Imagine a business where there are no human employees, no managers, yet it’s running day after day — making decisions, facing crises, and even losing money. Sounds like fiction? Not anymore. Welcome to the world of Firmulate, an experimental company that’s built in public, with its entire operation on display for anyone to observe, analyze, and learn from.
The Live Experiment: AI as a Company
At the heart of this project is a bold idea: what if artificial intelligence models could run a business, make decisions, and handle crises — all without human intervention? Firmulate’s live site showcases exactly that. Every workday, the company’s decision-making processes are versioned, documented, and open to scrutiny. It’s a real-time lab where four frontier AI models are tested against the same set of challenges, simulating a worst-week scenario for a small software startup.
Operating with 13 synthetic employees and a set of over 680 self-learned rules, the company faces familiar business crises, customer demands, and temptations to cheat or manipulate. Its goal? To see whether AI can manage complex, real-world tasks honestly and effectively, especially in high-pressure situations where human judgment is typically tested.
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.
Key Findings: AI’s Strengths and Weaknesses
One of the most striking results from this experiment is that all four models identified every crisis and refused every manipulation attempt — including fake CEO messages and subtle persuasion tactics. When presented with a fake report request, all models declined, demonstrating a strong capacity for integrity under pressure.
However, the real test was whether they could close a critical deal. Despite all models diagnosing the opportunity correctly and delivering the same sales pitch, only two managed to sign the €55,000 contract — the other two either left the deal on the table or failed to follow through. Interestingly, the decisive advantage came from reading a specific, buried document reference in the company’s files. The models that found this hidden piece of information secured a full-price deal worth over €4,583 monthly recurring revenue, effectively turning knowledge into profit.
Lessons for Business and AI
This live experiment highlights several important truths about AI’s role in business:
- Reading comprehension matters: In this case, the AI that looked beyond surface documents and uncovered hidden facts won the deal, demonstrating the importance of thorough information processing.
- Integrity under pressure: All models refused to be manipulated through social engineering or impersonation tactics, showing that AI can be trained to maintain honesty even when tempted.
- Discipline and follow-through: The most thorough model, Opus 4.8, had the deepest analysis but still left a deal unclosed due to discipline slips—highlighting that even the best AI needs guidance to execute perfectly.
For business leaders, this raises critical questions about deploying AI in customer-facing or decision-critical roles. Will the AI finish what it starts? Will it read and understand your internal files before acting? And crucially, will it remain honest when faced with potential shortcuts?
Building in Public: Transparency and Accountability
Firmulate’s approach is transparency taken to the extreme. Every decision, every rule, and every iteration is publicly accessible, allowing anyone to see how each AI model performs day by day. With a public cash countdown indicating the company’s ongoing financial struggles and a real-time leaderboard, it offers a rare glimpse into the gritty reality of building AI that can truly handle complex, ethical, and financial tasks.
The live site also features a quiz where over 240 real management decisions are used to challenge the models’ judgment—giving viewers an interactive way to understand the nuances of AI decision-making in business contexts.
The Bigger Picture: Why It Matters
This experiment isn’t just about AI playing pretend. It’s a vital step toward understanding how AI can be integrated responsibly into real organizations. As firms consider AI for CRM, support, or forecasting, the questions aren’t just about language or appearance. They’re about whether the AI can finish what it starts, stay honest under pressure, and adapt to unforeseen crises.
The ongoing live experiment at firmulate.com/live.html invites anyone curious about the future of work and AI to watch a small company fight for survival — with all its flaws, triumphs, and lessons on display.
Firmulate’s live AI-run company demonstrates that honesty, thoroughness, and discipline are vital for AI to effectively replace or augment human decision-making. Watching it battle for survival offers valuable insights into building trustworthy AI systems that can truly handle real-world business challenges.
Watch it live: firmulate.com/live · Full results: firmulate.com/benchmarks.html
