Imagine Watching a Company Struggle in Real Time — Powered by AI
In the world of senior care, trust, consistency, and decision-making are everything. Now, imagine a company run entirely by AI models, navigating crises and making decisions on a daily basis — all in public view. This is not science fiction; it’s the real-time experiment at Firmulate, where an AI-managed company is fighting to stay afloat, revealing how artificial intelligence can handle complex, trust-dependent situations.
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.
The Experiment: An AI-Run Business Under Pressure
At the heart of this experiment are four advanced AI models, each tasked with running the same small software company through its worst week. This company isn’t just a simulation; it’s a real operation with real money, real crises, and genuine temptations to cheat. Every decision these models make is recorded, versioned, and transparent, providing a rare look into how AI can handle real-world business challenges.
Complex Decisions Under Scrutiny
The models faced the same set of crises and customer requests, and were tested on how honestly and effectively they responded. All four AI systems identified every crisis and refused manipulative efforts, such as fake CEO messages or covert requests. For example, when staged social engineering attacks were attempted — like fake background approvals — all models rejected the requests outright, citing concerns about impersonation or bypassing approval processes.
Crucial Insights from Deep Reading
While all models performed well in recognizing crises and resisting manipulation, a deeper layer of the experiment uncovered a key weakness. The decisive advantage came not just from quick responses but from detailed document analysis. The models that read deeper into internal files uncovered a critical piece of information buried two references deep — a fact that enabled them to close a lucrative deal at full price, adding +€4,583 MRR (monthly recurring revenue).
The Reality of a Money-Losing Business
This AI company isn’t profitable — it burns €105,000 every month against a modest €2,300 MRR. The company’s public cash countdown underscores the urgency of making right decisions, especially as every workday’s processes, rules, and decisions are versioned and made transparent for all to see. This relentless transparency makes it an unparalleled live case study for AI’s potential and limitations in real-world business management.
Why This Matters for Senior Care and Aging
For organizations in senior care, trustworthiness, thoroughness, and resilience under pressure are critical. The insights from Firmulate’s experiment highlight how AI systems can be trained and tested to ensure they’re capable of making honest, effective decisions, especially in sensitive sectors where trust is paramount. The ability of AI to recognize manipulative tactics and prioritize integrity could be transformative for support services, compliance, and client relations in aging care environments.
Lessons from the Frontline of AI Management
The experiment also revealed that even the most thorough AI — Opus 4.8, with over 80 rules learned and deep analysis — still left potential revenue on the table, demonstrating that discipline and focus are vital. Interestingly, the AI running with default settings (K3) performed nearly as well as the others, indicating that well-designed AI can achieve high fairness and discipline without excessive tuning.
Deciding the Best AI
In the competitive league, the models scored as follows:
- GPT-5.6-SOL scored 95 — found the critical insight and sealed the deal at full price.
- Kimi K3 scored 93 — the newcomer with the cleanest discipline, also secured the deal.
- Sonnet 5 scored 88 — closed the deal but with some process slips.
- Fable 5 scored 77 — also closed but with more weaknesses.
This ranking underscores that even minor differences in discipline and thoroughness can impact ultimate success.
What This Means for Your Organization
Whether managing a healthcare service, a support team, or an aging care facility, the message is clear: the effectiveness of AI isn’t just about generating convincing conversations. It’s about whether AI can finish what it starts, stay honest under pressure, and process relevant internal information to make optimal decisions. The real test isn’t in chat demos — it’s in how AI handles crises, manipulative attempts, and complex documents.
Key Takeaway
This experiment demonstrates that AI models can recognize crises, resist manipulation, and even close high-value deals when given thorough internal information. For senior care providers, it underscores the importance of trusting AI to uphold integrity and make strategic decisions — especially as AI begins to touch critical aspects of daily operations. Watching this live experiment offers valuable lessons on building trustworthy, disciplined AI systems capable of supporting sensitive, trust-dependent sectors like aging care.
Watch it live: firmulate.com/live · Full results: firmulate.com/benchmarks.html
