AI DUNN Right Weekly - Issue #4

Practical AI insights for business growth

Hey AI Innovators! 👋

Welcome back to AI DUNN Right Weekly!

This week Apple dropped its M5 chip with 4x AI performance, OpenAI hit 800 million weekly users, and NVIDIA launched a desktop AI supercomputer for under $4K.

The gap between those leveraging AI and those ignoring it just became a canyon.

Here's what matters for your business:

• Apple M5 chip delivers 4x AI performance with on-device processing

• OpenAI reaches 800M weekly users and launches Apps SDK platform

• Oracle rebuilds its entire database architecture around AI

• NVIDIA's DGX Spark brings supercomputing power to your desk

• Google adds collapsible ads as AI chatbots threaten search dominance

Read time: 5 minutes

---

🚀 This Week's Game Changer

Apple M5: The AI Chip That Changes Everything

What happened: On October 15, Apple announced the M5 chip, and the numbers are staggering. Over 4x the GPU compute performance for AI compared to M4.

Up to 3.5x faster AI workflows. 45% better graphics performance. The new chip powers the 14-inch MacBook Pro ($1,599), iPad Pro ($999), and Vision Pro ($3,499), with each GPU core now having a dedicated Neural Accelerator.

Why it matters: Professional creative work just got democratized. AI tasks that took minutes now take seconds. Running large language models locally? Drawing AI-generated images? Video masking with AI?

All happening right on your device without sending data to the cloud. Students can now run models that required data center access last year. Small businesses can process sensitive data with AI without privacy concerns.

The bigger picture: This isn't an incremental update. It's Apple betting the farm on on-device AI. The M5 isn't just faster computing. It's computing that respects your privacy while delivering enterprise-grade AI performance. Apple just made AI accessible to anyone with a laptop.

Business impact:

• For Creators: Run professional AI workflows without expensive subscriptions

• For Startups: Process sensitive data locally without cloud privacy risks

• For Teams: Deploy AI tools that don't require constant internet connectivity

---

🛠️ Tools That Actually Work

OpenAI Apps SDK Transforms ChatGPT Into a Platform

ChatGPT hit 800 million weekly active users (up 100 million in one month) and unveiled the Apps SDK at DevDay 2025. Developers can now build and sell apps directly inside ChatGPT.

Companies like Canva and Coursera are streaming full functionality without users ever leaving the chat interface.

Translation: OpenAI is executing the classic platform playbook.

Build massive distribution first, then let developers monetize on top of your infrastructure. Welcome to the operating system of the AI era.

Oracle AI Database 26ai: Enterprise AI Gets Real

Oracle launched AI Database 26ai, integrating AI across the entire data stack from vector search to database management to application development.

The new Oracle Autonomous AI Lakehouse supports Apache Iceberg and works across all four major clouds: OCI, AWS, Azure, and Google Cloud.

It's interoperable with Databricks and Snowflake. Most companies have data scattered across multiple systems.

Oracle just made it possible to run AI agents that dynamically pull from all of it, combining private company data with public information.

NVIDIA DGX Spark: Your Personal AI Supercomputer

NVIDIA is launching the DGX Spark for $3,999, a desktop machine that handles advanced AI models without needing a data center.

Features include NVIDIA's GB10 Grace Blackwell Superchip, 128GB unified memory, up to 4TB storage, and a petaflop of AI performance.

You can now run models with up to 200 billion parameters from your desk. What required millions in cloud compute last year now costs less than a used car.

Major PC makers including Dell, HP, Lenovo, and ASUS are releasing their own versions. Available October 15.

Google Search Gets Collapsible Ads

On October 14, Google rolled out the ability to collapse sponsored search results.

Users can tap "Hide sponsored results" to remove ads from the top of search results. The "Sponsored" label remains visible, but the ads disappear.

Google also introduced AI-powered features in Search and Discover, including trending topic previews and sports news feeds.

The subtext: Google knows ChatGPT and Perplexity are eating their lunch. This is a strategic retreat designed to look like a gift.

---

💡 This Week's Prompt

The "Act As Expert" Framework

Stop writing generic prompts. Start with role assignment. Assigning expertise changes everything about the output.

Instead of: "Write a marketing email"

Try this: "You are a direct response copywriter specializing in SaaS products with 15 years of experience.

Write a 150-word email for mid-market B2B companies promoting our new project management tool. The email should focus on time savings and ROI. Use a conversational but professional tone."

Why it works:

• AI performs dramatically better when you provide specific context

• Role assignment activates relevant training data

• Specificity produces actionable results, not generic advice

  
  

Pro tip: Include relevant details like your industry, target audience, business size, and desired outcome. The first prompt gets generic advice. The second gets a strategy you can implement Monday morning.

---

💼 Business Opportunity

The AI Implementation Gap: Your Trillion-Dollar Opening

Here's the insight everyone's missing: Most businesses know they need AI. Almost none know how to implement it. This is your opening.

The opportunity isn't building the next ChatGPT. It's being the person who helps the local accounting firm use ChatGPT to cut invoice processing time by 60%.

It's showing the regional hospital how to use AI to summarize patient notes. It's teaching the real estate agency how to generate property descriptions in seconds instead of hours.

How To Capture This Market:

Start local. Identify 10 businesses in your area that could benefit from AI but clearly aren't using it. Offer a free 90-minute workshop. Show them three specific AI implementations they could deploy this week.

Build a toolkit. Create a library of prompts, workflows, and SOPs for common business functions. Customer service templates. Marketing copy generators. Data analysis frameworks. Make it plug-and-play.

Charge for implementation, not education. Don't bill for teaching them about AI. Bill for setting up their systems, training their team, and ensuring adoption. Implementation is worth 10x more than information.

The numbers: AI consultants are charging $2,000 to $5,000 per hour for implementation work. Not because they're AI experts. Because they can translate technical concepts into business outcomes. You don't need a PhD. You need the ability to explain things clearly and help people take action.

---

📌 Quick Bites

• AI spending projected to reach $375 billion in 2025, $500 billion by 2026

• Anthropic's Claude deployed to 470,000 Deloitte employees across 150 countries

• IBM partners with Anthropic to integrate Claude across enterprise software

• European Commission allocates €1B to accelerate AI adoption

• Character.AI faces lawsuit after chatbot allegedly contributed to teen's suicide

• AI-generated "actress" Tilly Norwood sparks Hollywood outrage from SAG-AFTRA

• Generative AI in healthcare market projected to grow from $1.1B (2024) to $14.2B (2034)

---

🔮 Strategic Takeaways

• Hardware Democratization: Apple's M5 proves on-device AI is ready for mainstream adoption, eliminating cloud dependency and privacy concerns.

• Platform Consolidation: OpenAI's Apps SDK strategy positions ChatGPT as an AI operating system, not just a chatbot, fundamentally changing how we think about software distribution.

• Enterprise Acceleration: Oracle and IBM partnerships signal that AI has moved from pilot programs to core business infrastructure at unprecedented scale.

• Economic Dependency Risk: Without AI-driven investment, some economists warn the US may already be in recession, raising concerns about market sustainability.

---

🎯 Action Items for This Week

For Corporate Teams:

• Evaluate M5-powered devices for teams handling sensitive data that needs local AI processing

• Test OpenAI's Apps SDK for internal workflow automation opportunities

• Review Oracle AI Database 26ai if your data is scattered across multiple cloud systems

For Small Businesses:

• Consider NVIDIA DGX Spark if you're running AI models that currently require expensive cloud services

• Explore Google's new AI search features to understand how your SEO strategy needs to evolve

• Start documenting which business processes could be improved with AI implementation

For Entrepreneurs:

• Research the AI implementation gap in your local market and identify underserved businesses

• Create 3 sample workflows showing how AI solves real problems in your industry

• Build a personal toolkit of prompts and processes you can deploy quickly for clients

---

🔮 Looking Ahead

Next week's AI DUNN Right Weekly will cover:

• M5 performance benchmarks: Real-world testing from early adopters

• OpenAI Apps SDK deep dive: Building your first ChatGPT app

• AI implementation case studies: Three small businesses that got it right

• The rise of AI consultants: Skills you need to charge premium rates

  
  

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 #4!

This week proved AI hardware is catching up to AI software. On-device processing eliminates the cloud bottleneck. Platform consolidation is happening faster than anyone predicted. And the businesses investing in AI implementation now will dominate their markets by 2026.

The gap isn't closing. It's widening. Time to pick a side.

Stay innovative,

Jackie @ AI DUNN Right

P.S. If you're still sending data to the cloud for AI processing when you could do it locally on an M5 device, you're leaving money and privacy on the table. This changes everything.