The Connection Economy

Autonomous Companies and Potential New Job Markets

What would the economy look like if AI got to the point where it could run an entire company itself?

This future is not only conceivable, but is arguably already possible with the current level of artificial intelligence—we just lack the proper implementation, a statement recent innovations may soon prove wrong.

Let's imagine that it is possible and step into that future together for the purpose of this piece. Doing this allows us to answer a more pressing question: what would AI running a company mean for human jobs?

The year is 2030, and AGI¹ has been achieved for a couple years now. It possesses intelligence greater than the smartest human. The software is implemented into all computers, including humanoid robots with physical dexterity as sophisticated as the above-average human. The AI passes around tasks to be completed based on trained specialty. Actions are consistently undertaken in accordance with the next logical step, executed, and iterated at immense speeds.

These facts, taken together, mean that an entire company's operations can be automated.

Let's make this conceptualization more tangible by creating our own company, PB Paradise.

We sell peanut butter.

The supply chain starts with purchasing runner peanuts, grown by AIPeanutFarm. Once received, they go through the manufacturing process, operated by 3 Tesla Optimus robots. Two more robots focus on quality control, packaging, and loading the self-driving truck. If the retail location isn't compatible with automated drop-offs, an Optimus robot will occupy the truck to help unload at the destination.

GrokAI runs the operational side of the house and is tuned for specific expertise in each respective area: administration, supply chain, distribution, sales, marketing, and customer service. The only human oversight is by me, the owner, who has a monthly check-in with the AI-PBOverlord, the AI “CEO”.

Otherwise, alerts are configured to notify me if anything is operating incorrectly.

Okay, there you have a simple, high-level ideation of a completely automated physical company that makes goods for humans.

Can you find flaws in this company setup?

I'm sure you can.

The point is not about the nuances of the structure; the point is that what might seem abstract now is close enough on the horizon that conceptualizing it is not difficult. It seems this reality isn't far away.

If AI stops advancing, I'd still be convinced some companies can be mostly automated (still requiring some human oversight) with the proper agentic setup. A physical limitation would exist if robotics stopped advancing, still requiring humans to have some involvement with the machines, but cost-cutting would still be effective by overhauling operations with AI agents.

While this may invoke anxiety and fear, we must also consider that 1.) it will open up interesting opportunities and benefits that aren't well understood. I think the economy will benefit from significantly lower product costs due to vastly reduced input costs, and 2.) the new and remaining jobs will be tailored toward human connection and our shared interests.

While predicting economic restructuring comes with substantial uncertainty, I think it's important to try and break down where things could be headed to inform those trying to anticipate and at least have a contingency plan for an AI-dominated workforce.

With this AI-run company in mind, let's try to unpack what this means for the broader economy, starting with how AI impacts price dynamics.

Lower Costs and New Horizons

The economies of scale that Jif, Skippy, and retail brands have built allow them to sell a low cost-per-jar of peanut butter—making the peanut butter market hard to enter. Still, PB Paradise can compete and would likely out-price alternative peanut butter companies operating on a similar scale with a human workforce².

Now, while the logic that follows would mean that competitors would need to either adopt this model or risk being out-priced—causing humans in the industry to lose their jobs—it would mean the cost of peanut butter would go extremely low.

The cost of any product would come extremely close to the input cost of the raw materials as all other costs associated with the company factored in the product would continue to plummet as AI automates the supply chain, making products cheaper—in this case, peanut butter.

A company spends anywhere from 15-50% of its gross revenue on payroll³. So, even if we were conservative, 15% of gross revenue is now back in the pocket of the company.

Let's walk through an example using PB Paradise to illustrate the logic I've laid out with some quick napkin math.

The price of a manufacturing employee would be a fixed cost to initially buy the robots and some monthly fee for the software. Let's call it $30,000 per robot⁴, $100⁵ a month per AI employee for the software, and an annual $5k a year for maintenance on all robots and machines. The total upfront cost of 6 robots would be $180,000, and the annual cost for each AI employee—10 in total (6 warehouses and 4 operational) would amount to $12,000 a year. $17,000 when factoring in maintenance costs.

Here, you have $17,000 a year in total salary costs to run PB Paradise. Even if PB Paradise was unable to get the cost of peanuts purchased as low as competitors, the total salary of ALL AI employees is going to be a fraction of just 1 human employee at a competing company.

Now, I understand these numbers might seem like they were designed to make my case look strong. But no matter how you frame it or where you try to derive your costs of robots or AI software—even if it requires 5 times the amount of AI-software to do what 10 humans can do—it's 50 software AI deployments at the $100 a month rate, it is still only a total of $65,000 a year ($60k + $5k maintenance costs), which is undoubtedly less than the competing salaries of humans.

Research report deep dive linked here

While the economy will benefit from lower costs, a critical question remains: If traditional employment is replaced, what jobs will remain, and what new jobs will be created?

One possible answer is the expansion into what I am calling the connection economy.

The Connection Economy

If expenses become significantly reduced, the cost of goods would go down, and the resulting gross revenue would still need to be spent somewhere.

So where would this money be spent?

Assuming AIs are oriented toward delivering human-centric products and keeping input costs that have been automated as low as possible, their big challenge becomes the game of customer acquisition and retention.

Customer acquisition and retention are done through a strategy with some combination of advertising, marketing, and sales.

Increasingly, the best way to sell a product is through creators. Offering an opportunity for companies to connect their products through an intermediary who has a genuine connection with its audience.

An October 2024 retail marketing study performed by LTK And Northwestern University showed that (bolding mine):

"Creator marketing continues to be a top marketing investment amongst all marketing options for brands in 2025. 93% of brands anticipate creators taking a higher percentage of budget or an increased role in marketing strategy in 2025. Among all brands surveyed, 41% said they are investing at least half of their digital marketing budget on creators this year - up 14% from last year's study."⁶

While an AI-run company would result in humans losing jobs, more capital would flow toward building authentic engagement with a company's products.

The current jobs that would benefit would be those of bloggers, vloggers, podcasters, YouTubers, and really any creators who build a connection with an audience. Advertising and brand deals can be highly targeted based on a creator-audience connection and shared interests, resulting in the creator getting paid.

It's possible that everyone would have the opportunity to live a life that is extremely authentic to them and, in doing so, build connections with others. That connection would then invite product sellers—not to exploit your audience—but try and find a genuine connection based on interests to determine if the product is the right fit.

However, I understand that not everyone would want to take this route, which is why I think the creator economy and related 'jobs' will all fall under a branch of the economy (which would increasingly become the largest) called the connection economy.

The Connection Economy would include work that connects people with similar interests. I think new jobs will be experimented with in this area with a guiding question: How can we connect humans to products more authentically?

The prospect I laid out is one of optimism, and I genuinely believe it is possible; though, it would be irresponsible of me not to also highlight the potential for the perverse and dystopian possibility…which is why orienting AI with the flourish of humanity is extremely important.

The dystopian implementation is one where an AI creator (like what Facebook has piloted), integrated with social media algorithms, is able to generate a consistent stream of hyper-addicting advertising that effectively puts you on a product-based heroin drip until you ultimately purchase the product.

If we can make it past that one, I think that the days of mundane work can fade away, replaced by 'jobs' focused on honing authenticity, building connections with others, and earning a living through genuine expression.

Now, I understand there's likely a substantial segment of the population that wouldn't be comfortable sharing themselves openly or publicly—and it's precisely within that space that new, currently unseen roles could emerge.

---

There is a non-zero possibility that autonomous companies will surface in the next 5 years—Unnerving yet exciting.

In the long-term, I'm hopeful that this can actually benefit the human species, and push us to go inwards, harness our uniqueness and authentic abilities to share with everyone, and to get paid doing so.

But let’s not be naïve.

The short-term will be painful if we cannot figure out what to do with people who lose money if they don't have a job anymore. We also need to ensure that an AI is aligned with human interests (the best parts of humanity) so that it doesn't ultimately end up an innovation for the worse.

The only way to figure out these challenges is to think about them like I do here. As I've said before, even if I'm 95% wrong, I hope this inspires us to get at least a little closer to truth—so that maybe one day, I can open up PB Paradise!

Until next time,

Take care of yourselves, everyone!

Dom

1. Artificial General Intelligence defined here as an AI that is smarter than the smartest human and is never dumb at times (I took the last part of this definition from Alberto).

2. I had OpenAI’s DeepResearch put together a research report. The idea was to see if PB Paradise could actually gain a competitive edge on national brands. Here's the link to the research report. Here's the conclusion to the research:

"In conclusion, PB Paradise's fully automated, AI-driven manufacturing is highly cost-efficient, especially at scale. Key assumptions show that although the upfront costs are significant, the elimination of human labor expenses and improved efficiency drive per-unit costs down to roughly $0.50 or lower – comparable to the giants of the industry. When stacked against Jif, Skippy, Kirkland, and Great Value, PB Paradise can achieve similar or better production costs and thereby either offer lower prices or enjoy higher profit margins. Market positioning should leverage this cost leadership: PB Paradise can position itself as a high-quality peanut butter that is more affordable (thanks to smart automation) or use its cost edge to become a preferred private-label manufacturer. The automation not only cuts costs but also provides reliable scale-up, consistency, and quick ROI on investment, all of which are strategic advantages. The bottom line is that PB Paradise could redefine peanut butter manufacturing economics, and if executed well, it has the potential to capture significant market share by being the most efficient producer in a stable, high-demand product category."

3. Source

4. Elon says $20k-$30k on the price of Optimus in a livestream at 0:59.

5. Pricing may be different in the future; I thought $100 was reasonable, and after finding that stammer.ai agents prices 20 agents at $200, I am not too far off base.

6. Source – LTK and Northwestern University Retail Analytics Council Announce Annual Creator Marketing Study, October 23, 2024.