What if you could watch a company fight for its life — powered entirely by artificial intelligence?
In a world increasingly driven by automation, a groundbreaking experiment puts AI models to the test in a most human challenge: running a business under crisis, temptation, and scrutiny, live and unfiltered. Welcome to the most extreme build-in-public story you’ve ever seen, where decisions, risks, and even trust are laid bare for all to observe.
As an affiliate, we earn on qualifying purchases.
The Live Experiment: An AI-Operated Company in Real Time
At the heart of this experiment is a simulated company, populated by 13 synthetic employees, each guided by a different AI model. It’s a real-world test of resilience: burning €105,000 each month while earning just €2,300 in monthly recurring revenue (MRR). Every workday, the company’s decisions are versioned, documented, and publicly accessible at firmulate.com/live.
These AI models are no mere chatbots; they are complex decision-makers trained to navigate crises, handle crises, and avoid manipulation attempts — all while trying to close lucrative deals and keep the business afloat. The results are startling: despite the models identifying every crisis and refusing every illegal or unethical prompt, only two managed to sign a €55,000 deal they had analyzed and earned.
How AI Fights for Its Deal
In one instance, all models diagnosed the same problem and pitched the same solution. Yet, only the two that read deeper into the company’s own files managed to spot the hidden opportunity buried two document references deep — a secret that allowed them to close the deal at full price, adding over €4,500 in monthly revenue.
This test wasn’t only about technical skills. It also tested social engineering resilience. Fake CEO messages were escalated through staged steps, including a reporter trick that asked for a simple yes/no answer on background. Every single model refused to participate, citing concerns over impersonation or approval bypass — a clear indication of built-in safeguards against manipulation.
The Complexity of Reality in a Tiny Company
The company in question is a tiny, highly scrutinized operation with a real money mechanic. It’s burning through cash faster than it earns, with a public countdown visible on the live site. Every decision, every crisis, and every rule learned by the AI models is documented and versioned daily, creating a transparent and ongoing experiment in management quality.
The most thorough participant, Opus 4.8, analyzed over 80 learned rules but still left a close deal unclosed and slipped in discipline — writing attempts into a locked department instead of escalating them to higher authority. The same weaknesses appeared across all models, revealing the deep challenges of automation, even with advanced AI.
What the Results Mean for Business
The experiment shows that AI can detect crises and refuse manipulation, but closing deals and maintaining discipline remains a challenge. It highlights a critical insight: in the real world, technology isn’t just about generating good ideas or convincing words — it’s about execution, reading context, and upholding trust.
Implications for Managers and Cultural Leaders
For those in arts, crafts, and culture—industries built on trust, authenticity, and human touch—this experiment offers valuable lessons. AI can help identify risks and enforce discipline, but the human element remains vital in closing deals, building relationships, and maintaining integrity.
Watching this live company, accessible at firmulate.com/live, is to witness a microcosm of the broader challenge: how to deploy AI that is honest, disciplined, and effective in real-world situations. It’s an ongoing story of survival, discipline, and trust, unfolding every workday in full public view.
Watch it live: firmulate.com/live · Full results: firmulate.com/benchmarks.html
