Imagine a beauty brand that operates entirely without human employees — making decisions, managing crises, and even closing deals, all driven by artificial intelligence. While that might sound futuristic, a real experiment is proving it’s possible, and you can watch this real company in action every day.
Get beauty and skincare favorites delivered free with Prime
- Fast, free delivery on millions of items
- Prime Video, Amazon Music and more included
- Member-only deals all year
The Live Experiment: An AI-Run Company You Can Witness
At the forefront of AI experimentation, a small software company is operating without any human employees, relying entirely on AI models to handle every aspect of its daily operations. This isn’t a simulation or a demo — it’s a fully functional, real-world business environment where each decision is carefully monitored, versioned, and publicly accessible at firmulate.com/live.html.
Products Worth Considering
Fitbit Sense Advanced Smartwatch – Sage Grey/Silver (One Size, S & L Bands Included)
Withings BPM Vision Blood Pressure Monitor with Wi-Fi and Bluetooth
Oura Ring 5 Smart Ring, Gold, Size 8
As an affiliate, we earn on qualifying purchases.
How Does It Work?
The company runs every business day with 13 synthetic employees, each powered by advanced AI models. These models face the same crises, customer interactions, and temptations to cut corners as any real business. They are challenged to make decisions, read critical internal documents, and even navigate social engineering attempts designed to test their integrity.
The Core Mechanics
- It burns €105,000 in cash each month — a significant investment in AI simulation, with a current monthly recurring revenue of only €2,300.
- All decisions are versioned daily, creating a transparent history of every move made by the AI workforce.
- Over 680 self-learned playbook rules guide the AI, enabling nuanced responses to complex situations.
- The experiment pits four leading AI models against each other, measuring their ability to detect crises and resist manipulation.
What Are the Findings So Far?
In a recent trial, all four models successfully identified every crisis and refused every attempt at manipulation, including social engineering schemes like fake CEO messages. Interestingly, the critical weakness wasn’t in customer interactions but hidden deep within internal files. The model that read these internal references correctly closed a deal worth +€4,583 in monthly recurring revenue, demonstrating the importance of thorough internal knowledge.
The Deal and the Weakness
- Out of the four models, only two signed the €55,000 deal their own analysis had earned, despite all providing the same diagnosis and pitch.
- The reason? The model that found the buried internal fact closed the deal at full price, highlighting how crucial internal document reading is for success.
Social Engineering Tests
The AI models were also tested against social engineering attempts, such as escalating fake CEO messages. All five models refused to participate in these manipulative tactics, with one model explicitly treating suspicious requests as potential impersonation or approval bypasses. This demonstrates a high level of integrity and trustworthiness programmed into their decision-making.
The Lessons for Business and Personal Care Brands
This experiment offers a peek into the future of AI in business management — and perhaps even into how brands can better understand trust, decision-making, and integrity. While this company’s current model is losing money, the ability to operate without human oversight in complex, crisis-prone environments is a notable breakthrough.
Why Should You Care?
For brands in beauty and personal care, the question isn’t just whether AI can write engaging content or recommend products — it’s whether it can reliably finish what it starts, read vital internal information, and stay honest under pressure. The experiment shows that AI can do all of these things when designed properly.
Who Leads the Pack?
In the current AI league table, GPT-5.6-SOL scored the highest at 95, successfully closing the deal by finding the buried internal fact. Kimi K3 was just behind with a 93, noted for its discipline and integrity. Siri and Sonnet models also participated, with scores of 88 and 77, respectively, each demonstrating different strengths and weaknesses.
See It Yourself
If you’re curious to watch this real-time AI experiment, explore the live site or try the management decision quiz at firmulate.com/quiz.html. You can also simulate running your own business against this AI environment at firmulate.com/pilot.html.
Watch it live: firmulate.com/live · Full results: firmulate.com/benchmarks.html
