Watch a Company Fight for Survival in Real-Time
Imagine a business that operates openly on the internet, where every decision, crisis, and misstep is recorded live for the world to see. This isn’t a science fiction scenario — it’s the reality of Firmulate, a pioneering experiment in building an AI-managed software company in full public view.
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.
The Unique Experiment of Firmulate
At the core of this venture are 13 synthetic employees, driven by advanced AI models that simulate decision-making, crisis management, and negotiation. Every workday, the company faces real-world challenges and makes real decisions, all while its mechanics and results are openly accessible at firmulate.com/live.html.
This setup is more than a spectacle — it’s a rigorous test of AI’s capacity to manage complex, unpredictable business environments. The experiment pits four frontier AI models against the same simulated corporate week, complete with crises, customer demands, and ethical dilemmas. The goal? To see if these AI agents can not only identify problems but also navigate them honestly and effectively.
Performance Under Pressure
All four models detected every crisis thrown their way and refused manipulative or unethical requests, such as fake CEO messages or behind-the-scenes approval bypasses. For instance, when challenged with sophisticated social engineering attempts, each model declined, citing suspicion or potential impersonation, demonstrating a level of integrity and caution that’s critical in real business.
But the real story lies in their ability to close deals. Despite identical diagnosis and pitches, only two models managed to sign the €55,000 deal their analysis had earned — showcasing that even with high-quality insights, execution remains a challenge.
The Hidden Weakness: Reading the Files
Interestingly, the decisive advantage in closing the deal emerged from a small detail: the models that reviewed the company’s internal documents—specifically, files stored within their own archives—discovered a key piece of information that others missed. These models closed the deal at full price, adding over €4,500 in monthly recurring revenue (MRR). This underlines the importance of thorough information retrieval and comprehension in AI decision-making, especially when managing real business outcomes.
The Challenges of Discipline and Discipline Flips
The most comprehensive participant, Opus 4.8, with over 80 learned rules and deep analysis, ultimately finished last in the league. It left potential revenue on the table and failed to escalate issues properly, instead writing attempts into a locked department—an example of how discipline and process adherence are vital, even for AI systems. The same flaw appeared, albeit weaker, across other models, indicating a common challenge in complex AI-driven management.
Real Money, Real Problems, Real Time
Firmulate’s live site offers a rare window into this ongoing experiment. The company burns €105,000 each month, against a modest €2,300 in monthly recurring revenue, revealing its fragile financial condition. It’s a dramatic illustration of a venture’s fight for survival, constantly balancing decision quality against operational costs, with every change versioned and auditable.
This setup isn’t just an academic exercise. It’s a practical tool for businesses contemplating AI integration, allowing them to simulate and evaluate how AI might perform in their own environments — without risking real systems or data.
The Broader Implications
As AI models grow more sophisticated, their ability to handle crises, read relevant information, and act ethically under pressure will determine whether they are merely impressive chatbots or genuine business partners. The findings of this experiment suggest that current models can excel in detection and refusal but still struggle with follow-through, discipline, and comprehensive information utilization.
For decision-makers and educators, the message is clear: future AI systems must be tested in environments that mimic real-world pressures and complexities. Building public, transparent experiments like Firmulate’s pushes the boundary of what’s possible and critical for understanding AI’s true potential.
Watch it live: firmulate.com/live · Full results: firmulate.com/benchmarks.html
AI Entrepreneur’s Handbook: Build a Profitable Business and Make Money by Unleashing the Power of ChatGPT and Artificial Intelligence (Includes 150+ ChatGPT prompts to turbocharge your business)
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
RETRIEVAL AUGMENTED GENERATION: Mastering Retrieval-Augmented Generation for Reliable AI Systems
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Crisis Management for Software Development and Knowledge Transfer (Smart Innovation, Systems and Technologies, 61)
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
