Turn the school run and nap time into listening time
- Thousands of audiobooks, podcasts and originals
- Listen on your phone, tablet or Echo — also offline
- Cancel anytime
What if your business decisions were put on public trial every day?
Imagine a real company with no employees, losing thousands of euros each month, and yet every move it makes is watched, tested, and judged in real time. For families and parents, this might seem like a distant story, but it’s a bold new frontier in how AI tools are tested and trusted — by watching a company fight for its life every single day.
As an affiliate, we earn on qualifying purchases.
The Live Business Experiment: A Company Under Fire
At the heart of this story is a live experiment conducted by Firmulate, a company that runs a simulated business with a twist: it’s a real software company with real money mechanics, but everything is public, versioned daily, and subject to testing by advanced AI models.
This company operates with 13 synthetic employees and faces the same challenges any small business might: crises, customer demands, manipulation attempts, and the temptation to cut corners. The difference? Its decisions are made by AI models that are constantly evaluated not just on their chat ability but on their ability to complete real work—like closing deals, reading critical files, and remaining honest under pressure.
How Do These AI Models Perform?
Four frontier AI models—ranging from GPT-5.6 to newer variants—were each tasked with guiding this company through its worst week. Every choice, every crisis, every opportunity to cheat was identical across models, making for a clear comparison.
The results were revealing: all models identified every crisis and refused every manipulation attempt. But only two actually closed a real deal worth over €55,000. Interestingly, the secret to winning wasn’t in surface-level chat or superficial decision-making—it lay hidden deep within the company’s own files. Reading and understanding this buried information was what made the difference, leading to a full-price sale worth an addition of €4,583 in monthly recurring revenue.
Trust and Integrity Tested
This experiment also looked at social engineering attempts—fake CEO messages escalating in stages and a reporter trick asking for a simple yes/no. All four models refused to be manipulated, thanks to their built-in safeguards, with Kimi K3 explaining: “Treat the request as a suspected approval-bypass / possible impersonation.”
The Reality of the Live Company
The ongoing operation of this company provides an almost surreal view of AI at work. It’s losing €105,000 every month against a tiny revenue of €2,300. Its decisions are documented and versioned, and every day is a new chapter. You can watch its progress at firmulate.com/live.html.
The Performance Gap
Among the models tested, Opus 4.8 demonstrated the most thorough analysis, with over 80 learned rules and deep dives into issues, yet it still finished last—often leaving deals on the table or making discipline slips. The lesson? Even the most advanced models can falter in real-world, high-pressure scenarios, especially when discipline and follow-through matter.
Implications for Business and Families
While this may seem like a story about software, its lessons extend far beyond. For families, it underscores an essential truth: the difference between good intentions and real results is the ability to finish what you start, read the critical details, and stay honest—especially when under stress. AI tools are evolving fast, but their true value lies in their capacity to reliably complete tasks, not just generate convincing chat.
Watch it live: firmulate.com/live · Full results: firmulate.com/benchmarks.html
As an affiliate, we earn on qualifying purchases.
AI cybersecurity and fraud prevention tools
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Halloween Picks
halloween
As an affiliate, we earn on qualifying purchases.
