AI DUNN Right Weekly — Issue #28
- Apr 6
- 8 min read
Practical AI insights for business growth
Hey AI Innovators! 👋
This week, something happened that I honestly did not expect to see quite so soon. A guy in Los Angeles built a $1.8 billion business from his living room... with $20,000, two months, and a stack of AI tools. Sam Altman predicted it in 2024. This week, it became real.
Meanwhile, Jack Dorsey replaced his managers with AI, OpenAI raised more money than any company in history, and Anthropic's mysterious new model is already being quietly briefed to governments.
Big, big week.
Here's what you need to know:
The first solo billion-dollar company just happened... and it sells weight-loss drugs
Jack Dorsey says AI should replace your manager (and he's already doing it at Block)
OpenAI raised $122 billion... the largest fundraise in venture capital history
Anthropic's secret "Mythos" model is powerful enough that they're briefing governments before launch
Canva's new Magic Layers is a game-changer for anyone using AI-generated images in their business
Read time: 7 minutes
Big Story
The First Billion-Dollar One-Person Company Is Real
Back in 2024, Sam Altman made a bold prediction: AI would eventually make a solo billion-dollar company possible. At the time, it sounded like optimistic tech talk. Maybe a decade away.
Meet Matthew Gallagher.
In early 2025, Gallagher launched a startup called Medvi from his house in LA. Budget: $20,000. Team: him and a laptop. Business model: sell GLP-1 weight-loss drugs (think Ozempic and Wegovy) online, outsourcing the prescriptions and shipping to telehealth partners.
He did not hire a marketing team, a tech team, or a customer service department. He used AI for all of it.
ChatGPT, Claude, and Grok — for writing code and content
Midjourney and Runway — for ad creatives
ElevenLabs and custom AI agents — for customer service
Two months later, Medvi was live. A year and a half later, the company had brought in $401 million in its first full year of revenue. This year, Medvi is on pace to do $1.8 billion in sales. His only full-time hire is his brother.
The New York Times covered it this week, and I have been thinking about it ever since.
What gets me is that this is not some flashy AI product or cutting-edge tech startup. It is a company that helps people get weight-loss prescriptions online. The product is almost ordinary. What is extraordinary is how Gallagher built and operated it... by treating AI tools the way most businesses treat a full workforce.
My take on this? This story matters even if you never want to build a billion-dollar company. It is proof that the gap between a small team and a large operation has closed dramatically. The tools Gallagher used are available to you right now, today. The question is no longer "can I do this?" The question is "what would I build if the size of my team was never the limiting factor?"
What's New This Week
Jack Dorsey Says AI Should Be Your Manager
Jack Dorsey, the founder of Twitter and CEO of Block (the company behind Square and Cash App), made waves this week when he co-authored a post arguing that AI can replace middle management entirely.
Block recently cut over 4,000 employees... more than 40% of its total workforce. Dorsey framed this not as a response to weakness, but as a deliberate bet on the AI era. His thesis: managers exist to route information up and down a corporate chain of command. AI can now do that faster, using a live "world model" of the entire business.
Under the new structure, everyone at Block falls into one of three roles: builders, problem-owners who are accountable for specific outcomes, and player-coaches who develop talent. No more approval layers. No more status-update meetings just to move something forward.
Block is remote-first, which means almost every decision and conversation already lives as a digital record. That data is exactly what AI needs to step into the role managers currently play... or at least to try. Will it work? Nobody knows yet. But Dorsey is betting the company on it, and a lot of lean AI-first teams are watching very closely.
OpenAI Just Raised $122 Billion
OpenAI announced a historic $122 billion funding round this week at an $852 billion valuation. That is the largest single fundraise in venture capital history, full stop. Amazon, Nvidia, and SoftBank anchored the bulk of it.
But the number itself is almost secondary to what comes next.
OpenAI revealed it is merging ChatGPT, Codex (its coding tool), and its AI agent tools into one unified "superapp." Think of it as a single platform that writes your emails, builds your website, manages your calendar, runs tasks on your behalf, and executes code... all from one place. An IPO is also reportedly on the table for later in 2026.
Here is the stat that really tells the bigger story: OpenAI's revenue is now hitting $2 billion per month, and enterprise customers account for more than 40% of that. Businesses are paying for this at scale. That tells you where this whole industry is heading.
Anthropic's "Mythos" Model Is Already Briefing Governments
A lot happened with Anthropic this week, and almost none of it was planned.
First, details about a new model called "Claude Mythos" leaked online. Mythos sits above the current Opus line in capability, with particular strength in reasoning, coding, and cybersecurity. It is apparently so advanced in the security space that Anthropic has been quietly briefing government officials about its potential risks... before the model is even released to the public.
Then Anthropic accidentally published the source code for Claude Code (their developer tool) to a public registry. Over 500,000 lines of code became briefly visible, including unreleased features and an AI terminal pet called "BUDDY" complete with species, rarity tiers, and personality stats including CHAOS and SNARK. The internet had a very good time with that one.
For everyday users, neither leak affects how Claude works right now. But what they signal is significant: Anthropic is building capabilities powerful enough to warrant government involvement before launch. That is a different tier of AI than what most of us are using today.
Tool of the Week
Canva Magic Layers
If you have ever used AI to generate an image and then wanted to change just one small thing... you know the pain. You have to regenerate the whole image, tweak the prompt, hope for the best, and repeat. It is one of the most frustrating parts of using AI for visual content.
Canva just fixed that with Magic Layers.
Upload any flat image (including AI-generated ones) and Magic Layers automatically separates it into individual editable layers. Background, text, objects, and logos all become distinct pieces you can move, resize, swap, or delete independently. No more starting from scratch to fix one detail.
Here is how to use it:
Go to the Canva homepage, select Magic Layers, and upload your image
Wait 30 to 60 seconds while Canva processes and separates the layers
Click any element to edit it directly: fix a typo, swap a product photo, change a date, adjust the background
Export your updated design when done
This is for anyone creating visual content for their business: social graphics, ads, presentations, website images... anything where one small change used to mean starting over. It is available on Canva right now. Go try it with one of your existing designs this week.
Quick Hits Worth Your Time
OpenAI shut down Sora, its standalone AI video app, and in the process canceled a planned $1 billion partnership with Disney. Sora was costing an estimated $15 million per day to run against just $2.1 million in lifetime revenue. The video technology is not gone... it is being folded into OpenAI's broader superapp instead.
Slack got 30 new AI features courtesy of Salesforce. Slackbot can now create reusable "skills" (repeatable tasks that run automatically across your workflows), connect to outside apps, transcribe and summarize meetings, and suggest next steps based on your calendar and recent activity. If you use Slack for work, this upgrade is worth a look this week.
OpenAI acquired TBPN, a daily live tech talk show that pulls about 70,000 viewers per episode and regularly hosts top tech CEOs. The deal was reportedly worth hundreds of millions. After a rough stretch of headlines, having a media platform they control is clearly part of OpenAI's plan to shape the narrative from a friendlier stage.
Google released Gemma 4, a family of open-source AI models under a fully permissive Apache 2.0 license. In plain terms: developers can now use, modify, and commercially deploy these models with zero legal restrictions. The smallest versions run entirely on a phone with no internet required. A significant move to win back developers from Chinese open-source rivals.
An Anthropic study of 81,000 people found something worth pausing on: AI is genuinely making people more productive... but most of them are actually working more hours, not fewer. When AI speeds you up, you fill the recovered time with more work. The breathing room disappears before you ever get to enjoy it. If that resonates, it might be worth being intentional about protecting some of that time before it gets swallowed.
Prompt of the Week
This week's big story got me thinking. Matthew Gallagher did not randomly throw AI tools at Medvi and hope for the best. He figured out exactly which business functions AI could own, and then he built around that. You can do the same kind of audit for your own work, right now.
I want to do an honest audit of my business to find
where AI could save me the most time or money.
Here is what I do:
[Describe your business or role in 2-3 sentences]
My team size is: [number of people, or "just me"]
The tasks I spend the most time on each week are:
1. [Task 1]
2. [Task 2]
3. [Task 3]
4. [Task 4]
5. [Task 5]
Please do the following:
- Review each task and tell me whether AI can currently
handle it fully, partially, or not yet
- For tasks AI can handle fully or partially, suggest a
specific tool or approach I can try
- Identify the 2-3 changes that would save me the most
time if I made them this month
- Flag anything I am still doing manually that most
people now automateWhy this works: Most people try AI tools at random and give up when the results feel generic. This prompt grounds everything in your actual day-to-day work. It gives you a personalized map instead of a blanket suggestion list, and it forces an honest look at where your time is actually going. That is where the real wins are hiding.
My Take
I keep coming back to Matthew Gallagher this week. Not because the revenue number is wild (though it is), but because of what the story means for the rest of us.
For a long time, the "AI will change everything" conversation felt abstract to me. Like it was happening somewhere else, for someone with more resources or more technical knowledge. This week, that shifted.
Gallagher did not invent anything. He did not write a line of AI code himself. He used the same tools sitting in front of you right now: ChatGPT, Claude, Midjourney, ElevenLabs. He just used them intentionally, systematically, and boldly. He treated them like a team, not like a toy.
That is the invitation this week is extending to all of us. Not "should I try AI?" but "what would I build if I treated these tools the way Gallagher did?"
What would I stop doing manually?
What could I attempt if the size of my team was never the ceiling?
You do not have to aim for a billion dollars. But I do think it is worth letting yourself dream a little bigger this week. The tools are ready. The question is whether we are.
That is what keeps me coming back to this every Monday. The ceiling keeps moving. Our job is to stay curious, stay practical, and keep showing up.
See you next week.
Jackie
Comments