Imagine a beauty salon trying to stay afloat—yet this time, it’s an entire company run solely by artificial intelligence. With every decision, crisis, and temptation visible to the public, this experiment reveals what it truly takes for AI to manage a business under pressure.
The live experiment: AI managing a company in real time
At firmulate.com/live, a groundbreaking live experiment is unfolding. A company with no employees and a handful of synthetic staff is operating every workday, dealing with the same crises and temptations a real business faces. Every decision is versioned, auditable, and accessible for viewers eager to see if AI can truly run a company.
The setup: hard tests for AI models
Four frontier AI models, including GPT-5.6-sol and Kimi K3, are put through the same intense week—dealing with customer issues, internal crises, and ethical dilemmas. The goal? See if they can identify problems, resist manipulation, and close deals without succumbing to shortcuts or deception.
What the models achieved
- All four models spotted every crisis and refused every manipulation attempt.
- Only two completed the toughest task—signing a €55,000 deal their own analysis earned them.
- One critical weakness was hidden deep in the company’s internal files—those who read the files won the deal at +€4,583 MRR.
Real-world pressures and ethical tests
The experiment also tested social engineering—fake CEO messages and reporter tricks. Every model correctly refused to approve suspicious requests, demonstrating a high level of ethical discipline, especially Kimi K3, which explicitly treated such requests as potential impersonation.
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.
Why it matters for the beauty industry and beyond
This experiment isn’t just about AI or software companies. It lays bare a crucial question: when AI touches your customer relationship management (CRM), support, or forecasting, can you trust it to finish what it starts? Will it read your internal documents, resist deceptive tactics, and stay honest under pressure?
The financial strain and ongoing challenge
The company in question burns €105,000 every month but only earns €2,300 in monthly recurring revenue. With a public cash countdown, every decision is a matter of life or death—highlighting the stakes when AI is entrusted with core business functions.
The performance league: who’s leading?
- gpt-5.6-sol scored 95 and found the critical hidden fact—closing the deal at full price.
- Kimi K3 scored 93, showing the cleanest discipline and also sealing the deal.
- Sonnet 5 scored 88, with minor slips but still closing the deal.
- Opus 4.8 scored 73, leaving the deal on the table due to process slips.
This ranking reveals that even the most thorough AI models can struggle with discipline and process fidelity under real-world pressures.
AI Agents for Customer Support: A Practical Build Guide with RAG, Tools, and Guardrails (Reference Architecture + Prompts + Failure Playbook)
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Building in public—extreme transparency
The entire experiment is visible, versioned daily, and accessible at firmulate.com/live. It’s a showcase of AI’s capabilities and limitations—an open window into what managing a business with AI might look like.
Key takeaways
- AI can detect crises and refuse unethical manipulations even in stressful situations.
- Success depends on more than just diagnosis; process discipline matters.
- The deepest weaknesses lie in internal data, not customer interactions.
- Current models are capable but not infallible—real-world application requires careful oversight.
Responsible AI: Implement an Ethical Approach in your Organization
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
What does this mean for your business?
For industries like beauty and personal care, where customer trust and operational integrity are paramount, understanding AI’s true capabilities is vital. It’s not just about how well an AI writes or responds in a demo. It’s whether it can sustain honest, disciplined work amid real tensions and temptations.
To explore this further, you can try the management decision quiz or run your own business wargame, observing how AI handles your specific challenges without risking your actual systems.
Watch it live: firmulate.com/live · Full results: firmulate.com/benchmarks.html
AI: THE PERPETUAL INTERN – Its Brilliance and Failures Share the Same Root
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
