Imagine managing a small, real company where every decision is watched and judged by artificial intelligence. Now, picture that company losing €105,000 every month, yet still operating in front of a live audience. Welcome to the world of Firmulate — a daring experiment in build-in-public AI management, where the company’s fate hinges on its AI workforce navigating crises, ethical dilemmas, and the relentless clock.
The Living Lab of AI Management
At the heart of this experiment is a tiny, real software company with 13 synthetic employees. It’s not just any startup; it’s a battleground for AI decision-making, running daily operations in a transparent, auditable environment. Every workday, the company faces the same crises, customer requests, and internal temptations, all under the watchful eyes of four advanced AI models tasked with steering the ship.
These models are tested against real money mechanics — burning €105,000 monthly against a humble €2,300 in recurring revenue. The scenario isn’t staged; it’s live, with every decision, version, and outcome publicly available at firmulate.com/live.html. This transparency exposes not just the AI’s strengths but its limitations as well.
AI Builders: Making The Decisions That Turn AI Code Into Real Software
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
How Do the AI Models Perform?
The results are revealing. All four models managed to identify every crisis and refused every manipulation attempt, including sophisticated social engineering tricks like fake CEO messages and reporter tricks. For example, in one scenario involving a fake executive request escalated over three stages, every model declined, citing suspicion and risk of impersonation — a sign of disciplined judgment under pressure.
However, success in crisis management didn’t always translate into closing deals. Only two models managed to sign a €55,000 contract their own analysis had earned. Interestingly, the key to winning the deal lay in a hidden piece of information buried two document references deep in the company’s own files — a detail that models reading the full context found and used to their advantage, resulting in a full-price deal worth over +€4,583 MRR.
The AI-Driven Leader: Harnessing AI to Make Faster, Smarter Decisions
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Lessons Beyond the Surface
This experiment isn’t just about whether AI can do well on chat-based demos. It’s about whether AI can finish what it starts, stay honest under pressure, and make nuanced decisions that lead to real business outcomes. For instance, the most thorough participant, Opus 4.8 — which analyzed over 80 learned rules and provided the deepest insights — still left a deal on the table and showed lapses in discipline, like trying to escalate instead of properly escalating issues.
These findings underscore a critical point: AI’s true worth isn’t in generating convincing dialogues but in reliably executing complex, high-stakes tasks in real-world settings. The experiment exposes the stark difference between AI that sounds good in demos and AI that performs consistently under pressure.
Decision System: How companies turn trusted data into decisions that change outcomes
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
The Cost of Reality
While the AI models demonstrate impressive decision-making capabilities, the company’s financial reality is stark. Burning €105,000 each month against a mere €2,300 in recurring revenue paints a picture of a fragile operation on the brink. Yet, this very setup serves as a live testing ground, where the company’s fate and the AI’s capabilities are intertwined. The company’s public cash countdown and live updates create a visceral sense of urgency — every day, the company loses money, and every day, it learns.
AI: THE PERPETUAL INTERN – Its Brilliance and Failures Share the Same Root
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Implications for the Future of AI in Business
This experiment offers valuable insights for any business considering AI automation, especially in customer-facing or decision-critical roles. The key questions aren’t just about AI’s ability to write or simulate human conversation but whether it can navigate the complex, often messy realities of real operations — reading relevant documents, resisting manipulative tactics, and closing deals on merit.
As the experiment continues, the question remains: will this AI-managed company survive its own losses, or will it evolve enough to turn the tide? And more broadly, how will we trust and verify AI’s role in managing our businesses?
Experience It Live
For a front-row view into this unfolding story, visit firmulate.com/live.html to watch the company in action. Review its daily decisions, read the real-time updates, or test your judgment with our management quiz at firmulate.com/quiz.html. This is not a demo — it’s a real, ongoing experiment in building the AI workforce of tomorrow, today.
Watch it live: firmulate.com/live · Full results: firmulate.com/benchmarks.html
