The Mega-Bill McDermott Prophecy Why Your AI Strat

I have a provocative tech theory. I cannot help but go into the future everyday. I have come back with this. Even people who expect human-level AI soon are still seriously underestimating how different the world will look when we have it.

The reason? It’s not just about individual AI capabilities—it’s about the collective advantages these systems will have. They can be copied, distilled, merged, scaled, and evolved in ways humans simply can’t.

As CEOs at Shopify, Duolingo, and Box release manifestos declaring their companies “AI-first” and Microsoft heralds the birth of the “frontier firm” in its 2025 Work Trend Index, my vision of fully automated enterprises is rapidly transitioning from speculative futurism to corporate strategy.

The Copying Advantage

Picture this: What if Google had a million AI software engineers? Not generic “workers,” but identical copies of their top talent—the AI equivalents of Jeff Dean and Noam Shazeer, with all their skills, judgment, and tacit knowledge intact.

[image: ChatGPT Image May 15, 2025, 06_29_08 PM]

This ability to turn capital into compute and compute into equivalents of your top talent is a fundamental transformation. Since you can amortise the training cost across thousands of copies, you could sensibly give these AI’s ever-deeper expertise.

For Dow Inc, this isn’t theoretical. The global materials science company has deployed agent systems to detect hidden losses and streamline shipping operations. Once fully scaled, they expect the system will save millions in the first year through increased accuracy in logistics rates and billing.

Similarly, at Wells Fargo, an AI agent built for 35,000 bankers across 4,000 branches has cut query response times from 10 minutes to just 30 seconds, with 75% of searches now happening through the system.

Micromanagement at Scale

[image: ChatGPT Image May 15, 2025, 06_57_29 PM]

But copying goes beyond replicating star performers. It transforms management even more radically than labour.

Human Bill McDermott simply doesn’t have the bandwidth to directly oversee 30 000 employees, hundreds of products, and millions of customers. But AI Bill’s bandwidth is capped only by the number of TPUs you give him to run on.

This enables a level of oversight that makes current management practices look primitive. I envision a world where all of ServiceNow’s middle managers could be replaced with AI Bill copies, reviewing every pull request, crafting every product strategy, and handling every negotiation—all flowing from a single coherent vision.

The concept directly parallels what Microsoft has identified as a key trait of frontier firms: the rise of the “agent boss.” According to their research, 28% of managers are already considering hiring AI workforce managers to lead hybrid teams of people and agents, while 32% plan to hire AI agent specialists within the next year.

Knowledge Transfer Without Limits

Mega-Bill, is the centralised intelligence, learns from everything experienced by its specialised copies—every customer conversation, every engineering decision, every market response.

[image: ChatGPT Image May 15, 2025, 07_02_43 PM]

Unlike Tesla’s FSD, this doesn’t have to be a naive process of gradient updating and averaging. Mega-Bill will absorb knowledge far more efficiently—through explicit summaries, shared latent representations, or even surgical modification of the weights to encode specific insights.

This perfect knowledge transfer represents what I call “the Knowledge Network Effects, a step change in how organisations can accumulate and apply knowledge.” While human organisations struggle with information silos and knowledge transfer bottlenecks, AI firms can instantly propagate insights across the entire operation.

As Microsoft’s report confirms, Frontier Firm employees are already experiencing the benefits. They’re more than twice as likely to say they’re able to take on additional work (55% vs. 20% globally) and substantially more likely to report having opportunities for meaningful work (90% vs. 73% globally).

Scale Without Constraint

In a fully automated firm, the cost to have an AI take a given role becomes just the amount of compute it consumes. This fundamentally changes which roles are considered scarce or valuable.

Future AI firms won’t be constrained by what’s scarce or abundant in human skill distributions—they can optimise for whatever abilities are most valuable. Want Andrej Karpathy-level engineering talent? Cool: once you’ve got one, the marginal copy costs pennies.

What becomes expensive in this world?

[image: image]

Roles which justify massive amounts of runtime compute. The CEO function is perhaps the clearest example. Would it be worth for Google to spend £75 billion annually on inference compute for mega-Sundar?

Just consider what this buys you: millions of subjective hours of strategic planning, Monte Carlo simulations of different five-year trajectories, deep analysis of every line of code and technical system, and exhaustive scenario planning.

Microsoft’s Work Trend Index supports this vision of scaling intelligence. Among the 31,000 knowledge workers surveyed, 81% expect AI agents to be moderately or extensively integrated into their company’s strategy within 18 months, with 45% of leaders saying expanding team capacity with digital labour is a top priority.

Distillation and Specialisation

What might specialised copies of these central intelligences look like? I’d suggest they would be highly specialised in function, tacit knowledge, and complex skills—while potentially sharing more factual knowledge than one might expect.

The cost of storing raw information is so unbelievably cheap that LLaMA 7B already knows more about the standard model and leaky drains than any non-expert.

Good and Featured Wikitext is less than 5 MB. I don’t see why all future models wouldn’t at least have Wikitext down.

This vision of specialised yet knowledgeable agents aligns with how frontier firms are already deploying AI. At Bayer, researchers on the Crop Science R&D team each save up to 6 hours per week using specialised research agents.

The Estée Lauder Companies has built an agent to identify and consolidate consumer insights, allowing teams to instantly access intelligence that previously required sifting through scattered reports.

Evolution at Unprecedented Speed

When I peeked into the future. The most profound difference between AI firms and human firms are their evolvability, I refer us all to the technology philosopher Gwern Branwen’s observations about corporate limitations.

Traditional corporations struggle to replicate themselves because they’re made of people—not interchangeable, easily copied widgets or strands of DNA. By contrast, AI firms can clone their culture, institutional knowledge, and operational excellence with perfect fidelity.

The scale of difference between currently existing human firms and fully automated firms will be like the gulf in complexity between prokaryotes and eukaryotes.

This evolvability advantage helps explain why frontier firms appear to be pulling ahead. Microsoft’s research shows that 71% of frontier firm workers report their company is thriving, compared to just 37% globally. They’re also far less likely to fear AI taking their jobs (21% vs. 38%).

The Economic Endgame

[image: ChatGPT Image May 15, 2025, 07_05_36 PM]

The ultimate question I pose to us all is existential:

If you can perfectly replicate any capability, why would you ever pay some markup for another firm, when you can just replicate them internally instead? Would the first firm that can figure out how to automate everything just form a conglomerate that takes over the entire economy?

While I’ve previously suggested Ronald Coase’s theory of the firm implies that dramatically lower transaction costs should lead to much larger organisations, I don’t believe this necessarily ends with one giga-firm which consumes the entire economy.

The reason? Even automated planning systems need to be grounded in some kind of “loss function”—a ground truth measure of success. In a market economy, this comes from profits and losses.

Internal planning can be much more efficient than market competition in the short run, but it needs to be constrained by some slower but unbiased outer feedback loop. A company that grows too large risks having its internal optimisation diverge from market realities.

The Here and Now

I concur with those of you thinking my vision may seem distant. However, Microsoft’s comprehensive research suggests we’re already witnessing the emergence of these frontier firms. The survey of 31,000 workers across 31 countries reveals that 24% of leaders say their companies have already deployed AI organisation-wide, while only 12% remain in pilot mode. I am suspicious of the extent of “organisation-wide deployed AI”—which could just be Copilot for MS Teams meetings. This is not deep, but it is a start and you must find more.

At Duolingo, CEO Luis von Ahn’s recent memo outlined concrete steps toward becoming what I would recognise as a fully automated firm: phasing out contractors for work AI can handle, tying AI usage to hiring and performance reviews, and making headcount contingent on maximising automation.

Similarly, Box CEO Aaron Levie has instructed teams to “use AI to automate more and save money,” while encouraging constant experimentation to find optimal use cases and upskilling every employee to be “AI-first.”

The Human Element: Skepticism and Concern

[image: image]

Not everyone shares this enthusiasm. I have written about this before, enterprise engineers are simply not convinced that this is possible anytime soon. They argue, and I agree, that there isn’t yet an enterprise grade deployment of any significant scale that would qualify anywhere close to a fully AI automated firm. Jaroslav Sýkora, a computer electronics hardware engineer with 15 years of experience, wrote: “I just can’t get over the idea that some computer system should ever work so well as to viably replace humans doing non-trivial jobs.”

Another commenter, Jai Rai, raised more fundamental concerns: “Efficiency is not everything. When you replace a Google engineer, you might increase Google’s bottom-line, but you also replace a source of demand for rest of the economy.”

These critiques echo what economist Daniel Susskind has identified as the three limits that will preserve human work: efficiency (it can be more effective to have AI and humans working together), preference (humans may prefer interacting with other humans), and moral judgment (society expects people to be responsible for consequential decisions).

Microsoft’s data suggests that humans still bring unique value. When asked why they turned to AI instead of colleagues, employees cited 24/7 availability (42%), machine speed (30%), and unlimited ideas (28%)—all things humans cannot provide. Notably, avoiding human traits like impatience or judgment ranked lowest.

The most valuable knowledge in enterprise technology resists digitisation by its very nature. In our work at Eclipse AI with financial services and healthcare clients, we consistently see three types of knowledge that AI struggles to capture:

1. Pattern recognition for undocumented failures - Experienced engineers who “just know” when something looks wrong
2. Contextual judgment for legacy systems - Understanding the quirks and historical decisions behind seemingly illogical design choices
3. Crisis heuristics - The instinctive problem-solving approaches that emerge during critical failures

These aren’t gaps we’ll close with bigger models or more training data. This is the quantum space where human cognition operates fundamentally differently than AI.

And in there lies the angle. We’re not building thinking machines. We’re building thinking partners.

The most successful enterprise AI implementations we’ve delivered don’t try to replace human intelligence—they amplify it. Our insurance claims processing vertical agent achieves 92% accuracy versus 47% for generalist approaches precisely because it’s designed to complement human expertise, not replace it.

Human + AI = Superhuman

[image: Screenshot 2025-05-10 at 13.47.17]

I will repeat two points that may seem counter to each other. First AI is better than humans.

Even people who expect human-level AI soon are still seriously underestimating how different the world will look when we have it.

The reason? It’s not just about individual AI capabilities—it’s about the collective advantages these systems will have. They can be copied, distilled, merged, scaled, and evolved in ways humans simply can’t.

Now the reason humans are better than AI.

[image: ChatGPT Image May 15, 2025, 11_08_45 PM]

The most valuable knowledge in enterprise technology resists digitisation by its very nature. Tribal and tacit knowledge are incomparable to Gen AI and its precusors. Experienced engineers who “just know” when something looks wrong. Marketing specialists understanding the quirks and historical decisions behind seemingly illogical design choices. In industry, machine operators instinctive problem-solving approaches that emerge during critical failures

These aren’t gaps we’ll close with bigger models or more training data. This is the quantum space where human cognition operates fundamentally differently than AI.

And in there lies the angle. We’re not building thinking machines. We’re building thinking partners.

Human + AI equals a superhuman and productivity for enterprise that is not just—one unit in and a linear output—it will be exponential.

Most will miss this opportunity. They’ll continue chasing the fantasy of human-free operations. Meanwhile, the winners will build cognitive partnerships that leverage the best of both worlds, creating capabilities that neither humans nor machines could achieve alone.

The question isn’t whether AI will transform your business. It’s whether you’ll understand what it actually transforms, and how to harness that transformation.

The future isn’t artificial intelligence.

It’s augmented humanity.

The most forward-thinking organisations are already preparing their workforces for this transition. Microsoft’s research shows that 47% of leaders list upskilling existing employees as a top workforce strategy, while 51% of managers say AI training will become a key responsibility for their teams.

The Great Rewiring

Markets facilitate an evolutionary process which selects not only goods and services, but the institutions that are best at turning the world into valuable goods and services. I think this will continue.

What’s changing is not the fundamental role of markets, but the nature of the firms operating within them. Intelligence—once the most constrained resource in business—is becoming abundant, available on tap, and able to be copied, merged, and evolved at unprecedented speed.

Intelligence is being commoditised.

[image: ChatGPT Image May 15, 2025, 10_50_13 PM]

The frontier firm, as Microsoft calls it, or the fully automated company, as my visions perceive, represents more than just an efficiency upgrade. It’s a fundamental rewiring of how organisations operate—as profound as the shift from prehistoric prokaryotes to the complex eukaryotic cells that enabled all higher forms of life.

For business leaders, the message is clear: The future belongs not to those who merely adopt AI, but to those who rebuild their entire operation around the unprecedented capabilities it offers.

About the Author: Chris Jones is CTO of Eclipse AI, where he helps organisations navigate the complex landscape of AI implementation. Drawing on his experience across software development, system architecture, and AI strategy, he brings a uniquely multidisciplinary perspective to the challenges of integrating artificial intelligence into business operations.