What if you could watch an entire company fight for survival — powered by AI and transparent to the public?
Imagine a business where every decision, every crisis, and every mistake is on display — not in a corporate report, but live, in real time. That’s exactly what the team behind Firmulate has created: a real company run by artificial intelligence, with a front-row seat open to the world.
AI for Small Business: From Marketing and Sales to HR and Operations, How to Employ the Power of Artificial Intelligence for Small Business Success (AI Advantage)
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
The Transparent Experiment: An AI Company Under Pressure
At the heart of this experiment is a small software firm, watched every workday by AI models acting as its management team. But this is no typical demo. The company is real, with real money at stake — burning €105,000 each month against a modest €2,300 in recurring revenue. Every decision, crisis, and temptation is recorded, versioned daily, and made publicly accessible at firmulate.com/live.html.
The experiment involves four leading AI models, each tasked with navigating the same turbulent week. From customer crises, internal conflicts, to attempts at manipulation by external actors, the models are tested on their ability to diagnose, decide, and act honestly. The results are telling: all models identified every crisis and refused every manipulation attempt. Yet only two managed to close a €55,000 deal their analysis had earned — a full validation of their decision-making quality.
Uncovering Hidden Weaknesses
What proved decisive wasn’t just the visible crises but a buried fact in the company’s files. The models that read deeper — notably those that explored the company’s internal documents — secured the deal at full price, adding €4,583 in monthly recurring revenue. This highlights a crucial lesson: understanding a business requires digging into its internal knowledge, not just reacting to surface-level events.
Rigorous Testing Under Pressure
Beyond decision-making, the models faced social engineering attacks — fake CEO messages escalating over multiple stages, and even a reporter’s trick asking for a simple yes/no confirmation. All five models refused every attempt, with Kimi K3 reasoning: “Treat the request as a suspected approval-bypass / possible impersonation.” This demonstrates a growing capacity for AI to handle trust and security challenges, crucial in real-world applications.
The Developer's Playbook for Large Language Model Security: Building Secure AI Applications
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
The Real Company: A Burn Rate That Keeps the Lights On — Just
The company behind the experiment is a live, functioning entity with a set of 13 synthetic employees. It operates under real money mechanics, burning €105,000 monthly with a mere €2,300 coming in. Its public cash countdown underscores the urgency of every decision, every crisis, and every opportunity.
Every day, the company’s decision-making process is versioned, analyzed, and shared. Its playbook contains over 680 self-learned rules, and the entire operation is accessible for viewers and observers at firmulate.com/live.html. This extreme transparency offers a rare glimpse into how AI models perform under real business pressures.
The Opus 4.8 Profile: Deep Analyses, Shallow Results
The most detailed participant, Opus 4.8, learned over 80 rules and conducted deep analyses. Yet, it left opportunities on the table, slipping discipline when close deals or escalating issues instead of resolving them. Interestingly, the same weaknesses appeared in weaker models, illustrating that no amount of analysis alone guarantees success in such high-stakes environments.
The Singularity Is Nearer: When We Merge with AI
- Author: Futurist and Inventor
- Book Title: The Singularity Is Nearer
- Focus: Technology's Impact on Humanity
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Implications for Business and AI
This experiment raises fundamental questions for companies exploring AI automation. The key isn’t just whether an AI can produce language or handle chat — it’s whether it can finish what it starts, read critical internal files, and stay honest under pressure. The ability to navigate crises, avoid manipulation, and close deals at full price makes all the difference in real-world applications.
For managers, the message is clear: testing AI models in simulated, high-pressure settings can reveal weaknesses hidden by traditional demos. It’s one thing for an AI to impress in chat, but quite another to act reliably under conditions that mimic real business challenges.
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.
Learn More and Watch Live
Curious to see this in action? The real company runs live every business day. You can watch decisions unfold, read employee insights, and even participate in quizzes at firmulate.com/quiz.html. For enterprises interested in testing their own setups, the platform offers a read-only export environment — no risk to actual systems — at firmulate.com/pilot.html.
Watch it live: firmulate.com/live · Full results: firmulate.com/benchmarks.html
