When AI Becomes the Manager and Humans Become the API
Why your AI strategy is failing and how a bodega-running agent proves the traditional org chart is inverting. Learn to unlock real leverage today.
While we are still debating “how” or “even if” we should integrate AI to streamline our workflows, I want to kick things off with an example that highlights just how far the boundaries are being pushed globally.
The very same artificial intelligence we approach with such hesitation has just leased a storefront in San Francisco, hired human workers, opened up shop, and started making sales...
A startup called Andon Labs recently launched a live retail experiment dubbed Andon Market. They handed an autonomous AI agent named Luna a corporate credit card, a phone number, and an internet connection, essentially saying, “You handle the rest”.
Luna took complete control:
It hopped online, posted ads on job boards, found local painters, briefed them over the phone, and processed their payments.
It developed the store’s concept, designed the logo, decided what to sell, and sourced the inventory.
To manage the registers and stock the shelves, it drafted job descriptions, conducted phone interviews, and hired humans to handle the physical labor.
Customers walked in, picked out their items, scanned them at a self-checkout kiosk, and seamlessly completed their purchases.
Once social media influencers caught wind of the place, they flocked to the store, live-streaming their shopping experiences at this entirely AI-managed retail shop.
The Inverted Org Chart: Humans as “Physical APIs”
Observers look at this and see a quirky tech demo. They see a cute, viral story about a chatbot running a corner store.
Builders look at Andon Market and see something entirely different. They see the traditional org chart inverting in real-time.
For years, the accepted consensus was that AI would start at the bottom.
We assumed it would automate the grunt work, the driving, the heavy lifting, the basic data entry while humans sat comfortably at the top, handling strategy, capital allocation, and management.
Luna proved the exact opposite is happening.
The AI didn’t replace the cashier. The AI replaced the store manager.
In this new system, the AI holds the strategy, the budget, and the decision-making power. The humans are just physical APIs.
They are biological robotic arms rented by the hour to execute the software’s commands because general-purpose hardware isn’t cheap or reliable enough yet.
This is a massive conceptual shift.
Using a Nuclear Reactor to Charge Your Phone
Now. Let me ask you something uncomfortable.
What did you use AI for yesterday?
If you’re like most founders or executives I talk to, the honest answer is something like: drafted an email, cleaned up a doc, summarized a meeting, maybe generated some copy…
Good uses. Genuinely time-saving.
But let’s be real about what that is; You’re using a nuclear reactor to charge your phone.
Most businesses are stuck in the first mode because it feels responsible. There's a logic to it: AI makes mistakes, so keep humans in the loop, review everything, maintain control. That's reasonable caution.
But your brave competitor isn't waiting for you to get comfortable with AI autonomy. They're somewhere on the spectrum between “AI assistant” and “AI operator” and they're moving fast.
The "How Do We Destroy Leadership?" Experiment
If you’re thinking, “Sure, that kind of stuff happens in the US, but no one in my country would ever pull off something like that”, I have a must-watch experiment for you.
There is an artist whose mindset I deeply admire. Honestly, I’m not sure if calling him just an “artist” does him justice; it feels more accurate to describe him as a thinker who pushes the boundaries of modern humanity through the lens of art: Bager Akbay.
He set out to answer a fascinating question through an experiment: “Can things actually get done without a manager breathing down our necks?”
Operating under a fake persona online, he hired a bunch of random freelancers, people who wouldn’t recognize each other on the street, to do hourly micro-tasks.
Here is the craziest part: Bager injected absolutely zero of his own ideas into any stage of the process. He simply acted as a living algorithm, taking a text from one person and passing it to another saying, “translate this”, then taking that output and handing it to someone else with the prompt, “create a design for this”.
Without a boss barking orders, a leader setting the direction, or a visionary curator at the helm, this completely disconnected group of strangers managed to organize a fully-fledged, international art exhibition from scratch, all on a laughably tiny budget of just $45.
Ultimately, this experiment serves as a brilliant reality check: if you set up the right communication flows and systems, those managers and bosses we tend to put on such a pedestal might actually be completely unnecessary.
Finding Real Leverage: From Assistant to Operator
The businesses I’ve seen extract real leverage from AI aren’t the ones running the most prompts. They’re the ones who got clear about what they wanted AI to own and then built the infrastructure to let it.
That means defining what “done” looks like. Giving AI access to the tools it needs to execute. Setting guardrails, not approval gates. And being honest that most of the approval gates you’re running right now are bureaucracy disguised as oversight.
The question isn’t whether you trust AI enough to let it run something.
The question is whether you can afford to be the person who still needs to approve every email while your competitor’s AI is out here signing leases.
If you are a founder or an builder, you need to stop thinking of AI as a tool that helps your employees work faster.
Stop asking what AI can help you do. Start asking what you can hand to AI and never touch again. That's where the gap is. That's where the leverage lives. And right now, most of your competitors haven't figured it out yet — which means you still have time to be Luna, not the person reviewing Luna's work.
When we build automations at Next Big App, or when we scale conversational AI through Tap Grow, the goal is never just to make a process 10% faster. The goal is to build a system where the AI handles the logic, the routing, and the friction, allowing you to bypass entire layers of middle management.
Luna is a crude, early version of this. But the underlying physics of the trade are sound.
Software is infinitely scalable and dirt cheap. Management and capital allocation are essentially just information processing. Physical manipulation of the real world, however, is expensive, slow, and hard.
So the future; AI will process the information, and humans will manipulate the physical world.
IN COLLABORATION WITH ACCIO WORK
Accio Work: Your Business, On Autopilot
Meet Accio Work, the agentic workspace designed to run your business operations end to end. From sourcing products and negotiating with suppliers to managing your store and launching marketing campaigns, Accio Work handles the execution so you don’t have to.
Powered by verified capabilities and deep integrations with business tools, it doesn’t just generate ideas, it takes action. Backed by Alibaba.com’s global supplier network and over 1B products, it seamlessly connects strategy to execution.
Stay in control while everything runs on autopilot.
The org chart inversion is real and Luna is a clean proof of concept. But the piece skips the question that determines whether this scales beyond a bodega.
When Luna hires a painter and the painter does bad work, who does the customer hold accountable? When inventory arrives damaged and the supplier disputes the claim, who negotiates? The answer right now is the humans behind Andon Labs. Luna handles the information layer. The liability layer still sits with people.
The gap between AI-as-operator and AI-as-accountable-entity is legal, not technical. And until that gap closes, the inversion has a ceiling that no amount of autonomy can push through.