figma and build the React components"
• "Extract the colour palette and spacing from our design system"

Technology:

• /security-review — PCI-DSS focused code review checklist
• /incident-report — Post-incident report in your standard template
• /api-doc — Generate API docs from code in pay.com.au's format
• /migration-plan — Database migration planning with rollback steps

Product:

• /release-notes — Customer-facing release notes from git log
• /customer-feedback — Synthesise customer feedback into actionable insights
• /pricing-analysis — Analyse pricing changes against competitor data

Finance:

• /settlement-reconciliation — Check settlement files against transaction records
• /compliance-check — Review against APRA/ASIC requirements
• /board-report — Monthly board report in your standard format

Marketing:

• /customer-case-study — Case study template with pay.com.au brand voice
• /competitor-update — Weekly competitor monitoring report

Support:

• /ticket-response — Draft responses using your support playbook
• /escalation-summary — Summarise a ticket thread for escalation to engineering

To suggest a new skill for the library, provide the following:

• Skill name — what would the /slash-command be? Keep it short and descriptive.
• What it does — describe the task in 1-2 sentences.
• Who uses it — which team(s) or roles would benefit?
• Inputs — what does Claude need to do this? (uploaded file? project context? specific data?)
• Expected output — what should the result look like? Include a sample if possible.
• Quality criteria — how do you know if the output is good? What are the must-haves?
• Frequency — how often does your team do this task?

Submit requests to your team's Claude administrator. The best requests include an example of a good output — this makes it dramatically easier to build an effective skill.

Once a skill request is approved:

• The skill is created as a SKILL.md file with YAML frontmatter (name, description) and markdown instructions
• It's tested with real examples to ensure output quality
• It's committed to .claude/skills/ in the relevant repository
• Every team member using Claude Code automatically gets it on next pull
• For non-Claude Code users, skills are available through the Cowork desktop interface
• Skills activate immediately — no restart needed (as of January 2026)

Regular conversations are private to you. Projects can be shared — you choose whether a project is private, view-only, or editable by the team. Admins can view usage statistics but not conversation contents.

Use your professional judgement, but as a general rule: avoid sharing raw customer PII (full card numbers, passwords, authentication tokens) unless you have explicit approval and data handling controls in place. Claude is a tool — the same data handling policies that apply to any third-party service apply here.

Usage limits reset weekly. If you hit your limit, you'll be notified and can resume when it resets. You can switch to a lighter model (e.g., Haiku) which has separate, more generous limits. Premium seat holders get 5× more usage than Standard seats. If you consistently hit limits, talk to your admin about a seat upgrade.

Yes. Claude has web search capability on the Teams plan. It can search for current information, recent news, documentation, and more. It will cite its sources so you can verify. For questions about internal company data, use integrations (Google Drive, Slack, etc.) or upload the files directly.

Sonnet 4.6 — Best for everyday use. Fast, smart, and great at most tasks. This should be your default.

Opus 4.6 — The most capable model. Use for complex reasoning, nuanced analysis, difficult coding problems, and when quality matters most. Slower and uses more of your quota.

Haiku 4.5 — Fastest and lightest. Great for simple tasks: quick summaries, reformatting, brainstorming, and translation. Uses the least quota.

Rule of thumb: Start with Sonnet. Upgrade to Opus for high-stakes or complex work. Drop to Haiku for simple, quick tasks.

The day-to-day experience is very similar — you chat, upload files, and get responses. The main adjustments:

• Your custom GPTs → recreate as Claude Projects
• ChatGPT "memory" → Claude has memory too, plus Projects for structured team context
• Canvas → Artifacts (same concept, different name)
• Image generation → not built in; use external tools
• You'll likely notice Claude follows instructions more precisely and writes more naturally

For internal questions about our Claude setup, policies, or access, reach out to your team administrator.

For product issues or questions about Claude itself, visit support.claude.com.

You can also use the thumbs up/down buttons on any Claude response to provide direct feedback to Anthropic.