Claude 101

01What is Claude?

Claude is an AI assistant by Anthropic, designed as a thinking partner for complex work — not just a question-answering chatbot. Claude is steerable, meaning you can direct its personality, tone, and behavior to match your needs.

• Writing & content creation — emails, reports, social posts, collaborative drafts
• Research & analysis — synthesizing documents, surfacing insights, processing large amounts of text
• Coding — writing, debugging, and explaining code across many programming languages
• Problem-solving & reasoning — complex cognitive tasks, math, strategic analysis
• Learning — adapts to your learning style and pace

Learning mode is a Claude experience that guides your reasoning process rather than providing direct answers — designed to build critical thinking skills, not replace them.

02Your First Conversation with Claude

All interactions with Claude begin with a prompt. Treat Claude like a knowledgeable coworker: speak naturally, concisely, and conversationally. The more relevant context you provide, the better the response.

The Prompt Framework

Every strong prompt has three elements. Think of them as stage-setting before asking Claude to act.

1. Set the stage — your role, objectives, and relevant context Claude should know
2. Define the task — the specific action you want (write, analyze, build, research, etc.)
3. Specify rules — style, tone, format, length, and any examples to guide output

Adding Context

Uploading files gives Claude immediate context without lengthy explanations. Claude analyzes both text and visual elements (charts, graphics) in uploaded documents.

• Supported file types: PDF, DOCX, CSV, TXT, HTML, PNG, JPEG
• Ask Claude to summarize a document's key points
• Share an image and ask Claude to describe or analyze it
• Attach a spreadsheet and ask Claude to identify trends
• Upload code and ask Claude to explain or debug it

Personalizing Claude

Two built-in features help Claude learn and adapt to you over time, reducing repeated explanations across conversations.

03Getting Better Results

The 4D Framework for AI Fluency

The 4D Framework, developed by Professors Rick Dakan (Ringling College of Art and Design) and Joseph Feller (University College Cork), identifies four competencies for effective AI collaboration.

• Delegation — deciding what work belongs to humans vs. AI, and how to distribute tasks strategically
• Description — communicating clearly with AI: defining outputs, guiding process, specifying behavior
• Discernment — critically evaluating AI outputs for quality, accuracy, and appropriateness
• Diligence — using AI responsibly and ethically; maintaining transparency and taking accountability

The prompt framework (stage, task, rules) maps to Description. The troubleshooting techniques above map to Discernment and Diligence.

Evaluating Claude with Evals

Your work is unique. Claude might excel at drafting marketing copy but need more guidance for technical documentation in your specific domain. Evals help you find out before committing.

1. Gather 5–10 examples of a recurring task you've already completed
2. Write prompts that would reproduce similar outputs (with the context you'd naturally have)
3. Compare Claude's responses to your known-good examples — check key info, tone, and gaps
4. Refine prompts, add examples, identify where human review is essential
5. If Claude matches your results, delegate confidently; if not, keep that task human-owned

04Claude Desktop App: Chat, Cowork, Code

The Claude desktop app offers three distinct modes — Chat, Cowork, and Code — each optimized for a different type of work. Cowork and Code both run on the Claude Code engine: local to your machine, capable of independent multi-step work, and able to spawn sub-agents.

Chat

Chat works identically to Claude.ai, with additional capabilities unlocked by running natively on your computer.

• Quick entry — double-tap Option (Mac) to open Claude as a compact overlay over any app
• Screenshots & window sharing — capture your screen so Claude sees exactly what you see (Mac)
• Dictation — talk through problems instead of typing (Mac)
• Desktop connectors — connect local tools and services through the desktop app

Cowork

• Folder access — point Claude to a local folder; it reads, processes, and saves work back there
• Scheduled tasks — recurring automated work (daily briefings, weekly roundups, inbox triage)
• Subagents — parallel workers for complex multi-source tasks
• Dispatch — hand tasks to your desktop Claude from your phone via the mobile app
• Browser use — Claude navigates websites in Chrome to gather data
• Computer use — Claude interacts with your desktop apps directly (research preview, Pro/Max, macOS only)
• Plugins — extend Claude with live data feeds, compliance tools, internal knowledge bases

Code

Code runs directly in your project with full access to your file system, terminal, and development tools. Choose Local (a folder on your machine) or Remote (a GitHub repo in a cloud environment where sessions persist even if you close the app).

• Ask mode — Claude proposes every change and waits for approval; you review a visual diff before anything is modified
• Code mode — Claude applies file changes automatically but checks before running terminal commands
• Plan mode — Claude outlines its full approach first; a plan viewer lets you review before work begins

05Projects

Use projects when you have reference materials you'll use repeatedly, consistent style or format requirements for Claude's responses, or team collaboration needs where multiple people should work from the same foundation.

Creating a Project

Setup takes a few minutes. Project instructions tell Claude how to behave across all conversations — specify tone, expertise level, response style, and context.

1. Go to claude.ai/projects and click + New Project; give it a descriptive name
2. Click Instructions and specify tone, expertise level, response style, and context; click Save instructions
3. Click + in the files menu to upload PDFs, DOCX, CSV, TXT, HTML, or link Google Drive

Collaboration (Team & Enterprise)

On Team and Enterprise (Claude for Work) plans, projects enable shared workspaces. Share with specific teammates or make a project visible to your entire organization.

06Artifacts

Claude automatically creates an artifact when content is substantial (typically 15+ lines), self-contained, likely to be reused, or complex enough to stand on its own. If Claude doesn't create one automatically, say: "Create this as an artifact."

Sharing Artifacts

Once created, you have three options for sharing an artifact depending on your audience and plan.

• Copy or download — available to all users; use buttons in the artifact window
• Share within your org (Team/Enterprise) — artifact stays internal and requires team authentication to access
• Publish publicly (Free, Pro, Max) — anyone with the link can view and interact; others can "remix" it into their own Claude conversation

Tips for Better Artifacts

• Be specific: describe the end user, required features, and constraints upfront
• Describe the end user — "This flowchart is for new employees" leads to different results than "for the engineering team"
• Iterate incrementally — request one change at a time to catch issues early

07Skills

Enabling Skills

Skills require Claude's sandboxed computing environment, so Code execution and file creation must be enabled before Skills can function.

1. Go to Settings > Capabilities
2. Toggle on Code execution and file creation
3. Scroll to the Skills section and toggle individual skills on or off

Creating Custom Skills

The easiest way to create a custom Skill is through conversation — Claude handles the technical file structure for you.

1. Start a new chat and describe the skill (e.g., "I want a skill for writing quarterly business reviews")
2. Answer Claude's questions about your workflow, desired outputs, and examples
3. Upload reference materials if available (templates, style guides, past examples)
4. Claude generates a structured skill file — save it and it becomes available immediately

Skills vs. Projects

Both features give Claude more context, but they serve opposite roles: Projects store *what* Claude knows; Skills define *how* Claude does a task. They complement each other — a skill can pull information from a project's knowledge base.

08Connectors

Connectors transform Claude from an assistant into an informed collaborator by giving it access to the same tools, data, and context you use daily. Depending on the connector, Claude can read, search, create, update records, and execute tasks across connected apps.

Setting Up a Web Connector

The connector directory at claude.ai/directory lists all available connectors organized by Web and Desktop tabs. Desktop extensions require the Claude Desktop app and are installed at Settings > Extensions.

1. Go to claude.ai/directory or click + > Connectors in any chat
2. Select a connector and click Connect
3. Sign in with your existing credentials for that service
4. Review and authorize the specific permissions requested
5. Test with: "Can you access my [tool name]?"

Permissions and Security

Permissions are scoped to what the connector specifically needs. Claude only sees data you already have access to — connecting your work email gives access to your inbox only, not colleagues'.

• Toggle individual permissions on/off within each application's menu
• Revoke at any time via Claude's settings or the third-party app's security settings

09Enterprise Search

Enterprise Search searches across all connected sources — SharePoint, Slack, Gmail, Google Drive, Microsoft Teams, and more — and synthesizes them into a single cited response. Think of it as a pre-built Project for your entire organization, with the knowledge base already loaded.

What You Can Ask

Enterprise Search is most valuable for questions that span multiple sources or require synthesizing information from across your organization.

• Get up to speed — "What happened yesterday while I was out?" / "Summarize key updates from last week"
• Policy & process — "What is our remote work policy?" / "How do I submit an expense report?"
• Research & analysis — "Why do customers choose competitors?" / "Summarize Q4 roadmap discussions"
• Onboarding — "How does our authentication system work?" / "Who should I talk to about the billing system?"
• Project tracking — "What were the key decisions from last week's leadership meetings?"

Setup Process

10Research Mode

How Research Works

Enabling Research automatically activates Extended Thinking — Claude plans its investigation before searching, then autonomously decides what to search next based on what it finds.

1. Plan — Extended thinking activates; Claude breaks down the request and identifies investigation angles
2. Search — Claude runs multiple searches that build on each other, pursuing leads without you directing each step
3. Synthesize — Claude combines findings from the web and connected integrations (Gmail, Drive, etc.)
4. Cite — every claim in the report links back to its source for easy verification

Writing Effective Research Prompts

Since Research takes 5–45 minutes, a well-crafted prompt pays off. Steer Claude's use of connected integrations explicitly when needed.

• Be specific about goals: instead of "Tell me about EVs," use "Analyze the EV battery market — key players, technology trends, and supply chain risks for investors"
• Specify the structure you want: "Compare venue options across: location, meeting space, catering, and pricing"
• Include constraints: budget, geography, timeline, or audience
• Steer integration use: say "Pull relevant context from my Google Drive" or turn off web search for internal-only research

11Claude Across Roles and Tools

Claude applies across professional roles. The Use Case Gallery at claude.com provides step-by-step guides for each function — browsing it is the fastest way to find prompts tailored to your specific work.

Additional Claude Products

Beyond Claude.ai, Claude's intelligence is embedded in specialized tools designed to meet you where you already work — each tailored to a specific workflow.