Skyfall AI Will Buy a Real Company and Let AI Run It as CEO

The experiment

While OpenAI, Anthropic, and Google race to build AI agents that can handle individual workplace tasks, a startup founded by former Microsoft AI leaders is taking a radically different approach. Skyfall AI plans to acquire a small B2B SaaS or e-commerce company for up to $1 million and attempt to run it almost entirely with artificial intelligence serving as CEO.

The company will publicly document the experiment's successes and failures as it tries to double the acquired company's revenue while progressively reducing human involvement. According to details first reported by Forbes, Skyfall's AI system will oversee pricing, marketing, customer support, finance, and operations.

"If you want a truly autonomous business, you have to buy a business and operate the whole thing end to end," Skyfall cofounder and CEO Sam Pasupalak told Forbes. "Unless you run a business with minimal human intervention, you'll never know whether an autonomous enterprise is actually possible."

Why it matters

This experiment represents a fundamental challenge to the prevailing enterprise AI strategy. Rather than automating jobs piecemeal—a marketing agent here, a sales agent there—Skyfall is testing whether AI can handle the abstract reasoning, long-term planning, and incomplete-information decision-making that defines executive leadership. If successful, it would validate a path toward business automation far more ambitious than today's task-specific agents. If it fails publicly, it will provide rare transparency into the actual limits of current AI systems in complex, evolving environments.

The founders' pedigree

Pasupalak and cofounder Kaheer Suleman met studying computer science at the University of Waterloo nearly two decades ago. They cofounded Maluuba, an early deep learning startup that built conversational AI systems reaching hundreds of millions of devices through partnerships with Samsung, LG, and BlackBerry. Microsoft acquired Maluuba for approximately $160 million in 2017, transforming it into the foundation of Microsoft AI in Canada.

The technical argument

Skyfall's founders believe current large language models have reached their enterprise limits. They've developed a benchmark called Morpheus that tests whether frontier AI models can continually adapt as business conditions evolve. According to the company's research, models like GPT-5.5 and Gemini perform well in familiar conditions but deteriorate as competitors launch products, customer behavior shifts, and market dynamics change.

Their proposed solution: "Enterprise World Models" that simulate how decisions ripple across an organization rather than simply applying pre-trained knowledge. "Instead of predicting future states in token space, we're predicting them in a latent space," Suleman explained to Forbes. "That allows us to focus only on the features that actually matter to the dynamics of the business."

Not the first attempt

Skyfall isn't entering completely uncharted territory. Anthropic's Project Vend, developed with AI safety startup Andon Labs, gave a Claude-powered agent control of a real vending business. Early versions made comical mistakes—employees convinced it to stock novelty tungsten cubes and sell them at a loss—but later iterations became competent enough that Andon Labs called the operation "almost boring."

What distinguishes Skyfall's experiment is scale and transparency: acquiring a revenue-generating business, publicly measuring whether AI can grow it while reducing human intervention, and documenting failures alongside successes.

The human role

Even Pasupalak doesn't envision AI replacing every leadership function. "When it comes to motivating employees and building relationships with people, that's still very difficult for AI to replicate," he told Forbes. Suleman added that humans should maintain oversight and accountability even as AI handles operational workload.

The experiment will test whether AI can manage an organization where strategy, competition, customer behavior, and market conditions constantly evolve—a far more demanding proposition than the relatively constrained retail environment Anthropic has already automated.

Details of the experiment were first reported by Victor Dey at Forbes.