AI DUNN Right Weekly - Issue #5
- Jacqueline Dunn
- Oct 26
- 7 min read
Practical AI insights for business growth
Hey AI Innovators! 👋
Welcome back to AI DUNN Right Weekly! This week OpenAI entered the music generation game, synthetic influencers sparked ethical debates, and Anthropic launched Skills that actually make Claude useful for specific tasks. Here's what matters for your business:
• OpenAI building AI music composer to rival Suno and Udio
• Synthetic influencer marketplace raises questions about authenticity
• Claude Skills eliminate context window bloat with compartmentalized expertise
• China's $1,370 humanoid robot challenges U.S. pricing
• The subscription audit prompt that could save you thousands annually
Read time: 5 minutes
🚀 This Week's Game Changer
Claude Launches "Skills" - Compartmentalized AI Expertise
What happened: Anthropic released Skills in preview, repeatable, customizable instructions that Claude can follow in any chat. Skills are folders containing instructions, scripts, and resources that Claude only loads when relevant. Access them via Settings > Capabilities > Skills (might need a US VPN to activate).
The innovation: Previously, you'd cram everything into system prompts, brand guidelines, visual rules, compliance requirements, clogging your context window with irrelevant information.
Now, Claude scans available skills and loads only what's needed for each specific task.
Business impact:
For Content Teams: Build individual skills for different content types, combine tone of voice skill with newsletter format plus source material to generate first drafts
For Operations: Skills for expense reports, meeting notes, data analysis that work consistently without reprompting
For Agencies: Client-specific skills that maintain brand voice and requirements without manual copying
The bigger picture: Skills might replace MCPs (Model Context Protocols) in many cases since they're lighter weight and easier to manage. One entrepreneur turned a 2-3 hour monthly expense report process into 5 minutes by creating a custom skill. The first-time setup took 20 minutes, but now it's completely automated.
Why it matters: This closes the gap between "AI could theoretically help" and "AI just did my work in 10 minutes." Skills are currently in preview for Claude Pro, Max, Team, and Enterprise users, with a cookbook available on GitHub for advanced implementations.
🛠 AI Tool Spotlight
ChatGPT Atlas - OpenAI's AI-Powered Browser
What it does: OpenAI launched ChatGPT Atlas, an AI-powered browser designed to help you search, organize, and interact with the internet more intelligently. Instead of just answering questions, Atlas can navigate websites, pull information from multiple sources, and organize research automatically.
Key features:
Intelligent browsing that executes tasks across multiple websites
Automated research compilation from various sources
Natural language commands for complex web workflows
Built-in organization for gathered information
Integrates directly with ChatGPT's existing capabilities
Best use cases: Researchers compiling data from multiple sites, entrepreneurs conducting competitive analysis, teams gathering information for reports, anyone who loses hours manually navigating between tabs and copy-pasting information.
Real example: One user described asking Atlas to gather grocery items from a recipe video they were watching. Atlas identified the ingredients, checked availability across stores, and compiled a shopping list. All through natural language. What would take 20 minutes of manual work happened in 2 minutes with one command.
Getting started: Atlas is currently in testing with select ChatGPT users. The tool represents OpenAI's push beyond being "just a chatbot" into becoming infrastructure for AI-powered workflows.
Why it's a breakthrough: This isn't Chrome with AI features. This is AI that controls your browser. It opens tabs, fills forms, clicks buttons, and executes multi-step tasks while you watch. The technology that lets it complete assignments can also automate your expense reports, book complex travel itineraries, or research competitors. The question becomes: are we choosing between getting faster and getting smarter?
⚡ The 5-Minute AI Academy
The Subscription Audit That Could Save You Thousands
Most businesses hemorrhage money on software subscriptions they barely touch. If you have 30+ tools like most companies, you're likely spending £300-600 monthly (£3,600-7,200 yearly) on subscriptions you could replace with AI in under an hour each.
The Problem: Linktree just raised prices 60%. Typeform charges £30-45 monthly for forms. Calendly costs £10-15 for basic booking. Each one made sense when you signed up. Each one is "only" a small monthly charge. But multiply that by 30 tools and it adds up brutally, especially when bootstrapping.
The Solution: AI coding tools like Lovable, Cursor, or Claude Code can now rebuild these simple SaaS tools in 30 minutes to 2 hours. One entrepreneur built a Linktree replacement in 10 minutes, added polish in another 20 minutes, then immediately cancelled his subscription.
Here's the systematic approach:
Step One: Audit Your Subscriptions List every subscription your business pays for using bank statements and credit card bills.
Then use this prompt:
I have these active subscriptions in my business:
[paste your list with costs and what they do]
Analyse each one for:
1. Replaceability score (1-10, where 10 is "trivial to rebuild with AI coding tools")
2. Core functionality (what it actually does vs all the features you don't use)
3. Estimated time to build a basic replacement using vibe coding
4. Annual cost saved if replaced
Rank them by best ROI: highest cost savings + easiest to build first.
Step Two: Pick Your Targets Replace tools scoring 7+ on replaceability, costing £10+ monthly, where you use less than 30% of features. Leave alone: core business tools with years of data, deeply integrated systems, or tools your entire team relies on.
Step Three: Screenshot Everything Open the tool you want to replace. Capture the main admin interface, forms or templates you've created, the output view, and settings you actually care about. You want the AI to see exactly what you use.
Step Four: Build It Open your chosen AI coding tool. Attach screenshots and describe what you want, stripping out features you never touch. Be specific about the core functionality you need.
Step Five: Iterate Until It Works Your first version won't be perfect. Test it, find bugs, give specific feedback like "The buttons are too close together on mobile, add more spacing." This usually takes 30 minutes to 2 hours depending on complexity.
Expected results: Cancel subscriptions that cost £10-45 monthly each. Build custom tools that do exactly what you need without bloated features. Save thousands annually that can be reinvested into growth, hiring, or staying in your pocket.
Pro insight: If you solve a problem your industry has, you've potentially found a product. Clean up your replacement, add user authentication and a database, and you could sell it for £5/month to people facing the same subscription pain.
🧠 Quick Wins: 5 AI Tools Worth Investigating
Based on this week's newsletter coverage and emerging capabilities:
🎵 OpenAI Music Model - AI that generates original music from text or audio promptsUse case: Custom music for marketing content without licensing nightmares, expanding short clips, producing instrument layers
🤖 Doublespeed AI Synthetic Influencers - Controversial AI-generated personas for social platformsUse case: Testing ad narratives, seeding trends, boosting engagement metrics (ethically questionable, under investigation by Meta and TikTok)
💻 Lovable Cloud - No-code app building that eliminates backend complexityUse case: Replacing expensive SaaS subscriptions with custom-built tools in under an hour
📊 Shakespeare - Builds complete websites from prompts with decentralized architectureUse case: Quick landing pages, personal sites, client projects without traditional development time
🎤 Fish Audio - Generates expressive AI voices with emotion control and cloningUse case: Voiceovers, audiobooks, multilingual content without re-recording everything
📈 Business Intel: This Week's Market Movers
🎶 OpenAI Enters Music Generation The AI-music market is projected to hit $3.1 billion by 2028, growing 30% annually. Startups like Suno have already raised $125 million this year, while YouTube reports a 400% rise in AI-generated songs since January. OpenAI's entry with Juilliard partnership could force the entire music industry to adapt faster than during the streaming revolution.
🎭 Synthetic Influencer Economy Explodes Experts estimate over 15 million synthetic accounts operate across major social platforms, up 200% since 2023. With influencer marketing spend projected to exceed $24 billion in 2025, synthetic influencers could undercut human creators and distort engagement metrics. Regulators in the U.S. and EU are weighing disclosure rules as the line between organic and artificial influence blurs.
🧠 Wikipedia Traffic Declines 8% Wikipedia seeing year-over-year traffic drop after filtering bot traffic. The irony: while AI causes traffic decline, Wikipedia data is more valuable than ever, appearing in every training dataset. Without Wikipedia, what will future models train on? Their own hallucinations?
⚖️ Barrister Caught Using AI Generates Fake Cases Immigration barrister cited 12 legal authorities in tribunal, 10 didn't exist. Used ChatGPT-like software, failed to check accuracy, then tried to hide it. Follows Deloitte charging Australia $440,000 for an AI-generated report full of hallucinations. The real question: how many other professionals are doing this and haven't been caught yet?
🤖 China's $1,370 Humanoid Robot Noetix Robotics unveiled Bumi, a humanoid robot priced at just $1,370 (cost of an iPhone). Despite compact build, Bumi can walk, balance, and dance. Signals China's push to make humanoid robots mainstream while the U.S. battles high production costs.
📚 This Week's Curated Reading
Based on key developments from this week's AI newsletters:
• Skills vs MCPs: Claude's Skills implementation shows how AI works best when expertise is compartmentalized rather than crammed into system prompts, reducing context window bloat and improving consistency
• Vibe Coding as Gateway: Teaching fundamentals first kills interest, letting people build immediately then backing into learning actual code creates sustainable engagement, especially for young learners
• AI Parasitic Relationships: Wikipedia exemplifies the ultimate one-sided dynamic, AI companies train on their data, Google snippets steal their content, now AI overviews mean nobody clicks through, yet it remains crucial for maintaining clean training data
• Subscription Bloat Reality: Most businesses use less than 30% of features in their SaaS tools, paying for bloat that AI coding can now replace in under two hours per tool
• Synthetic Influence Ethics: The line between organic and artificial influence blurring faster than platforms can respond, raising questions about authenticity, disclosure, and the future of creator economies
🎯 Action Items for This Week
For Corporate Teams:
Enable Claude Skills and create your first custom skill for a repetitive task
Audit subscriptions using the provided prompt to identify replacement candidates
Test ChatGPT Atlas workflows for research and data compilation tasks
For Small Businesses:
Screenshot your three most expensive SaaS tools and assess replaceability
Try building a Linktree replacement using Lovable as a learning exercise
Review which features you actually use in your current subscription stack
For Entrepreneurs:
Calculate your monthly subscription burn using bank statements from last 3 months
Identify your top replacement candidate (high cost + low feature usage)
Spend one afternoon building your first subscription replacement
🔮 Looking Ahead
Next week's AI DUNN Right Weekly will cover:
Practical guide to creating your first Claude Skill from scratch
Step-by-step walkthrough: Building a Typeform replacement in under an hour
ChatGPT Atlas deep dive: Safe automation practices and security considerations
When to replace vs when to keep paying: The ROI calculation framework
Have questions or topic requests?
Reply to this email. I read every message and use your feedback to shape future issues.
That's a wrap for Issue #5!
This week proved AI is moving from "nice to have" to "essential cost-cutting tool." The businesses that systematically audit subscriptions and rebuild simple tools will save thousands while competitors keep bleeding monthly fees. The subscription economy just met its match.
Stay innovative,Jackie @ AI DUNN Right
P.S. - That subscription audit prompt? Run it this weekend. Pick one tool scoring 9 or 10 on replaceability and spend Sunday afternoon building your replacement.
You'll be shocked how easy it is, and even more shocked when you cancel that subscription Monday morning. The satisfaction is worth the two hours invested.
Comments