Claude 101

0101. What Claude Is

Claude is an AI assistant built to function as a thinking partner. Rather than just answering simple questions, it handles summarization, search, creative writing, Q&A, coding, and more—while remaining reliable and predictable across all of those tasks.

• Helpful: assists with a broad range of tasks and provides genuinely useful responses.
• Harmless: avoids toxic, discriminatory, or illegal outputs and declines requests that could cause harm.
• Honest: operates transparently and does not deceive users.

Claude is steerable and collaborative. You can direct it on personality, tone, and behavior. Users consistently report that Claude is easier to guide toward a desired output and produces fewer harmful responses than comparable tools—meaning less effort to get the result you need.

Plans and Device Access

Claude apps are available across Free, Pro, Max, Team, and Enterprise plans. Every plan type includes access to the web, desktop, and mobile apps.

0202. Core Capabilities

Claude's capabilities span a wide range of tasks, going well beyond simple question-and-answer. It can both automate and augment your work—acting as a collaborator on complex, open-ended problems.

Writing, Research, and Analysis

For writing, Claude collaborates on social media posts, professional emails, and detailed reports. Because it accepts direction on tone and style, you can iterate with it until your own voice comes through clearly.

For research and analysis, you can upload documents and Claude will surface meaningful insights from complex material. Its large context window allows it to process extensive materials—potentially hundreds of pages—within a single conversation without losing track of earlier content.

Coding and Problem-Solving

Coding is one of Claude's strongest areas. It can write, debug, and explain code across multiple programming languages, with strong performance on real-world coding tasks.

For complex cognitive tasks—mathematical problems, strategic analysis, research—Claude handles everything from quick lookups to deep multi-step reasoning. Extended thinking makes it well-suited for problems that require careful, structured analysis.

Learning Support

Claude adapts to your learning style and pace. Whether you are picking up a new skill, exploring an unfamiliar domain, or working through a challenging concept, it adjusts its explanations to meet you where you are.

0303. Ways to Access Claude

The same Claude intelligence is available through multiple interfaces, each suited to a different type of work. Most users interact with Claude through Claude.ai—the web, desktop, and mobile experience—which is the focus of this course.

Claude Code is an agentic coding tool designed for developers but usable for general file manipulation on the desktop. It goes beyond conversation—it can directly edit files, run terminal commands, and create git commits.

Claude + Slack lets you chat with Claude from any channel's AI assistant header or by @mentioning it in threads. When connected, Claude searches your workspace's channels, direct messages, and shared files to give more contextually relevant responses.

Claude for Excel adds a sidebar inside Microsoft Excel where Claude can read, analyze, modify, and create workbooks. It is best suited for model analysis, assumption updates, error debugging, template population, formula explanations, and multi-tab navigation.

0401 The Claude.ai Interface

Claude is an AI collaborator that amplifies your capabilities across your work. You bring the context and expertise that makes the work meaningful; Claude brings AI intelligence to help you act on it.

When you open Claude.ai, the sidebar on the left gives you quick access to the main navigation. Your text input area sits at the bottom of the screen, ready for your first prompt.

• New chat icon — starts a fresh conversation
• Chat history — lets you return to previous conversations
• Projects — organize conversations with persistent context and custom instructions
• Artifacts — turn ideas into shareable apps, tools, or content

Your prompts can range from simple questions — like brainstorming names for a new feature — to complex requests to co-create full documents or code. No matter the goal, Claude can support you.

0502 Writing Effective Prompts

The best approach when speaking to Claude is the same as speaking to a coworker — naturally, concisely, and conversationally. Prompts do not need to be formal commands; plain language works well.

The Three-Part Framework

A strong prompt covers three elements before you hit send.

Putting the Framework Together

Combining all three elements in a single prompt gives Claude everything it needs to produce a well-targeted response. The investor pitch deck example from the video demonstrates this: role and context come first, then the specific research task, then format requirements and citation rules.

0603 Adding Context and Tools

Uploading files and connecting external tools gives Claude access to information beyond what you type. Claude can analyze both text and visual elements — such as charts and graphics — in the files you attach.

• PDF — documents with text and embedded images
• DOCX — Word documents
• CSV — tabular data
• TXT — plain text files
• PNG and JPEG — images, screenshots, and charts

Once a file is uploaded, it appears as an attachment in the chat and Claude automatically parses its content. You can then prompt Claude directly about anything in the file.

1. Upload a document and ask Claude to summarize the key points
2. Share an image and ask Claude to describe or analyze what it sees
3. Attach a spreadsheet and ask Claude to identify trends in the data
4. Upload code and ask Claude to explain it or find bugs

The Search and Tools menu extends Claude's access to real-time information. You can toggle on web search to let Claude pull current data, or connect Google Drive to access your own files. Claude selects which tools to use automatically while solving your request — you do not need to specify when.

0704 Choosing the Right Model

Claude offers multiple models, each optimized for different types of work. The model selector appears under your text input when you start or open a chat. Changing the model starts a new chat.

For straightforward tasks, start with Sonnet and leave thinking mode off to get the fastest output. Switch to Opus or enable thinking mode when a task genuinely requires more analytical depth.

Research mode is well-suited for competitive analysis, market research, or literature reviews. You submit the question and Claude handles the investigation while you focus on other work.

0805 Iterating and Personalizing

Claude conversations are meant to be iterative. Chaining short prompts together — rather than writing one long request — creates a natural dialogue where you guide the output based on what Claude returns.

Refining Claude's Responses

• Ask follow-up questions: request more detail, a different angle, or clarification on a specific point — e.g., "Can you expand on the second point?"
• Provide feedback: tell Claude what worked and what didn't — e.g., "The tone is too formal, make it more conversational"
• Redirect: steer Claude back if it went the wrong direction — "I was asking about X, not Y"
• Restart: open a new chat to fully refresh context when a conversation has drifted too far off track

The more frequently you interact with Claude, the better it becomes at understanding your communication style and preferences. Continued conversation — not just one-off prompts — is where Claude's full value surfaces.

Memory and Styles

For example, if you tell Claude you work in marketing at a B2B company, Memory stores that so Claude applies it in future conversations without being told again.

Together, Memory and Styles reduce the setup work at the start of each conversation. Claude arrives with context about who you are and already knows how you prefer to receive information.

0901. Common Challenges When Starting Out

When Claude's response isn't what you expected, that's a signal to refine your approach—not a reason to stop. Most issues trace back to a few predictable causes, and each has a clear fix.

The Iteration Mindset

Your first prompt rarely produces a perfect result—and that's expected. Think of your initial prompt as the start of a conversation, not a one-shot request.

• Treat first drafts as starting points: review what Claude produces, identify what's working and what isn't, then refine.
• Give specific feedback: "Cut the first two paragraphs and make the conclusion more action-oriented" is more useful than "Make it shorter."
• Know when to start fresh: if a conversation has gone off track, opening a new chat with a clearer prompt is often faster than trying to redirect it.

1002. AI Fluency and the 4D Framework

The 4D Framework was developed through research collaboration between Professor Rick Dakan (Ringling College of Art and Design) and Professor Joseph Feller (University College Cork). It identifies four core competencies that, when combined, help you get the most from your AI interactions.

How the 4Ds Connect to This Course

The prompt framework from earlier lessons—setting the stage, defining the task, specifying rules—is rooted in Description. The troubleshooting techniques above draw on Discernment and Diligence. You've already been practicing these competencies throughout this course.

• Delegation: choosing which tasks to hand off to Claude and which to keep
• Description: writing clear prompts with context, constraints, and format expectations
• Discernment: evaluating whether a response is accurate, complete, and appropriate
• Diligence: verifying facts, disclosing AI's role, and staying accountable for the final output

1103. Evaluating Claude for Your Workflows

Your work is unique. Claude might excel at drafting marketing copy but need more guidance for technical documentation in your specific domain. Simple evals reveal where Claude adds the most value—and where you need to stay more involved.

• Understand where Claude adds the most value in your specific workflow
• Identify tasks where you'll need to provide more context or examples
• Build confidence in Claude's outputs for recurring tasks before depending on them

A Simple Eval Approach

You don't need complex infrastructure to run evaluations. A practical approach uses examples you already have.

1. Gather 5–10 examples of a task you do regularly—emails you've written, reports you've created, analyses you've completed.
2. Write prompts that would generate similar outputs, including the context you'd naturally have when doing this work.
3. Run your prompts and compare Claude's responses to your examples: Does Claude capture the key information? Is the tone and style appropriate? What's missing or could be improved?
4. Adjust your prompts, add examples to show Claude what good looks like, or identify where human review is essential.

1204. The Delegation Diligence Loop

The loop gives you a structured way to determine whether AI is appropriate for a specific analytical task—and exactly what context it needs to perform that task well.

1. Identify a specific analytical task you do regularly that you want to delegate to AI.
2. Find past data where you already completed the analysis—you need the correct answers in order to evaluate AI's results.
3. Work with AI to reproduce your past analysis using clear description.
4. Evaluate the results against what you know to be true. Note what AI got right and what it missed.
5. Refine your approach based on the gaps and test again.
6. If AI can match your known results, you have a validated approach for similar future tasks. If not after several refinements, you've learned this task shouldn't be delegated.

Rio, a program director at Valley Veterans Services, applied this loop to his quarterly analysis of program attendance and employment outcomes—work that regularly took him hours. In his first attempt, AI correctly identified the correlation between attendance and job placement but missed a critical insight around a combined housing and job placement program. Rio refined his description to request that AI consider program type, and AI caught the gap. He then tested cohort analysis by enrollment date, discovering that AI would infer enrollment dates when they weren't in the data—something he didn't want. Each gap became a documented note about what context to include in future prompts, leaving him with a validated approach and clear rules for when to apply it.

What If You're Not Comfortable with the Data?

If you're not familiar enough with a dataset to spot AI's process gaps yourself, AI can still help—this time as a thinking partner rather than a validator.

• Bring your question or idea to AI and ask what a solution could look like, the same way you'd work with a data analyst on your team
• Request clarifications and explanations as you go so you can follow the process and understand the final output
• Use AI to handle tasks like writing Excel formulas or reformatting messy data
• Keep asking questions until you understand the output well enough to take accountability for it

1301 The Claude Desktop App

The Claude desktop app gives you three ways to work with Claude: Chat for quick back-and-forth exchanges, Cowork for complex multi-source work, and Code for building software. Each mode surfaces controls and information tailored to the work it serves.

Cowork and Code share the same underlying engine. Both run Claude Code locally on your machine, both can handle larger tasks independently, and both can spin up sub-agents for sustained workflows. The difference is what they are pointed at: documents and research for Cowork, codebases for Code.

• Connectors, Skills, and Claude in Chrome are available across all three modes
• Plugins and Hooks are available in Cowork and Code, but not in Chat
• Computer Use is exclusive to Cowork
• Quick entry, dictation, and screenshots are exclusive to Chat (Mac)

1402 Chat

Chat works the same as claude.ai but gains additional capabilities from running natively on your computer. It excels at questions, brainstorming, drafting, and iterative problem-solving — any work where you want fast back-and-forth rather than a long autonomous task.

• Quick entry: Double-tap Option (Mac) to open Claude over whatever you are working on; it responds in an overlay that stays on top
• Screenshots and window sharing: Capture a screenshot or share a window so Claude sees exactly what you are looking at — faster and more precise than describing your screen (Mac)
• Dictation: Talk through a problem instead of typing, useful when thinking out loud or away from your keyboard (Mac)
• Desktop connectors: Connect local tools and services so Claude can work with apps and data sitting directly on your machine

Quick entry keeps the app behind it active. Screenshot an unfamiliar dashboard, ask what the metrics mean, and read the answer without switching windows. Or open quick entry, switch to voice, talk through a presentation structure — Claude drafts the outline from what you said.

Desktop connectors let Claude reach local tools and services — not just cloud APIs. This makes Chat useful for tasks that involve files, apps, or data on your computer rather than only in the cloud.

1503 Cowork

Cowork is designed for work that requires pulling from many sources, reasoning across all of them, and producing something finished — not a chat reply, but a deliverable. Before starting, Claude often asks a short set of questions to pin down scope, format, and constraints, then builds a reviewable plan in the sidebar.

• Folder access: Point Claude to a folder; it reads what is relevant, works with it, and saves finished output back to the same place
• Scheduled tasks: Define a recurring task (daily briefing, weekly roundup, morning inbox triage) and Claude runs it automatically each time the app is open, catching up on missed runs when you return
• Subagents: Background workers that handle subtasks in parallel, coordinated by Claude into one deliverable
• Dispatch: A persistent thread that lets you assign Cowork tasks from your phone via the Claude mobile app, using your desktop files and connectors while your computer is awake
• Projects: Dedicated workspaces grouping related Cowork tasks with their own files, context, instructions, and memory — stored locally on your desktop

Browser use and computer use extend Cowork's reach beyond what you upload or share. With the Chrome connector, Claude can navigate websites, interact with pages, and pull data directly into the task — useful for competitor research or any site without an API. When no connector or Chrome integration exists, computer use lets Claude navigate your computer directly, clicking, typing, and opening apps.

Plugins add capabilities Cowork does not have natively: live financial data, internal knowledge base search, or compliance-specific workflows. A protected environment limits Claude to the folders you explicitly share — it cannot read or modify anything outside them.

• Query decisions across meeting notes, Slack, email, and slide decks in a single request
• Research new markets without opening a browser tab — Cowork visits sites, reads reports, and delivers a structured brief with sources
• Work through 50+ documents as if reviewing five — Claude reads every page and surfaces patterns that only emerge from reading the full set
• Automate repetitive morning admin (messages, status updates, meeting prep) with a scheduled task set up once

1604 Code

The Code tab gives you Claude Code running directly inside the desktop app — a full development environment where Claude reads your codebase, writes and modifies files, runs terminal commands, and tracks every version with git. A built-in terminal shows commands as they run, and visual diffs show exactly what changed.

Development Environments

Code lets you choose where work happens — on your local machine or in a cloud environment linked to a GitHub repository. The choice affects what Claude can access and whether sessions persist when the app closes.

Interaction Modes

Three interaction modes let you control how much Claude does before checking with you. You can switch between them as the work evolves — tighter control when reviewing unfamiliar code, more autonomy for routine tasks.

• Ask: Claude proposes every change and waits for your approval — you review a visual diff and accept or reject before any file is modified
• Code: Claude applies file changes automatically but checks with you before running any terminal commands
• Plan: Claude outlines its full approach before touching anything; a dedicated plan viewer lets you review and revisit the strategy as work progresses

1701 What Are Projects?

Projects solve a common problem: having to re-upload files and re-explain context every time you start a new conversation. Instead of treating each chat as independent, a project gives Claude persistent memory of your documents, goals, and preferences across all conversations within it.

• Store reference materials Claude should always have access to
• Organize related chats around a specific topic or work area
• Apply consistent behavior rules across every conversation
• Collaborate with teammates who need access to the same shared context

When to Use a Project

Projects are most valuable for ongoing work—not one-off questions. A project makes sense when your workflow involves materials or rules you would otherwise repeat in every conversation.

• Reference materials you use repeatedly (meeting notes, reports, historical data)
• Consistent response requirements (always cite sources, always follow our template)
• Team collaboration where multiple people work from the same foundation

1802 Setting Up Your First Project

Creating a project takes only a few minutes and follows three sequential steps: basic setup, writing instructions, and building the knowledge base.

Step 1 — Basic Setup

1. Hover over the left sidebar and click "Projects," or go to claude.ai/projects
2. Click "+ New Project" in the upper right corner
3. Give the project a descriptive name (e.g., "Q4 Marketing Campaign")
4. Add a brief description of the project's purpose
5. Choose visibility: private for personal use, or shared with your organization (Claude for Work users only)

Step 2 — Write Project Instructions

Project instructions tell Claude how to behave in every conversation within the project. Click "Instructions" to open the panel, write your instructions, then click "Save instructions."

Step 3 — Build the Knowledge Base

The knowledge base holds documents that Claude references across all chats in the project. Find the files menu on the right side of the project's main page and click "+" to add content.

• Supported file types: PDF, DOCX, CSV, TXT, HTML, and more
• Google Drive connection to link documents directly
• Reference documents such as brand guidelines, style guides, and templates
• Background materials such as research reports and requirements docs
• Examples of work you want Claude to emulate

1903 How Projects Scale: RAG Mode

When the combined size of your project's knowledge base approaches the context window limit, Claude automatically switches to RAG mode. This happens without any action on your part—you will see a visual indicator that the project is RAG-enabled, but the experience remains the same.

• Expands project capacity by up to 10x
• Claude intelligently retrieves only the most relevant sections for each query
• You can still upload documents, chat, and receive context-aware responses
• No manual configuration required

If you upload content that you need for a single conversation without adding it to the knowledge base permanently, simply upload it during the conversation. It stays separate from the project knowledge and does not affect the knowledge base size.

2004 Collaboration and Permissions

For users on Claude for Work (Team and Enterprise) plans, projects can be shared with other members of the organization. This turns a personal workspace into a team resource where everyone benefits from the same context, instructions, and knowledge base.

Permission Levels

Sharing a Project

1. Open the project you want to share
2. Click "Share project" to the right of the project name
3. Add members by name or email, or paste a list of email addresses for bulk sharing
4. Alternatively, share with "Everyone at [your organization]" to make it discoverable in the Team tab

Multiple members can contribute documents, create chats, and work together within the same project environment. Owners can modify permissions or remove access at any time.

2105 Example Projects and Best Practices

Projects work across a wide range of professional and personal contexts. The following examples show the kind of materials and use patterns that benefit most from a project structure.

Best Practices

• Start focused — one specific use case per project, then expand as needed
• Keep the knowledge base current — outdated documents lead to outdated responses
• Write specific instructions — vague instructions produce inconsistent results
• Name documents descriptively so Claude can find and relate the right files
• Reference documents by name in your messages to help Claude focus its search

2201 What Artifacts Are

When Claude generates a long document or complex code, that content is easy to lose in the scroll of a conversation. Artifacts fix this by placing the output in a separate pane where you can preview it, interact with it, and act on it directly.

Claude applies four criteria to decide whether content becomes an artifact:

• Significant and self-contained — typically over 15 lines
• Likely to be edited, iterated on, or reused
• Complex enough to stand on its own without the surrounding conversation
• Something you will want to reference or use later

When an artifact appears, it opens in a window to the right of your chat. From there you can toggle between a rendered preview and the underlying source, copy the content to your clipboard, or download the file to your computer.

2302 Artifact Types

Claude can create six categories of artifacts. Each suits a different kind of deliverable — the right choice depends on what you need to do with the output.

SVG images render directly in the artifact window, so you see the finished graphic immediately rather than raw markup. HTML artifacts combine HTML, CSS, and JavaScript into a single file — useful for self-contained prototypes you can open in any browser without a server.

Document artifacts support export formats including Word, PDF, PowerPoint, and Excel — making them suitable for content that needs to move outside Claude into other tools or workflows.

2403 Creating and Refining Artifacts

Creating an artifact starts with describing what you want. Claude determines from your request whether the output qualifies as an artifact. The more precise your description, the more useful the result.

• "Create a flowchart showing our customer onboarding process"
• "Build an interactive dashboard that lets me input monthly expenses and see a breakdown"
• "Design a landing page for a productivity app with a hero section and feature list"
• "Write a project brief template I can reuse for new initiatives"

Describing the end user changes the output meaningfully. "This flowchart is for new employees" leads to different design choices than "This flowchart is for the engineering team." Give Claude that context before it generates.

Iterating on an Artifact

Once an artifact exists, you refine it through the conversation. Each follow-up request can target one specific change — add a feature, adjust a layout, fix a calculation — without rebuilding from scratch.

1. Start with a clear, specific initial request
2. Review the artifact in the preview pane
3. Request one targeted change per follow-up message
4. Toggle to the code view to understand what was generated
5. Download or copy the artifact when it meets your needs

2504 Sharing and Publishing Artifacts

Three sharing paths are available after creating something useful. Which one to use depends on your Claude plan and your intended audience.

Copy and Download

Copy and download buttons sit in the lower right corner of the artifact window. Use them to grab the content for use outside Claude — pasting into a document, committing to a repository, attaching to an email, or opening in a browser.

• Copy: puts the raw content into your clipboard
• Download: saves the artifact as a file to your computer
• View code: shows exactly what Claude generated under the hood

Share Within Your Organization

Team and Enterprise users on Claude for Work can share artifacts internally with colleagues. The shared artifact stays within the organization — recipients must belong to the same Claude for Work account to access it.

Publish Publicly

Free, Pro, and Max users can publish an artifact to make it accessible to anyone with the link. Click "Share" or "Publish" in the upper right corner of the artifact. To remove access, return to that artifact and remove public access — unpublishing is available at any time.

• Only the selected version becomes public — your chat remains private
• Viewers do not need a Claude account to see or interact with the artifact
• Others can "remix" the artifact — opening it in their own Claude session to modify and build on it
• Published artifacts are not indexed by search engines and will not appear in Google results

2601 What Skills Are

Skills run behind the scenes when you ask Claude to create Excel spreadsheets, PowerPoint presentations, Word documents, or PDFs. Beyond document creation, Skills can encode entire repeatable workflows—a quarterly variance analysis, a brand voice review process, or a compliance checklist—so Claude follows the same rigorous steps every time.

The core value of Skills is consistency. When a workflow is captured in a Skill, Claude executes the same methodology each time it is invoked—reducing variation and eliminating the need to re-explain your process in every conversation.

• Excel spreadsheets with formulas and structured data
• PowerPoint presentations built from source material
• Word documents with consistent formatting and structure
• PDF reports that summarize and organize data
• Custom repeatable workflows specific to your organization

When Claude uses a Skill, it is mentioned in Claude's chain of thought as it works through your request. The final output is a downloadable file you can save to your computer or directly to Google Drive.

2702 Types of Skills

There are two categories of Skills: those built and maintained by Anthropic, and those you or your organization create for your own workflows.

Anthropic Skills

Anthropic Skills are created and maintained by Anthropic. They provide enhanced document creation capabilities for Excel, Word, PowerPoint, and PDF files. These Skills are available to all paid users.

• Excel spreadsheets (.xlsx) with formulas and formatting
• PowerPoint presentations (.pptx)
• Word documents (.docx)
• PDF files (.pdf)

Custom Skills

Custom Skills let you teach Claude your specific ways of working. Once created, they are private to your individual account and automatically invoked whenever you work on relevant tasks.

• Applying company brand guidelines to presentations
• Structuring meeting notes in a specific internal format
• Executing your organization's data analysis workflows
• Running compliance checklists with consistent steps
• Drafting documents that follow internal style guides

2803 Enabling and Managing Skills

Enabling Skills takes just a few steps in your account settings. The process is the same for Pro and Max plans; Team and Enterprise plans have additional organization-level controls.

1. Navigate to Settings > Capabilities
2. Ensure Code execution and file creation is toggled on
3. Scroll to the Skills section
4. Toggle individual Skills on or off as needed

Once enabled, all available Skills—both Anthropic's built-in ones and any custom Skills you have uploaded—appear listed in your settings. You can toggle each one individually to control which Skills are active.

To work with existing files using Skills, you will also need to allow limited network access when prompted. This enables Claude to access your uploaded files within its contained environment and produce updated output versions.

2904 Using Skills in Practice

The primary design goal of Skills is transparency—you should not need to think about them. Claude selects and applies the appropriate Skill based on the nature of your request.

• "Create an Excel spreadsheet tracking monthly expenses with formulas for totals"
• "Turn this meeting notes document into a PowerPoint presentation"
• "Generate a PDF report summarizing this data"
• "Build a financial model in Excel with scenario analysis"

Each of these prompts automatically triggers the relevant Skill. You can see which Skill is being used by watching Claude's chain of thought as it works through your request.

Working with Existing Files

Skills also enable Claude to work with files you upload. Provide an .xlsx, .pptx, .docx, or .pdf file, and Claude will create an updated version—performing analyses, adding slides, or suggesting edits—without modifying your original file.

Anthropic's built-in Skills are tested and maintained by Anthropic. Custom Skills you create are private to your individual account and are not shared with other users.

3005 Creating Custom Skills

The most accessible way to create a custom Skill is through conversation with Claude. No coding or manual file creation is required—Claude handles the technical structure for you.

The Creation Process

1. Start a new chat and tell Claude what you want to create — for example, "I want to create a skill for writing quarterly business reviews"
2. Answer Claude's interview questions about your workflow, desired outputs, and example use cases
3. Upload reference materials such as templates, style guides, brand assets, or past work you are proud of
4. Save the Skill file Claude generates — it immediately appears in your Skills list and is ready to use

The interview step is where most of the value is created. Claude asks targeted questions: what should this Skill do, what makes good output for this type of work, and when would you use it. Providing concrete reference materials—templates, examples of past work, style guides—gives Claude specific anchors for the Skill's behavior.

Managing Your Skills

Once saved, your custom Skill appears alongside Anthropic's built-in Skills. Claude automatically invokes it whenever you work on a relevant task—no manual triggering needed.

Because Skills are invoked automatically, the more precisely you define the Skill's scope during creation, the more reliably Claude will apply it at the right moments.

3106 Skills vs. Projects

Both Skills and Projects give Claude additional context, but they serve fundamentally different roles: projects store knowledge, skills perform tasks.

The two features complement each other. A Skill can reference knowledge stored in a Project—for example, a "customer call prep" Skill might pull from customer profiles uploaded to a Project's knowledge base. The Project provides the what (information); the Skill provides the how (process).

• Projects: passive reference, always available within the project context
• Skills: active procedures, triggered when the task matches the Skill's purpose
• Combined: a Skill draws on Project knowledge to execute a task with full context

A customer support team might use a Project to store product documentation and FAQs, while a Skill encodes the step-by-step process for drafting responses that meet their quality standards. Each tool does what the other cannot.

3201. What Connectors Are

Connectors give Claude access to the tools, files, and data you already use at work. Instead of copying and pasting information into each conversation, Claude can read from and act within your connected applications directly.

Depending on the connector and the permissions you grant, Claude can search your files, retrieve documents, analyze data, create content, update records, and execute tasks—all from within a single conversation.

There are two categories of connectors. Web connectors link Claude to cloud services accessed through the browser. Desktop extensions run locally through the Claude Desktop app and provide access to native applications and your local file system.

Web connectors are the easier starting point because they work in any browser and require no additional software. Desktop extensions expand Claude's reach to local applications but require installing the Claude Desktop app first.

3302. Finding and Setting Up Connectors

Anthropic maintains a directory of recommended connectors organized into two tabs: Web (cloud services) and Desktop extensions (local tools). You can access it directly at claude.ai/directory, or by clicking the + button in the lower left of the chat window and selecting Connectors.

Setting Up a Web Connector

Setting up a web connector takes only a few minutes and follows the same pattern regardless of which service you choose.

1. Navigate to the connectors directory, or open + > Connectors in any chat
2. Select the connector you want and click Connect
3. Sign in with your existing credentials on the service's login page
4. Review the permissions Claude is requesting and authorize access
5. Return to Claude and test with a simple request, such as "Can you access my [tool name]?"

Installing Desktop Extensions

Desktop extensions require the Claude Desktop app, which runs on macOS and Windows. They let Claude interact with local applications, your file system, and native features that are not accessible through the browser.

1. Download and install the Claude Desktop app
2. Open the app and navigate to Settings > Extensions
3. Browse available extensions and click Install
4. Follow any additional setup steps specific to that extension

3403. Using Connectors in Your Work

Once a connector is active, Claude draws on it automatically when your request requires information from that source. You don't need to name the connector in every message—just ask as you would ask a colleague who has access to the same tools.

Project Management

With a project management connector such as Asana, Linear, or Jira, Claude can query task status, create new items, and summarize project progress without you switching applications.

• "What are my highest priority tasks due this week?"
• "Create a new task for reviewing the Q4 budget proposal"
• "Summarize the status of our product launch project"

Communication and Documentation

Communication connectors such as Slack and Gmail let Claude search past conversations and draft replies. Documentation connectors such as Notion, Google Drive, and Confluence let Claude retrieve and summarize internal knowledge.

• "Find the email thread where we discussed the vendor contract"
• "Draft a reply to the latest message in the #marketing channel"
• "Search our documentation for our brand voice guidelines"
• "Summarize the meeting notes from last week's product review"

3504. Security and Permissions

When you connect Claude to an external service, you grant it access to read—and in some cases modify—data within that service. Understanding how permissions work helps you make informed decisions about what to connect.

How Permissions Work

Connector permissions are scoped to what the connector needs. You can toggle individual permissions on and off within each application's settings menu. Claude can only access data you have access to—connecting your work email does not expose your CEO's inbox, only your own.

• Scoped access: each connector requests only the permissions it needs, not blanket access
• Your data only: Claude sees what you see—nothing from other users' accounts
• Revocable at any time: disconnect through Claude's settings or through the third-party service's own security settings

Custom Connectors and Trusted Sources

Beyond Anthropic's directory, you can find or build custom connectors—the MCP standard is open, so anyone can create one. Apply the same caution you would with Skills: only install connectors from sources you trust.

• Check the connector's publisher before installing
• Review permissions before authorizing—only approve what the connector genuinely needs
• Remove connectors you no longer use to reduce your permission surface

3601 What Enterprise Search Is

Enterprise Search works like a pre-built Project for your entire organization. Your company's knowledge base is already loaded, so you can ask questions immediately without setting up connectors yourself each time.

Unlike regular chats with connectors enabled, Enterprise Search is specifically designed for information gathering. It uses custom instructions configured by the Anthropic team, making it optimized for organizational knowledge retrieval rather than general conversation.

When you submit a question, Claude searches across all connected tools simultaneously — SharePoint documents, Slack conversations, Gmail threads, Google Drive files — and returns a single unified response with citations so you can trace every fact back to its source.

3702 Questions Enterprise Search Can Answer

Enterprise Search is most valuable for questions that span multiple systems or require pulling together information from across the organization. Five distinct use-case patterns cover the majority of practical needs.

Onboarding is a particularly strong use case. New team members can ask how systems work, who owns what area, and what tools the team uses — questions that normally require interrupting experienced colleagues.

3803 Setting Up Enterprise Search

Setup follows a two-step sequence: an admin (Owner) configures the organization-level integration first, then individual users authenticate their own accounts to start searching.

Admin Configuration

The Enterprise Search project is enabled by default for all Team and Enterprise organizations, but an Owner must complete initial setup before anyone can use it.

1. Click "Ask Your Org" in the left sidebar
2. Click "Set up for your org" to begin (or "Disable" to turn the feature off)
3. Connect a Documents connector (Google Drive or SharePoint) and a Chat connector (Slack or Microsoft Teams) — both are required; Email is optional
4. Click "+ Add more" to add any additional tools the team needs
5. Customize the project name — it appears as "Ask [Name]" in everyone's sidebar
6. Add a description, then click "Finish set up"

User Authentication

Once an admin completes setup, the "Ask {Org Name}" project appears starred in every team member's sidebar. Users then connect their personal accounts to the services they want to search.

1. Click the project in your sidebar
2. Follow the guided onboarding flow to connect to recommended services
3. Authenticate with each service you want to search (Slack, Google, Microsoft 365, etc.)
4. Start asking questions about your organization's knowledge

3904 Security and Permissions

A common concern when connecting organizational data to an AI tool is whether sensitive information might become visible to people who shouldn't see it. Enterprise Search addresses this through permission inheritance.

• Search results reflect your existing access rights in each connected tool — no new permissions are granted
• Your conversations with Enterprise Search remain private
• Connected data is not indexed or stored separately by Claude

This design means that access control stays where it was originally set — in SharePoint, Google Drive, Slack, or whichever system owns the data. Enterprise Search queries those systems on your behalf rather than ingesting and re-storing the content.

4001 What Is Research?

Standard Claude responses draw on what the model already knows. Research changes this by actively gathering new information—autonomously deciding what to search next based on what it finds, pursuing leads, and filling gaps without you needing to direct each step.

Think of it as a skilled research assistant who can spend hours gathering information, cross-referencing sources, and compiling a comprehensive report—except the work completes in minutes rather than hours.

• Conducts multiple searches that build on each other
• Autonomously decides what to investigate next based on prior findings
• Explores different angles of your question automatically
• Synthesizes findings from the web and any connected integrations
• Delivers a well-organized report with verifiable citations

Research is particularly valuable when a thorough understanding requires pulling together information from multiple sources, comparing different perspectives, and synthesizing findings into actionable insights. It is not designed for quick factual lookups.

4102 When to Use Research

Choosing the right Claude capability for your task saves time and produces better results. Research is powerful but intentionally slower than a quick web search—use it when thoroughness and citations matter more than speed.

Tasks where Research excels

Research produces the most value for tasks where thoroughness, multiple perspectives, and verifiable citations matter more than how fast you get an answer.

• Market analysis and competitive research
• Planning complex projects like team offsites or product launches
• Synthesizing information from email, calendar, and documents
• Creating technical documentation that draws from multiple sources
• Preparing briefings that require current, verified information

When to choose a lighter tool

If you need today's stock price, a company's address, or any answer that requires only one or two sources, use web search instead. Research would spend time gathering information you simply do not need.

Extended thinking is the better choice when you are working on math problems, debugging code, or logical analysis—tasks where the answer comes from reasoning through a problem rather than gathering external information.

4203 How Research Works

When you enable Research, Claude follows a structured four-step cycle. Each step builds on the previous one, moving from planning through searching to synthesis and final citation.

Planning and searching

Step 1: Claude activates extended thinking to break down your request, identify what information it needs, and plan how to investigate different angles of your question before any searches begin.

Step 2: Rather than running a single search, Claude conducts many searches that build on each other. It determines what to investigate next based on what it finds, pursuing promising leads and filling in gaps as it goes.

Synthesizing and citing

Step 3: After gathering information from multiple sources—including the web and any connected integrations like Gmail, Google Calendar, or Google Drive—Claude compiles everything into a comprehensive, well-organized report.

Step 4: Every claim in a Research report links back to its source. This makes it straightforward to verify information and dig deeper into specific findings when needed.

4304 Writing Effective Research Prompts

Because Research can take between 5 and 45 minutes depending on complexity, a well-crafted prompt pays dividends. Vague prompts produce broad reports; specific prompts produce focused, actionable ones.

1. Be specific about your goals—state what decision or outcome the research supports
2. Specify the sections or structure you want—Claude organizes findings around the structure you provide
3. Include relevant constraints such as budget ranges, timelines, or geographic requirements
4. Ask Claude to help refine your prompt before enabling Research if you are unsure how to frame your question

The strong prompts share a pattern: they name a specific outcome (investment decisions, venue selection), list the dimensions to cover, and include implicit constraints that help Claude focus on relevant options rather than casting too wide a net.

4405 Research with Connected Integrations

When Google Workspace or other integrations are connected, Research gains access to your own context—emails, calendar events, and documents—alongside public web sources. This lets Claude bridge internal and external knowledge in a single report.

• Summarize email and Slack discussions about a project, then research industry best practices for similar initiatives
• Review calendar commitments for an upcoming week and research each company you are meeting with
• Find internal documents about your pricing strategy and compare them to how competitors are positioning themselves publicly

You can also turn off web search entirely to limit Research to your connected tools only. This is useful for internal-only questions—for example, finding what your team discussed about a product launch across Slack messages and shared documents.

Disabling web search and querying only internal tools turns Research into a personal knowledge management layer—useful for onboarding, tracking project history, and reviewing decisions made across scattered documents.

4501. Claude Across Professional Roles

Claude applies to virtually every professional role. The use cases here are organized by job function, so you can quickly identify which ones match your daily work and start applying them right away.

The gallery continues to grow, making it a useful resource to revisit as your role or responsibilities change over time.

General Professional Use Cases

Three use cases apply regardless of industry or title — useful for almost any professional who manages projects, communicates with stakeholders, or works with customer feedback.

• Generate project status reports — keep stakeholders informed with clear, consistent updates
• Analyze patterns in user feedback — extract insights from customer comments and survey responses
• Package your brand guidelines in a skill — create a reusable Claude skill that applies your brand standards automatically

4602. Sales, Marketing, and Finance

Sales

Sales professionals can use Claude to accelerate deal preparation, build competitive intelligence resources, and produce clear pipeline reports.

• Build a battle card library — create competitive intelligence resources that help your team win deals
• Prepare for sales deals — research prospects and organize talking points before important meetings
• Create sales reports — turn pipeline data into clear, actionable reports

Marketing

Marketers can use Claude to analyze performance data and repurpose content efficiently across different channels and audiences.

Two high-impact starting points: analyzing campaign performance to extract strategy-informing insights from metrics, and adapting content across platforms to repurpose existing material for different channels without starting from scratch.

Finance

Finance professionals can use Claude to build analytical models, draft complex documents, and make sense of spreadsheets they did not create.

• Build financial models — create and refine financial projections with Claude's assistance
• Draft investment memos — structure and write investment analyses more efficiently
• Understand and extend an inherited spreadsheet — decode complex logic and safely add new functionality

4703. HR, Legal, and Research

HR

HR teams can use Claude to create onboarding materials that are comprehensive and tailored to specific roles, rather than relying on generic company-wide templates.

Creating role-specific guides at scale is time-consuming when done manually. Claude can draft and adapt these materials quickly, letting HR professionals focus on reviewing and refining rather than writing from scratch.

Legal

Legal professionals work with large volumes of documents and tight timelines. Claude can organize case materials and surface patterns that would otherwise require hours of manual review.

By handling organizational and pattern-finding work, Claude allows legal teams to direct their expertise toward judgment and strategy rather than document sorting.

Research

Researchers can use Claude to plan systematic literature reviews and to verify statistical calculations against raw data before publishing or presenting findings.

These role-specific examples are a starting point. The Use Case Gallery contains the full collection with step-by-step instructions, organized to help you find scenarios that fit your specific work context.

4801. Claude Beyond Claude.ai

Claude.ai is one interface to Claude, but the underlying intelligence is available in several specialized tools built for the environments where you already work. The goal of each tool is the same: instead of moving your work into Claude, Claude moves into your work.

• Claude Code — software development in your terminal or IDE
• Claude in Slack — team collaboration within your workspace
• Claude for Excel — spreadsheet analysis via an in-app sidebar
• Claude for PowerPoint — slide editing via an in-app sidebar
• Claude for Chrome — browser assistance via a Chrome extension

Claude Code is the most autonomous of these tools, capable of running terminal commands and managing entire development workflows. The sidebar tools (Excel, PowerPoint, Chrome) give Claude visibility into your current context and apply changes directly within the application.

4902. Claude Code

Claude Code integrates with your existing development environment. You do not switch to a separate interface — it works alongside your terminal and IDE so your current tools remain in place.

Core capabilities

• Build features: describe what you need in plain English and Claude writes the code, runs tests, and creates commits
• Debug: paste an error message and Claude analyzes your codebase to identify and fix the problem
• Navigate unfamiliar codebases: ask questions about how different parts connect and work together
• Automate repetitive tasks: fix lint errors, resolve merge conflicts, write release notes

Claude Code also works inside Slack. When a bug report or feature discussion appears in a channel, tagging @Claude can start a Claude Code session that uses the surrounding conversation as context.

The full development loop — describing a feature, having Claude write and test it, then commit the result — happens without leaving your terminal.

5003. Claude in Slack

Claude integrates directly into Slack as a participant in channels and threads. It can also bring Slack content — conversations and shared documents — into your Claude sessions.

• Draft responses to messages, summarize long threads, or clarify complex discussions without leaving Slack
• Prepare for meetings by pulling together relevant conversations and shared documents from your workspace
• Help new team members understand ongoing projects by reviewing channel history
• Answer questions about industry trends, technical concepts, or company information mid-conversation
• Hand off coding tasks from a bug report: tag @Claude and it starts a Claude Code session using surrounding context

The Slack and Claude Code bridge is especially useful for development teams. Instead of copying a bug report into a separate tool, you tag Claude in the thread and it can open a coding session with the issue already in context.

Claude in Slack reduces tool switching. You stay in the workspace your team already uses, and Claude participates in conversations where the work is already happening.

5104. Claude for Excel

When you open Claude for Excel, it explores the full workbook — not just listing tabs but explaining the data flow. For a multi-tab headcount planning model, it will trace how an Assumptions tab feeds into Headcount, which feeds into Compensation, rolling up into a Summary tab.

Exploring and debugging workbooks

When you inherit a workbook you did not build, Claude can describe it the way its original author would. During an initial walkthrough, it also proactively flags errors — in a demo, it identified a cascading #REF error in cell E15 of the Headcount tab before being asked to look for issues.

• Identify #REF!, #VALUE!, and circular reference errors and trace them to their source
• Explain what the broken formula was trying to calculate
• Apply the fix directly — dependent cells update automatically through existing formula relationships

Scenarios, visualizations, and new formulas

For scenario modeling, you describe the change in plain language. Asking Claude to update attrition from 10% to 15% causes it to make the edit and then walk you through what changed across all dependent tabs — headcount projections, compensation costs, and summary figures — because the existing formula relationships do the recalculation.

• Create pivot tables from headcount data by describing the breakdown you need
• Generate bar charts or stacked bar charts without navigating Excel menus
• Add calculated columns that reference other sheets, preserving all formula dependencies

5205. Claude for PowerPoint

Claude for PowerPoint runs as a sidebar in Microsoft PowerPoint. It drafts, edits, and restructures presentations through conversation while preserving your existing template and brand styling.

• Rewrite or tighten slide copy: shorten bullets, add speaker notes, or adjust tone for a specific audience
• Restructure an existing deck: reorder sections, split dense slides, or merge overlapping ones
• Apply consistent formatting across all slides — titles, bullet styles, and layouts — without editing each slide manually
• Get layout or chart-type suggestions for individual slides based on the point being made

All edits appear directly in your open presentation. There is no export or import step — Claude's changes are visible immediately in the deck you are working on.

The tool is especially useful for the later stages of deck creation: tightening copy, making formatting consistent, and adjusting structure once a first draft already exists.

5306. Claude for Chrome

Unlike opening Claude.ai in a separate tab, Claude for Chrome retains context as you move between tabs and tasks. This makes it useful for workflows that span multiple web tools — CRMs, dashboards, and internal applications that do not have their own Claude integrations.

• Summarize articles, research papers, or web pages while browsing
• Draft email responses or help manage your inbox
• Automate repetitive form filling
• Test website features or navigate multi-step workflows without manually clicking through each step

The cross-tab context retention is what distinguishes Claude for Chrome from a standalone chat window. Claude can pull information from multiple open tabs and apply it to what you are currently working on.

As a research preview, the extension's behavior and available actions may change over time. Using it on well-known, trusted websites minimizes the risk of unexpected actions.

5401. What You've Covered in Claude 101

Claude 101 has taken you from first conversation to a working understanding of Claude's full capability set—prompting, organizing work, extending Claude to external tools, and applying it across professional roles.

Getting Started and Getting Better Results

• Claude is an AI assistant built to be helpful, harmless, and honest—a thinking partner for complex work, not just a chatbot.
• You can access Claude through web, desktop, and mobile apps, with conversations syncing across devices.
• Effective prompts set the stage (context), define the task (action), and specify rules (format and style).
• Treat first responses as starting points and refine through conversation—iteration is the core skill.

Common challenges—generic responses, wrong tone, missing context—are correctable by adding specificity. The four AI Fluency competencies give you a mental model for diagnosing and fixing these issues.

Organizing Work and Expanding Claude's Reach

• Projects create dedicated workspaces with persistent knowledge, custom instructions, and team collaboration.
• Artifacts are standalone outputs—documents, code, diagrams, interactive tools—created alongside your conversation.
• Skills are instruction packages that teach Claude specialized workflows, including built-in and custom options.

Connectors link Claude to external tools—Google Workspace, Slack, Notion, and more—so it works with your actual organizational data rather than generic knowledge. Enterprise Search provides a dedicated project for searching across your organization's knowledge sources, while Research mode conducts systematic, multi-source investigations automatically.

• Claude applies across professional roles: sales, marketing, finance, HR, legal, and research.
• Beyond claude.ai, Claude is available through Claude Code, Slack, Excel, and Chrome.
• Research mode handles multi-source investigations that would take hours to complete manually.

5502. Resources for Continued Growth

Anthropic provides a set of free courses and documentation that extend what you've built here. These resources cover AI concepts, product-specific workflows, and reference material for daily use.

Learning and Reference Resources

• AI Fluency courses — free courses on effective AI collaboration.
• AI Capabilities and Limitations — free introductory course on what AI can and can't do.
• Use Case Gallery — step-by-step guides and prompts for specific workflows.
• Anthropic Help Center — detailed documentation and troubleshooting.
• Prompting documentation — comprehensive guide to getting the best results from Claude.

Product-Specific Courses

• Claude Code in Action — free course on using Claude for software development workflows.
• Introduction to Claude Cowork — free course on Claude's desktop companion for multi-step work.
• Connector Directory — browse and connect your tools to Claude.

Each product course goes deeper into a single surface—useful once you know which part of Claude's ecosystem fits your daily work best.

5603. Putting Knowledge Into Practice

The skills from this course sharpen with use. The gap between understanding Claude and using it well closes through repeated, real-work application—not through more reading.

Start simple and build from there. You don't need Projects, Connectors, or Research mode on day one. A well-crafted prompt on a single real task will already reveal what Claude can do.

1. Choose a single recurring task from your actual workload.
2. Write a prompt using the context-action-format structure from the course.
3. Review Claude's response critically—use your domain expertise to evaluate it.
4. Refine through conversation until the output meets your standard.
5. Note what worked and apply that pattern to future tasks.

Claude is designed to be a collaborator, not a replacement. The best results come when you bring your expertise, context, and judgment to the conversation—Claude handles the generative heavy lifting; you supply the standards.

5701 Certificate of Completion

The final step in Claude 101 is a completion quiz. Passing it earns your Claude 101 Certificate of Completion.

The quiz draws on material from all four main sections of the course.

1. Meet Claude — what Claude is, your first conversation, getting better results, and the desktop app
2. Organizing your work and knowledge — projects, artifacts, and skills
3. Expanding Claude's reach — tool connections, enterprise search, and research mode
4. Putting it all together — use-cases by role and other ways to work with Claude

After completing the quiz, your certificate reflects mastery of the full Claude 101 curriculum, from basic conversations to advanced research and enterprise features.