Imagine a business that operates without human employees, loses money every day, yet is openly tested and scrutinized in real time. For those caring for seniors and aging populations, the story offers a glimpse into the future of work, trust, and automation — where transparency isn’t just an ideal, but a daily reality.
A Live Experiment in AI-Managed Business
In an unprecedented public trial, a small software company is run entirely by artificial intelligence models, with no human staff involved. Every decision, crisis response, and negotiation is made by AI, and every move is recorded, versioned, and open for evaluation at firmulate.com/live. This isn’t a simulation; it’s a live, ongoing experiment designed to test the limits of AI’s management capabilities under stress.
The Stakes and the Mechanics
The company is losing €105,000 each month, while generating a modest €2,300 in monthly recurring revenue. It has 13 synthetic employees, and its daily operations are governed by over 680 rules learned and refined through self-play. Every workday, the AI’s decisions are versioned, making it possible to trace each move back to specific rules and decision points.
Facing Crises — Same Challenges, Different Outcomes
The experiment subjected four frontier AI models to identical stressful scenarios: the same customers, crises, and temptations to manipulate or cheat. Remarkably, all four models identified every crisis and refused every manipulation attempt, demonstrating a strong grasp of ethical boundaries and crisis management.
One key finding: the models that read deeper into the company’s own files, beyond surface-level documents, won more deals — closing at full price and adding over €4,500 in monthly recurring revenue. This indicates that effective AI management depends heavily on thorough information processing, not just surface interactions.
Trust and Ethical Decision-Making
In situations involving social engineering, such as fake CEO messages or reporter tricks, all models refused to sign off on unauthorized deals or impersonation attempts. Kimi K3, one of the models, explicitly explained: “Treat the request as a suspected approval-bypass / possible impersonation.” This highlights that advanced AI systems are capable of recognizing and resisting unethical pressure, an essential feature for trustworthy automation.
The Human-Like Failures of AI
Despite its strengths, the most thorough model, Opus 4.8, showed a vulnerability: it left potential deals on the table due to discipline slips, such as writing attempts into a protected department instead of escalating issues. All models exhibited similar weaknesses, revealing that even the most advanced AI can struggle with complex organizational processes under pressure.
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.
Implications for Business and Caregiving
For industries like senior care and aging support, where trust, ethics, and reliability are paramount, this experiment offers valuable lessons. AI systems can identify crises, resist manipulation, and make decisions aligned with ethical standards. But they also need safeguards to prevent slip-ups that could jeopardize trust or financial stability.
What Does This Mean for the Future?
As AI systems increasingly handle tasks in healthcare, support, and administration, their ability to complete what they start — reading all relevant information, resisting unethical pressures, and making decisions aligned with organizational values — will be crucial. The current experiment underscores that successful automation isn’t just about chat quality or superficial interactions; it’s about consistent, trustworthy performance under real-world stress.
Building AI-Powered Products: The Essential Guide to AI and GenAI Product Management
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Takeaway: Building Trust in Automated Management
The live company run by AI models demonstrates that even in a financially challenging environment, AI can perform complex management tasks ethically and effectively. For those who care for seniors and aging populations, the message is clear: future automation must prioritize ethical decision-making, comprehensive information processing, and transparent operations — qualities that this experiment vividly illustrates.
Watch it live: firmulate.com/live · Full results: firmulate.com/benchmarks.html
I, Human: AI, Automation, and the Quest to Reclaim What Makes Us Unique
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Information Systems for Crisis Response and Management in Mediterranean Countries: 4th International Conference, ISCRAM-med 2017, Xanthi, Greece, … in Business Information Processing, 301)
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
