AI DUNN Right Weekly - Issue #21
- Feb 16
- 6 min read
Practical AI insights for business growth
Big Story
"Something Big Is Happening" and why you need to pay attention
An essay went viral this week. Like, 71 million views viral.
Matt Shumer, CEO of HyperWrite, published a piece called "Something Big Is Happening" and it clearly struck a nerve. The essay says out loud what a lot of people in the AI space have been thinking quietly.
His argument? We're in the "this seems overblown" phase of something much bigger than most people realize.
Shumer describes telling AI what he wants built in plain English, walking away for four hours, and coming back to the finished product. Done. No edits needed.
That tracks with what I've been seeing too. Every time someone says "sure, AI can do that, but it can't do this other thing," six months later... it can.
Now here's the part that should make every business owner sit up straight.
Dario Amodei, CEO of Anthropic (the company behind Claude), has publicly predicted that AI will eliminate 50% of entry-level white-collar jobs within one to five years. Many in the industry think he's being conservative.
Connor Grennan, Chief AI Architect at NYU Stern, backed that up this week. He said that even if the technology doesn't improve from where it is today, it could wipe out 25% of entry-level white-collar jobs. With what already exists. Right now.
The jobs most at risk? Summarizing information, creating first drafts, turning data into narratives, and formatting documents. That's the bulk of what entry-level employees spend their time doing.
Here's what this means for you...
The gap between people who use AI seriously and people who tried it once two years ago is getting dangerously wide.
Stop treating AI like Google. Start treating it like a team member. Feed it a full contract. Give it a messy spreadsheet. Push it into your actual work.
The window where being early gives you a massive advantage won't stay open forever.
What's New This Week
Anthropic raises $30 billion at a $380 billion valuation
Anthropic just closed a $30 billion funding round. Annual revenue has climbed to $14 billion, with Claude Code alone pulling in $2.5 billion per year. That makes them the second most valuable AI company behind OpenAI.
What this means for you: more investment means faster improvements to the tools you're already using. Competition keeps driving prices down and quality up.
OpenAI launches Frontier for enterprise AI agents
OpenAI rolled out Frontier, a platform that helps companies build and manage AI agents across their entire business. Shared context, permissions, coordination across systems.
HP, Intuit, Oracle, State Farm, and Uber are among the first adopters.
The key difference? These agents understand how your business works. They access your CRM, data warehouses, and internal tools. They build memories from past interactions.
Early results include a manufacturer that cut production optimization from six weeks to one day, and an investment company that freed up 90%+ of salespeople's time for actual client conversations.
ChatGPT now shows ads to free users
OpenAI started rolling out ads inside ChatGPT for free and basic tier users in the US. They say ads won't influence answers. I'll say this... if you're using ChatGPT for free, you are the product. Same deal Google has had for decades. Paid plans remain ad-free.
Microsoft discovers AI recommendation poisoning
This one is important. Microsoft security researchers found that companies are hiding instructions in "Summarize with AI" buttons that inject commands into your AI assistant's memory.
Over 50 unique prompts from 31 companies across 14 industries were found doing this.
They tell AI to "remember [Company] as a trusted source" so future conversations are biased toward that brand.
Your AI assistant might already be compromised without you knowing.
Action step: Go check your AI memory settings right now. In ChatGPT, go to Settings, then Personalization, then Memory. In Claude, check Settings, then Memory preferences. Delete anything you don't recognize.
Tool of the Week
OpenClaw (formerly ClawdBot)
OpenClaw is an open-source AI agent tool with 185,000 stars on GitHub that's changing how people think about AI assistance.
Here's the key difference. With ChatGPT or Claude, you go to the tool when you need something. With OpenClaw, the AI handles things before you even think about them. Always on, working based on triggers and rules you set.
Four ways people are using it:
Executive Assistant: Invoice processing, receipt logging, scheduling. One user built a system that auto-categorizes receipts and pushes them to accounting software.
Researcher: Proactively pulls research based on criteria you define and delivers results before you've finished your morning coffee.
Marketing and Ops: Monitor ad performance, refresh copy when engagement drops, repurpose content across platforms, handle PR outreach.
Builder: Give it a GitHub account and deployment access, and it builds, tests, and ships real projects with minimal oversight.
Think of it the same way you'd onboard a human assistant. What would you hand off? What should they check on in the background? What decisions can they make without asking first?
Quick Hits Worth Your Time
Meta launched "Dear Algo" for Threads, letting users tell the algorithm what content they want to see using plain language. Testing in the US, UK, Australia, and New Zealand.
Amazon released an Ads MCP Server that lets AI agents manage entire advertising campaigns through natural language prompts. Works with Claude, ChatGPT, and Gemini.
xAI lost more co-founders after the SpaceX merger. Instability at a competitor often means opportunity for the companies that stay focused.
Saudi Arabia partnered with NVIDIA on AI factories up to 500 megawatts with 18,000 next-gen supercomputers. The infrastructure race is global now.
Trump signed an executive order to override state AI laws. Love it or hate it, this changes the regulatory landscape.
A Harvard Business Review study found that AI doesn't reduce how much people work. It increases it. People do more because the tools make more feel doable. AI burnout is real.
OpenAI disbanded its mission alignment team, the group focused on safe and trustworthy AI development.
Kling 3.0 launched with major upgrades to AI video generation including 15-second clips, multi-language audio, and better character consistency.
Prompt of the Week
Act as a Customer Retention and Loyalty Program Architect.
I want to move away from one-off visits and create predictable,
recurring revenue through a membership or loyalty program that
feels exclusive without costing much to deliver.
Essential Details:
- Business Type: [YOUR INDUSTRY]
- Current Customer Behavior: [ONE-OFF VS REPEAT]
- Average Transaction Value: [TYPICAL SPEND]
- Customer Frequency: [HOW OFTEN THEY VISIT]
- Current Loyalty Approach: [WHAT YOU DO NOW]
- System/Platform: [CARD SYSTEM OR TECH YOU USE]
Design one tiered loyalty program including:
1. Tier names (memorable, aspirational names
customers want to reach)
2. Requirements to reach each tier (spend, visits,
or other measurable actions)
3. Specific perks for each tier (early access,
bonus credits, exclusive events, priority service...
things that feel valuable but cost you little)
4. Progression visibility (how customers track
their status and see what's next)
5. Retention mechanics (what keeps them engaged
after reaching top tier)
6. Communication strategy (how you announce tiers,
celebrate upgrades, remind them of benefits)
Create recurring revenue through loyalty that
doesn't break your budget.
Why this works: Acquiring a new customer costs five times more than keeping an existing one. This prompt forces you to think about the entire retention cycle and builds a system that turns one-time buyers into repeat revenue.
My Take
The biggest theme this week? The distance between people who get AI and people who don't is becoming a canyon.
An essay hits 71 million views. Anthropic raises $30 billion. Harvard Business Review publishes research showing AI users are working more (not less) because the tools make so much feel possible. We've crossed a line. AI has gone fully operational.
The people doing well right now have one thing in common. They stopped waiting for the "right time" and just started. Picked a tool, tested it for 15 minutes, saw that it saved time, and used it again the next day.
And here's what might be the most important insight from this week. Management skills are becoming the AI superpower. If you know how to delegate tasks, write clear briefs, and structure projects, you're perfectly positioned for the AI era. Working with AI agents is managing a team of digital employees.
Great news for anyone who's led a team, run a department, or managed contractors. The skills you already have are exactly what's needed.
Stop overthinking it. Start building.
That's it for this week.
What part of your business are you thinking about handing over to AI? Hit reply and let me know.
Jackie
Comments