What You Will Learn

• Master Claude's Chat Mode vs Work Mode
• Set up a marketing workspace with CLAUDE.md and MEMORY.md
• Build reusable skills to encode your brand voice
• Automate content repurposing into blog, newsletter, and social assets
• Integrate MCP connectors for visuals, lead data, scheduling, and automation
• Create live dashboards and implement AI SEO to track competitors and boost visibility

Study Guide

# Master Claude for Marketing in 72 Minutes (FULL COURSE) ## Introduction: Why This Course Changes Everything Let me be straight with you. Most people use AI tools like Claude the wrong way. They open a chat window, type a prompt, get a response, copy it, and paste it somewhere. Then they do it again. And again. It's like using a chainsaw to butter your toast,technically possible, but you're missing the point entirely. This course is different. We're going to build something that functions like a full marketing department, operating from inside Claude. Not a chatbot that gives you advice. An actual system that creates content, generates leads, tracks competitors, produces visual assets, and schedules posts across every platform you use. All from one workspace. All orchestrated through natural language. Here's what you'll walk away with: - A complete understanding of Claude's work mode and why it's the foundation of any serious AI marketing operation - A structured workspace that organizes your content, leads, and assets automatically - Reusable skills that encode your brand voice so every piece of content sounds like you,not like generic AI output - A content repurposing pipeline that turns one video into a blog post, newsletter, and social media content - Connections to the tools you already use through MCP connectors - A lead generation system that finds verified contacts and drafts personalized outreach - Live dashboards that track competitors and your own performance - A scheduling system that publishes content without leaving Claude - An AI SEO strategy that gets you cited in ChatGPT, Claude, and Gemini responses This isn't theory. These are operational systems you can set up today. Let's get into it. --- ## Section 1: Understanding Claude's Two Modes ### Chat Mode vs. Work Mode Claude operates in two fundamentally different ways, and understanding the difference is the difference between playing with a toy and running a machine. **Chat mode** is what you've probably used before. You type a question, Claude responds with text. It's conversational, it's useful for brainstorming, and it's completely limited. Claude can't touch your files. It can't save anything. It can't connect to your tools. Every conversation is a fresh start with no memory of what came before. **Work mode** is where the real power lives. When you enable work mode, Claude operates inside designated folders on your computer. It reads files. It creates documents. It organizes content. It connects to external applications. It runs scheduled tasks while you sleep. It's not having a conversation with you,it's doing actual work. Think of chat mode as asking a consultant for advice. Think of work mode as hiring an employee who has keys to your office, access to your files, and a list of tasks to complete. ### Why Work Mode Matters for Marketing Marketing is production. You're producing blog posts, emails, social content, images, videos, and reports. Production requires file systems, organization, and consistency. Chat mode gives you isolated conversations. Work mode gives you an operational workspace where everything accumulates and compounds. When Claude works inside your marketing folder, it can: - Save every generated asset to the right subfolder automatically - Reference your previous content to maintain consistency - Scrape and analyze external data from competitors - Build and maintain databases of leads and performance metrics - Execute recurring tasks on a schedule The shift from chat mode to work mode is the shift from asking questions to getting things done. ### The Permission System Work mode runs on a strict permission model. You choose which folders Claude can access. This isn't a limitation,it's a feature. It creates security boundaries while giving you precise control over what Claude touches. When you set up your workspace, you grant access to a specific folder. Everything Claude reads, creates, or modifies happens inside that boundary. Your personal files, your other projects, your system files,all off-limits unless you explicitly include them. This means you can run a complete marketing operation without worrying about Claude accidentally accessing something it shouldn't. --- ## Section 2: Setting Up Your Marketing Workspace ### The Folder Structure Before you do anything else, create a dedicated folder for your marketing operations. Name it something obvious,"Claude Marketing" works fine. This folder becomes your central command center. Inside that folder, create subfolders for different content types: Claude Marketing/ ├── content/ (blog posts, long-form articles) ├── social/ (platform-specific posts) ├── emails/ (newsletters, outreach drafts) ├── leads/ (prospect data, CSV exports) └── assets/ (images, videos, graphics) This structure means Claude automatically organizes everything it creates. No more hunting through your Downloads folder for that blog draft you generated three weeks ago. ### The Onboarding Questionnaire When you first open work mode in your marketing folder, Claude should run through an onboarding questionnaire. This is a structured interview that captures essential information about your business. The questions cover: - What your business does and what you offer (keep it to two lines or fewer) - Who your target audience is - Which marketing functions you want to run from this workspace - Your brand voice preferences Here's what a real answer might look like: "We provide influencer marketing services connecting AI creators with brands. We want to run content creation, social posting, email newsletters, lead generation, and outreach from this folder. Our brand voice is simple, direct, and written at a fifth-grade reading level." That single response gives Claude everything it needs to start operating with your brand in mind. ### CLAUDE.md: The Workspace Constitution CLAUDE.md is the instruction file that defines how Claude behaves inside your workspace. It contains your company details, tone guidelines, operational preferences, and behavioral rules. Here's the critical thing: keep this file concise. Claude reads CLAUDE.md with every single message you send. A bloated file consumes context windows and slows down responses. Focus on the essentials,what Claude needs to know to operate effectively. A good CLAUDE.md might include: - Your business description and offerings - Your target audience - Your brand voice guidelines - Your content standards - Any hard rules (like "never send emails without approval") That's it. The details live in skills, which we'll cover in the next section. ### MEMORY.md: The Persistent Brain MEMORY.md is where Claude stores contextual information that accumulates over time. This file grows as you work. It captures: - Key facts about your business - Decisions made during previous sessions - Stored offers and their results - Important contacts and relationships - Lessons learned from campaign performance You can update the memory file two ways. You can edit the markdown file directly, or you can tell Claude conversationally: "Update the memory file,we changed our product line." The memory file is what transforms Claude from a tool with amnesia into a system that learns and compounds knowledge. --- ## Section 3: Building Skills for Consistent Brand Output ### What Skills Actually Are Skills are reusable instruction sets that Claude executes when triggered by a slash command. Type "/email_voice" and Claude runs through a complete set of instructions for generating emails in your specific voice. Type "/blog_post" and Claude follows your established structure for long-form content. The genius of skills is consistency. Every time you generate content, it follows the same standards. You don't have to restate your preferences. You don't have to hope Claude remembers what you asked for last time. The skill encodes everything. ### Creating Skills from Real Data The most powerful skills aren't written from scratch,they're extracted from your actual output. Claude can analyze your real marketing materials and reverse-engineer your voice. Here's how it works with email: 1. Connect Claude to your Gmail through the connector 2. Instruct Claude to scrape your last 20-30 sent emails 3. Claude analyzes greeting patterns, sentence structures, vocabulary choices, sign-off conventions, and tone tendencies 4. Claude generates a skill file that captures your authentic voice The same process works for blog posts, newsletters, LinkedIn content, or anything else you've published. By processing 20-30 real examples, Claude constructs a skill that faithfully reproduces your voice,not a generic AI approximation. This is the difference between content that sounds like you and content that sounds like AI pretending to be you. ### Skill Components A well-constructed skill typically includes: - **When to use**: The trigger conditions and context - **What it does**: A description of the task and expected output - **Steps**: The sequential workflow instructions - **Hard rules**: Non-negotiable constraints - **Style references**: Voice parameters, formatting conventions, structural preferences - **Storage directives**: Where generated files should be saved For example, an email skill might include the hard rule "draft only,never send emails directly." That constraint ensures Claude assists without overstepping. ### Managing Your Skills You can review, edit, and organize your skills through the customization panel. You can edit the markdown files directly or make changes conversationally: "Change the email skill to eliminate exclamation points in subject lines." This iterative refinement lets your brand voice evolve without rebuilding skills from scratch. As your preferences change, your skills change with them. --- ## Section 4: The Content Repurposing System ### The Architecture of Repurposing You've probably heard that you should repurpose content across platforms. The problem is that manual repurposing takes forever. You record a video, then you have to write a blog post, then a newsletter, then LinkedIn posts, then Instagram captions. That's hours of work. The content repurposing orchestrator solves this. It's a parent skill that coordinates multiple child skills,each one responsible for a specific content format. Here's the workflow: [Primary Content: YouTube video URL] ↓ [Content Repurposing Orchestrator] ↓ +----------------+----------------+ ↓ ↓ ↓ Blog Post Skill Newsletter Skill LinkedIn Post Skill ↓ ↓ ↓ Saved to content/ Saved to emails/ Saved to social/ ### How Execution Works When you trigger the orchestrator with a content URL, Claude: 1. Retrieves and reads the source content (using Firecrawl for platforms Claude can't access natively) 2. Passes the content to each child skill sequentially 3. Each child skill applies its specific formatting, voice, and structural requirements 4. Generated files are saved to the appropriate subfolders automatically A 30-minute YouTube video becomes three ready-to-publish assets in minutes. A podcast episode becomes a newsletter, a blog post, and five social media updates. One piece of content generates an entire distribution ecosystem. ### The Benefits This system delivers efficiency that manual repurposing can't touch: - **Consistency**: Every piece of content follows established brand standards - **Speed**: What took hours now takes minutes - **Scalability**: New channels can be added as new child skills - **Voice preservation**: All outputs reflect your authentic brand voice ### Advanced Skill Composition Skills can be nested within other skills, creating tiered automation. Individual content-type skills combine into the repurposing orchestrator. The orchestrator can combine with lead generation and publishing automation into an entire marketing operating system skill. The hierarchy keeps everything modular. You can update one child skill without affecting the others. You can add new formats without rebuilding the system. --- ## Section 5: Visual Asset Generation ### The Gap in Claude's Capabilities Here's a limitation you need to know: Claude cannot generate images natively. It's a text-based model. But that doesn't mean you can't produce visual assets from within Claude. The solution is an MCP connector to a creative studio platform. These platforms offer MCP servers that connect Claude to image generation models, video creation tools, and graphic design capabilities. Once connected, you can generate visuals without leaving your workflow. ### Connecting a Creative Studio The integration process follows a standard pattern: 1. Copy the studio's remote MCP server URL 2. Add a custom connector in Claude with that URL 3. Authorize the connection (requires an account with the creative platform) 4. Set permissions to "always approve" for frictionless operation 5. Test generation capabilities Once connected, the creative studio appears in your connector interface alongside your other tools. ### Image Generation Workflows **Explainer infographics**: Request an infographic on any topic. Claude researches facts using Firecrawl for web data, selects visual elements, and generates a professional infographic. You can specify aspect ratio (square for Instagram, tall for Stories) and style preferences. **Product infographics**: Provide a product URL. Claude scrapes the product specifications and features, then generates an informative infographic ideal for e-commerce listings or promotional materials. This workflow scales easily,you could process 5-10 products in a single session. **Style transfer**: If you see an infographic style you like, provide it as a reference image and instruct Claude to generate new content in that style. Claude can also create a new skill for future use, building a growing library of visual styles. ### Video Advertisement Generation The product-to-ads skill takes visual generation to the next level. It creates user-generated content (UGC) style video advertisements with synthetic actors. Here's the process: 1. Upload product images and specify target audience characteristics 2. Claude generates a synthetic UGC actor matching the target demographic 3. The actor is placed in a realistic scene holding the product 4. Claude writes a short-form UGC script (5-30 seconds) 5. The final video ad is rendered with the actor speaking the script The output features realistic character consistency, natural dialogue, and simulated scene cuts that make the video appear organically produced. These videos work for TikTok, Instagram Reels, YouTube Shorts, or paid advertising campaigns. The ethical considerations here matter. Synthetic actors in advertising raise questions about disclosure and consumer perception. Consider how you handle transparency when using AI-generated presenters. --- ## Section 6: Lead Generation and Personalization ### The Lead Data Problem Finding accurate, actionable lead data is one of the hardest parts of marketing. Claude has a vast knowledge base, but it cannot reliably locate verified emails, current job titles, phone numbers, or company funding data on its own. That information lives in specialized data platforms. This is where MCP integrations become essential. Connecting Claude to a lead generation platform solves the data accuracy problem. ### How the Lead Discovery System Works **Step 1 , Define target criteria**: Tell Claude what you're looking for. For example: "AI SaaS tools, funded, US-based companies seeking AI creator partnerships." **Step 2 , Claude triggers the data platform**: Claude queries the connected lead database platform, which searches across 150+ data sources using waterfall enrichment to find verified matches. **Step 3 , Structured results**: The platform returns organized data including: - Contact names and job titles - Verified work emails - Direct phone numbers - Company location and employee count - Funding history and revenue ranges - Technology stack signals - Website traffic metrics **Step 4 , CSV export**: Claude generates a CSV file with all lead data, easily shareable across teams or imported into CRM systems. ### Waterfall Enrichment Explained Waterfall enrichment is a data acquisition method where the platform searches sequentially across multiple databases until it finds valid, matching information. Instead of relying on a single source that might have incomplete or outdated records, the system queries sources in sequence, finding the first reliable match. This approach routinely triples data coverage and quality compared to single-source data acquisition. Your emails, phone numbers, and company details are verified and current. ### Personalization at Scale The combination of lead data with your email voice skill creates a powerful outreach system: 1. Claude identifies the top contacts from each target company 2. The email voice skill generates personalized drafts for each individual 3. Each email references company-specific details (funding, product relevance, shared connections) 4. Drafts appear in Gmail for review, with custom subject lines and tailored openings 5. You approve and send,or schedule for automated distribution This is personalization at scale. Every email is unique, references specific details about the recipient's company, and sounds like you wrote it. ### The Context Advantage Advanced lead platforms maintain a context file about your business,company description, ideal buyer personas, existing partnerships. This context enables more intelligent lead matching. The system doesn't just find "marketing directors." It finds marketing directors who match your specific ideal customer profile. --- ## Section 7: Live Artifacts and Marketing Dashboards ### What Live Artifacts Are Live artifacts are interactive mini-applications that run inside Claude. They pull fresh data from connected sources and present it in a visual, interactive interface. Unlike static reports, live artifacts update continuously and accept user input for actions like adding notes or flagging opportunities. ### The Three Components A functional dashboard requires three components working together: 1. **Scheduled task**: An automation that runs at specified intervals (e.g., every morning at 7:00 AM) to collect and process data 2. **Data scraper**: Firecrawl extracts content from external platforms (competitor YouTube channels, blogs, industry news) 3. **Database**: A connected database like Supabase stores structured historical data in tables ### Example: Competitor Content Tracker Let me walk you through a practical implementation,tracking competitor YouTube performance. **Data collected per video**: - Title and publication date - View count and view velocity - Engagement metrics (likes, comments) - Outlier score (performance relative to channel average) - Content topic and format **Dashboard features**: - Sortable list of competitor videos by performance - Outlier flagging for exceptional performers - Idea board for saving inspiration and internal content ideas - One-click addition of new content concepts This dashboard runs automatically. Every morning, it collects the latest data. When you open Claude, you see a live view of what your competitors are doing. ### Building Custom Trackers The dashboard creation process is conversational. You tell Claude what you want to track, and it builds the system. 1. Specify what to track (competitors, own performance, keyword rankings, product mentions) 2. Identify the data sources (YouTube, Instagram, blogs, news sites) 3. Define dashboard metrics and views Claude then builds the system by creating the appropriate database tables, configuring the scraping schedule, developing the interactive artifact interface, and connecting all components into a working dashboard. ### Example: Personal Video Performance Tracker Beyond competitor tracking, you can monitor your own content performance. A personal YouTube tracker displays all your published videos with view counts, like percentages, and engagement rates. The same system works for Instagram, LinkedIn, or blog content,any platform accessible through Firecrawl. --- ## Section 8: Connecting Unsupported Applications ### The Gap Problem Not every marketing tool offers a native Claude connector. Email platforms, niche CRMs, specialized scheduling tools, and custom databases may not appear in the connector browse menu. Without integration, you'd have to manually transfer data between Claude and your essential tools. That's a workflow killer. ### The Universal Adapter Solution Automation platforms like Zapier provide a universal integration layer supporting thousands of applications. The MCP connector for these platforms creates a two-way bridge between Claude and nearly any marketing tool. **Setup process**: 1. Create an account with the automation platform 2. Create a new MCP server 3. Select Claude work mode as the client 4. Connect desired applications to the MCP server 5. Add the MCP server URL as a custom connector in Claude 6. Configure permissions and begin operating ### Real-World Application Let's say you use a newsletter platform that has no native Claude support. Through the automation MCP, it becomes fully integrated: - Connected applications appear in the connector interface with their available actions - Actions include: list posts, view posts, list segment members, create subscribers, update contacts - Workflows execute directly from Claude prompts (e.g., "Add these 15 leads from the attached CSV to the newsletter subscriber list") - The entire operation happens without leaving the Claude interface ### When to Use an Automation Platform vs. Native Connectors **Native connectors** are preferable when they exist. They're simpler, faster, and more directly integrated. **Automation platform MCPs** are necessary when: - The desired tool has no native connector - Multiple niche tools need consistent integration - Custom workflows span several applications - You're non-technical and prefer not to configure your own MCP server The rule is simple: use native when available, use automation platforms when you need to bridge gaps. --- ## Section 9: Content Distribution and Scheduling ### The Scheduling Challenge Here's a common scenario: Claude generates a week's worth of content,posts, carousels, videos, newsletters. Now you need to publish it across Instagram, TikTok, LinkedIn, and everywhere else. If you're manually copying and pasting, you've just destroyed all the efficiency you gained from automated content creation. ### The Universal Scheduler Integration Social media management platforms connect to Claude via custom MCP connectors, enabling direct distribution. **Setup**: 1. Create an account and connect your desired social channels (Instagram, TikTok, LinkedIn, X, Facebook, Pinterest, YouTube, Threads, Blue Sky) 2. Generate an MCP server URL from the platform's API settings 3. Add as custom connector in Claude with "always allow" permissions 4. Begin scheduling content conversationally ### Scheduling Workflow **Within Claude work mode**: - Generate post content (text, images, videos) - Instruct Claude: "Schedule this for Instagram and LinkedIn at 7:00 PM Texas time today" - Claude uses the scheduling MCP to create the scheduled posts with captions **Advanced scenario with visual assets**: - Images generated in work mode may require file transfer to code mode - The same scheduling MCP connector operates from code mode - Claude handles uploads, caption generation, and cross-platform scheduling ### The Complete Content Pipeline When content creation combines with scheduling, a full end-to-end pipeline emerges: 1. Create source content (video, long-form article, podcast) 2. Run the content repurposing orchestrator 3. Generate platform-specific visual assets with the creative studio MCP 4. Schedule all assets across social platforms via the scheduler MCP 5. Monitor performance through the live artifact dashboard 6. Refine based on analytics and repeat The entire cycle runs from one workspace. You're not jumping between tools. You're not copying and pasting. You're directing an operation. --- ## Section 10: AI Search Engine Optimization ### The Changing SEO Landscape Traditional SEO focuses on ranking websites in Google search results. That still matters. But a transformative shift is underway. Users increasingly bypass traditional search engines and ask AI assistants directly for recommendations and information. When someone asks "Who is the best influencer marketing agency?" to ChatGPT, Claude, or Gemini, the AI composes an answer from its training data,and cites sources. Being that cited source represents enormous traffic potential. This is AI SEO: optimizing your content to rank prominently in responses generated by large language models. ### Data Types That Matter for AI SEO Connect an SEO analysis tool like Semrush to Claude to access: | Metric | Value | |--------|-------| | **Keyword search volume** | Monthly searches for target terms | | **Keyword difficulty** | Competition level for ranking | | **AI visibility** | How often the site appears in AI-generated responses | | **Competitor keywords** | Terms competitors rank for that you don't | | **Question data** | Specific queries from target audiences | | **Content gaps** | Opportunities competitors haven't addressed | ### The AI SEO Audit Process When Claude combines an SEO platform connector with content scraping, you get a comprehensive audit: 1. Claude analyzes your website structure, copy, and keyword targets 2. It identifies high-value, low-difficulty keywords matching your actual offerings 3. It generates copy recommendations for title tags, meta descriptions, and page content 4. It flags gaps where existing copy is vague or not search-optimized 5. It provides competitor comparisons showing what rival sites do well 6. It produces actionable recommendations for immediate implementation ### Example: Title Tag Optimization Let me give you a concrete example. A website with the tagline "Built by creators for creators" might receive the recommendation to change its title tag to "AI Influencer Marketing for Creators and Brands." The reasoning: the original tagline is evocative but vague. The recommended version explicitly states what the business does, immediately improving relevance signals for AI engines. AI systems need clarity to understand what your site offers. Evocative language doesn't help when a language model is trying to match your content to a user's query. ### Competitor Opportunity Analysis Beyond site analysis, the system evaluates competitors: - Identifies 2-3 primary search competitors - Analyzes their content strategies (post frequency, formats, topics) - Determines what performs well for them - Identifies gaps in their coverage that you can exploit - Compiles findings into readable markdown reports This gives you a roadmap for content creation that targets opportunities your competitors have missed. --- ## Section 11: Building the Complete Marketing Operating System ### Integration of All Components Now let's bring everything together. Each system we've covered,content creation, visual assets, lead generation, SEO, distribution,combines into a comprehensive marketing department running from a single workflow. The architecture looks like this: ┌─────────────────────────┐ │ CLAUDE WORK MODE │ │ Marketing Workspace │ └───────────┬─────────────┘ │ ┌───────────────┬───────┴────────┬───────────────┐ ↓ ↓ ↓ ↓ ┌──────────┐ ┌──────────┐ ┌──────────┐ ┌──────────────┐ │ Content │ │ Visual │ │ Lead │ │ AI SEO │ │ Creation │ │ Assets │ │ Discovery│ │ Optimization │ └─────┬────┘ └─────┬────┘ └─────┬────┘ └──────┬───────┘ │ │ │ │ ↓ ↓ ↓ ↓ ┌────────────────────────────────────────────────────────┐ │ Distribution & Scheduling MCP │ │ (Blot/Social Platforms) │ └────────────────────────────────────────────────────────┘ ### The Individualized Skill Stack Your system should reflect your unique needs. There's no one-size-fits-all configuration. A coach might prioritize: - Content repurposing for podcast episodes - Lead generation for coaching prospects - Newsletter production - AI SEO for personal brand visibility An e-commerce marketer might focus on: - Product listing infographics - UGC video ads - Competitor product tracking - Amazon/Shopify optimization A B2B company might emphasize: - LinkedIn content generation - Account-based lead discovery - Competitive intelligence dashboards - Email outreach at scale The components are modular. You assemble the stack that serves your specific goals. ### The Continuous Improvement Loop Your system improves over time through several mechanisms: - Skills accumulate knowledge from your corrections and feedback - Memory files store lessons about what content performed well - Analytics dashboards reveal what competitors are doing - New skills can be created on demand from reference materials - The AI SEO process identifies emerging keyword opportunities This isn't a static setup. It's a living system that gets smarter and more effective the more you use it. ### Getting Started Today Here's your action plan: 1. Download Claude Desktop 2. Create a dedicated marketing folder on your computer 3. Set up the workspace structure with subfolders 4. Run the onboarding questionnaire 5. Install necessary MCP connectors (Firecrawl, Supabase, Zapier, Semrush, creative studio, scheduler) 6. Create custom skills for your specific needs 7. Build your content repurposing pipeline 8. Configure your first dashboard Start with one component. Master it. Then add the next. --- ## Conclusion: The Shift from Chatbot to Operating System Here's what you now understand that most marketers don't: Claude is not a chatbot. It's an operating system for marketing operations. The difference is everything. A chatbot gives you answers. An operating system produces results. It creates your content, generates your visual assets, finds your leads, tracks your competitors, optimizes your search presence, and distributes your work across every platform you use. The key insight is integration. The power doesn't come from any single component,it comes from connecting them all through MCP connectors and skills. Each piece amplifies the others. Content repurposing feeds your distribution channels. Lead generation feeds your outreach. Competitor tracking feeds your content strategy. AI SEO feeds your visibility. The second key insight is accessibility. None of this requires programming skills. The entire system operates through natural language commands and pre-built connectors. If you can describe what you want, you can direct the system to produce it. The third insight is compounding. Every piece of content you create improves your skills. Every campaign you run informs your memory file. Every dashboard you build reveals new opportunities. The system doesn't just execute tasks,it learns and improves. Start small. Set up your workspace folder. Create one skill. Connect one tool. Generate one piece of content. Then expand. The complete system we've covered in this course isn't built in a day, but every component you add makes everything else more powerful. The marketers who master these systems now will have a significant advantage as AI becomes more integrated into every aspect of the industry. You now have the complete framework. The tools are available. The path is clear. Go build your marketing operating system.

Frequently Asked Questions

This FAQ is meant to answer the real questions that show up once you try to use Claude as a practical marketing engine, not just a chat toy. It walks through setup, skills, connectors, automation, and AI SEO in a sequence that makes sense whether you're brand-new to AI or already deep into tools. Use it as a reference: scan the questions, jump to what you need, then come back when you hit the next constraint in your system.

What is the difference between standard Claude and Claude Co-work mode?

Key idea:
Standard Claude is like talking to a smart strategist; Co-work mode is like having that strategist sit at your desk and work inside your files.

Standard chat mode:
You type prompts, Claude replies. Everything stays in the chat. Great for brainstorming ideas, rough drafts, quick explanations, and strategy questions. But once you close the window, the work is scattered and you're copy-pasting into your tools by hand.

Co-work mode:
Claude gets access to a specific folder on your computer. It can read documents, create new ones, rename files, and organize entire campaigns into subfolders. It can also run skills, call connectors, and operate like a junior marketer who actually ships work into your systems.

The shift is this: chat mode gives you answers; Co-work mode gives you assets and systems
,blog posts, email sequences, social calendars, CSVs, and dashboards that live on your machine.

What is a workspace folder and how do I create one?

Key idea:
A workspace folder is Claude's "office" on your computer where all marketing work happens in a structured way.

How it works:
You choose or create a folder (for example, "Claude Marketing") and grant Co-work mode access. Inside that folder, Claude stores:
- CLAUDE.md (instructions)
- MEMORY.md (context)
- Subfolders like /content, /social, /emails, /leads, /assets

How to set it up:
Open the desktop app → switch to Co-work → select "Choose folder" → create or pick your marketing folder → allow ongoing access. From that point on, when you say "Write a blog post" or "Export these leads," Claude can save the files straight into the right subfolder.

The result: everything your AI produces lives in one clean, browsable structure
instead of getting buried in random chats.

What are the CLAUDE.md and MEMORY.md files, and why are they essential?

Key idea:
These two markdown files are your operating system: one sets the rules, the other stores long-term context.

CLAUDE.md:
This is your instruction manual. It explains who you are, what your business does, your brand voice, and how you want Claude to behave in this workspace. Claude reads it on every message, so it should stay short and sharp: core positioning, voice principles, non-negotiable rules, and file conventions.

MEMORY.md:
Used well, CLAUDE.md defines how Claude should think; MEMORY.md defines what it should remember
so you don't repeat the same backstory in every session.

What are "skills" in Claude Co-work and how do they function?

Key idea:
A skill is a reusable playbook that tells Claude exactly how to perform a recurring task, step by step.

What's inside a skill:
Each skill is a markdown file that explains:
- When to use it
- What the task is (e.g., "write a newsletter")
- Detailed steps and formatting rules
- Where to save the finished file

Once a skill exists, you trigger it with a slash (/) command or a natural language request like "Write a blog post about X," and Claude follows the playbook instead of guessing from a short prompt.

The benefit: you write the instructions once, then reuse them forever
so every blog post, email, or carousel follows the same structure, tone, and quality bar without you rewriting a 400-word prompt every time.

How can I clone my own brand voice so AI-generated content sounds like me?

Key idea:
You don't "tell" Claude your voice; you let it study real examples and codify the patterns into skills.

Practical workflow:
1) Connect Gmail (or another email tool) via a connector.
2) Let Claude analyze 20-30 of your sent emails: greetings, sentence length, rhythm, phrases you repeat, how you sign off.
3) Do the same with blog posts, LinkedIn posts, or newsletters by pasting them in or adding files to the workspace.
4) Ask Claude to create separate voice skills: "email voice," "LinkedIn voice," "long-form essay voice," etc.

You then review and tweak those skills: "Use fewer adjectives," "Shorten sentences," "No hypey language." Claude updates the skill file so the corrections become permanent.

The result: Claude doesn't sound like 'an AI'; it sounds like a disciplined version of you
that stays consistent across platforms.

What is a content repurposing skill and how does it work?

Key idea:
A content repurposing skill is a conductor that tells several other skills what to do with one source asset.

How it behaves:
You feed it a source, like a YouTube URL or a long article. The repurposing orchestrator skill then:
- Scrapes or reads the content (often via Firecrawl)
- Sends it to your blog post skill
- Sends it to your newsletter skill
- Sends it to your LinkedIn skill (and other platform skills you define)
- Saves each output to the correct subfolder

Instead of manually asking, "Now write a newsletter, now a thread, now a carousel," you fire one command and the system fans out the work.

Effect:
One talk, podcast, or video turns into a small content ecosystem in minutes.
This shifts you from "What should I post today?" to "Which assets should I repurpose this week?"

How do I create skills for different content platforms?

Key idea:
The process is the same for every platform: show Claude examples, clarify the rules, then turn that into a skill file.

Step-by-step:
1) Collect 3-5 strong examples for that platform (yours or ones you want to emulate).
2) Paste them into Claude or add them as files in the workspace.
3) Answer Claude's questions about audience, tone, length, CTAs, and formatting.
4) Ask: "Turn this into a reusable skill for [platform]."
5) Review the generated markdown skill, adjust anything off, and save it in the skills folder.

The result is a library: "/linkedin_post", "/newsletter", "/short_form_script", "/carousel_copy"
each with its own tone, structure, and non-negotiables. You can refine any skill over time by editing the file or asking Claude to update it based on your feedback.

What is an MCP connector and why do I need it?

Key idea:
An MCP connector is a standardized bridge that lets Claude control other tools directly from your workspace.

What it enables:
By adding connectors, Claude can:
- Scrape websites (Firecrawl)
- Pull SEO data (Semrush)
- Access lead databases (Clay)
- Generate images/videos (Higsfield)
- Manage social posting (Blot)
- Talk to thousands of apps via automation platforms

Without connectors, Claude is stuck with text and your local files. With them, it becomes an operator that can reach out, pull data in, and trigger actions elsewhere.

Setup pattern:
You usually paste a server URL or choose from a library, drop in an API key, and set permissions (always allow vs ask each time). Once done, you can say things like: "Find 100 leads matching this profile" or "Generate a product infographic from this URL," and Claude handles the tool calls behind the scenes.

What is Firecrawl and when would I use it?

Key idea:
Firecrawl is Claude's scraper sidekick for pages it can't access directly, especially social platforms and some websites.

Use cases:
With Firecrawl connected, Claude can:
- Pull full transcripts and metadata from YouTube videos
- Extract content from Instagram posts, X threads, or long landing pages
- Grab product details from e-commerce URLs for infographics or comparison sheets
- Collect competitor blog posts for analysis and repurposing

Firecrawl converts messy web pages into structured markdown that Claude can read, summarize, and feed into skills.

Why it matters:
Most of your useful marketing data lives on sites Claude can't read natively. Firecrawl plugs that hole so your workflows can start from real competitor content, your own channels, or any URL,not just what you manually paste in
.

How do I generate images and videos directly inside Claude?

Key idea:
Claude doesn't create images or videos by itself; it orchestrates creative tools through connectors like Higsfield.

Practical flow:
1) You install the Higsfield MCP connector.
2) Inside Claude, you describe what you want: "Create an explainer infographic about this feature," or "Generate a 20-second UGC-style ad from this product URL."
3) Claude handles prompt crafting, calls the right Higsfield model, and returns generated assets into your workspace or chat.

Think of it as: Claude = creative director, Higsfield = studio
Claude knows your brand voice, your offer, and the context of the campaign. Higsfield has the heavy visual models. Together they produce consistent images, infographics, and videos that you can schedule directly through your social connectors or drop into sales pages and ads.

What types of visual content can I create with the Higsfield skills?

Key idea:
Higsfield skills give you repeatable templates for the most useful marketing visuals: infographics, product graphics, and UGC-style videos.

Core skills include:
- Explainer infographics: Claude researches a topic, then designs a clear visual explanation in a specific style and aspect ratio.
- Product infographics: Claude scrapes a product page, picks the most persuasive features, and lays them out in a graphic for Amazon, Shopify, or social.
- Product-to-ad UGC videos: You provide product images and target audience. Claude designs a synthetic actor, writes a script, and Higsfield renders a short video that feels like organic creator content.

You can also say, "Use this previous graphic as a style reference and create a new skill around it." Over time, you build your own visual system library
,consistent colors, layouts, and moods, all callable from a short instruction.

Why should I use Clay rather than having Claude scrape leads directly?

Key idea:
Claude is great at language, not at hunting down accurate, up-to-date contact data.

Where Clay fits:
Clay aggregates data from more than a hundred sources: email finders, company databases, social signals, funding trackers, tech stack tools. It also uses waterfall enrichment (checking multiple sources until it finds a verified match).

If you ask Claude on its own, "Find me the email of the VP Marketing at X company," it will often guess or hallucinate. With Clay connected, Claude instead says, "Ask Clay," and Clay returns:
- Verified emails and phones
- Titles and departments
- Company size, funding, and revenue ranges
- Tech stack and intent signals

The difference is simple: Claude writes the outreach; Clay supplies the right people and their real contact info
so you're not blasting guesses into the void.

How does the automated outreach workflow function?

Key idea:
Think of it as a pipeline: define who you want, pull the leads, then send personalized emails in your voice.

Typical flow:
1) Claude interviews you about your ideal customer (industry, size, role, geography, tech stack, problems).
2) It passes those criteria to Clay, which finds matching companies and decision-makers.
3) Claude enriches the results: funding rounds, traffic, tools they use, and so on.
4) Using your email voice skill, Claude writes tailored messages referencing specifics about each company and role.
5) Drafts are created in Gmail (or saved as files) for your review, not sent blindly.
6) Claude saves a CSV of the lead list plus email copy to your /leads folder.

The key is control: you stay in the loop for approvals, while Claude and Clay do the grunt work
of research, enrichment, and first-draft writing.

What is a "live artifact" and how can it help with marketing?

Key idea:
A live artifact is an interactive dashboard inside Claude that updates from your data sources instead of living in a static spreadsheet.

What it can show:
Examples include:
- A YouTube competitor tracker with recent videos, performance, and outlier scores
- A personal content tracker for your own posts across platforms
- A keyword opportunity board for AI SEO
- A lead pipeline view connected to your database

You can filter, sort, click buttons, and add notes directly inside the artifact. It's not "just text"; it behaves more like a lightweight app.

Behind the scenes, scheduled tasks collect data (often via Firecrawl) and store it in a database like Supabase. The artifact pulls from that data and renders it in a visual interface.

The benefit: you get a living control panel for your marketing system
without jumping between ten tools or building complex BI dashboards.

What is Supabase and why is it needed for tracking marketing data?

Key idea:
Supabase is the database layer that lets Claude store and query structured data over time, not just for a single session.

Why it matters:
Anything recurring,competitor tracking, content performance, keyword monitoring,needs history. Claude alone can generate a snapshot, but it doesn't persist data automatically. Supabase gives you tables, rows, and queries that Claude can manage for you.

Example: a competitor YouTube tracker.
- A daily task scrapes competitor channels via Firecrawl.
- Claude cleans and structures the data (title, views, posting date, outlier score).
- The data is saved to Supabase tables.
- A live artifact reads from Supabase and shows trends over weeks or months.

You don't need to be a database engineer. Claude can design the schema, write the queries, and keep it running
while you focus on decisions: which topics to copy, which to avoid, and where the gaps are.

How do scheduled tasks work in Claude Co-work?

Key idea:
Scheduled tasks are calendar-based automations that keep your systems updated without you logging in.

How they behave:
You define a schedule ("every morning at 7," "every Monday," "on the first of each month") and what should happen. For example:
- Scrape competitor YouTube channels → store in Supabase → refresh the dashboard
- Pull AI SEO data for a keyword set → update a content priority board
- Export new leads from Clay → append to a /leads CSV and notify you

At the scheduled time, Claude runs the workflow, calls the connectors it needs, and updates the relevant files or artifacts.

Think of it like hiring a reliable assistant who runs the same checklist at the same time, every time. Your dashboards stay current, and your decisions stay data-driven
without you pressing manual buttons.