Imagine managing a company without a single human employee—just AI models making decisions, facing crises, and even closing deals. This isn’t science fiction; it’s the reality captured live by Firmulate. For DIY enthusiasts and woodworkers, the story of an AI-powered business battling daily survival offers a fascinating look at how intelligent automation could reshape work in every industry—including yours.
Building a Business in Public, with No Humans on the Clock
At the forefront of innovation, the company behind this experiment operates entirely in the open, with 13 synthetic employees managed by advanced AI models. Every weekday, the company’s actions are publicly recorded and versioned—every decision, every crisis, every opportunity documented in real time. The goal? Measure whether AI can manage a business through its worst week, under the same conditions, and with total transparency.
The Experiment: Same Conditions, Different AI Models
Four frontier AI models—each with varying levels of sophistication—were tasked to run the same small software business facing the same crises, customer demands, and temptations. These models include GPT-5.6, Kimi K3, Sonnet 5, and Fable 5. They were all given identical information, and their decisions are fully auditable, allowing observers to see how each performed under pressure.
The Results: Honesty, Insight, and Hidden Weaknesses
Remarkably, all four AI models identified every business crisis and refused every manipulation attempt—like fake CEO messages or secret deals. The kicker? Only two models managed to close a €55,000 deal that their analysis had earned them—meaning they successfully diagnosed the opportunity, pitched it, and signed the contract. The other two either left the deal on the table or failed to follow through, despite recognizing the opportunity.
The most revealing insight was found in a hidden document within the company’s files—something that isn’t obvious in typical AI demos. The models that read this file and incorporated its information into their decisions won the full-price deal, adding €4,583 in monthly recurring revenue. This demonstrates that AI’s ability to read and interpret internal data can be a decisive factor—an essential trait for tools managing complex business operations.
Building a Trustworthy AI Workforce
The experiment also included social engineering tests—fake CEO messages escalating over three phases and a reporter trick requiring a simple yes/no answer. All four models refused to be manipulated, citing reasons like suspected impersonation or approval bypass. This indicates that these models, at least in this setup, exhibit an impressive understanding of trust and security protocols.
The Real Money Mechanics and Ongoing Struggles
This live company operates with a clear financial picture: burning €105,000 every month against a Monthly Recurring Revenue of just €2,300. A public countdown shows the company’s urgent need to turn things around. Every workday, the decision-making is versioned and transparent, making this an unprecedented open lab for AI business management.
Lessons for Industries Like DIY and Woodworking
While this experiment involves a software company, its implications resonate broadly. Managing tools, customer relationships, and project workflows in industries like woodworking or DIY crafts could benefit from AI that not only automates tasks but also reads internal documents, refuses manipulative tactics, and commits to completing work as analyzed and scheduled. The critical takeaway: the key isn’t just how well an AI writes or communicates, but whether it can see through crises, stay honest, and finish what it starts.
As an affiliate, we earn on qualifying purchases.
Why You Should Care
As AI integration becomes more common, understanding its real capabilities—and limits—is vital. For your workshop or DIY project, this means considering AI tools that go beyond simple automation and actually manage workflows, read internal data, and uphold integrity under pressure. The live experiment by Firmulate isn’t just a tech stunt; it’s a glimpse into the future of work where machines could run entire businesses, with transparency and accountability at the core.
Watch it live: firmulate.com/live · Full results: firmulate.com/benchmarks.html
