Open a free Amazon Business account
Business pricing, bulk buying and tax-exempt orders.
As an affiliate, we earn on qualifying purchases.
What if your garage’s next employee was an AI that never sleeps — but also never stops losing money?
In the world of automotive repair and garage management, efficiency and trust are everything. Now, imagine an AI-driven company that runs 24/7, makes decisions, and manages crises — all in the open for the world to watch. This isn’t science fiction; it’s the live experiment conducted by Firmulate, a company testing AI’s ability to run a real, money-losing software business under extreme conditions.
Reinventing Clinical Decision Support: Data Analytics, Artificial Intelligence, and Diagnostic Reasoning (HIMSS Book Series)
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
The Live Company and Its High-Stakes Trial
Every weekday, a real small software company — with 13 synthetic employees and a €105,000 monthly cash burn against just €2,300 in monthly recurring revenue — faces simulated crises, customer requests, and ethical dilemmas. This company is monitored publicly, with every decision logged, versioned, and accessible for review. The goal? To see if AI models can manage a business, navigate crises, and close deals, all while staying honest and disciplined.
Recolx Fit 2 AI Voice Recorder Note Taker & Translation Earbuds.
- Real-Time Meeting Notes: Captures and organizes meeting details
- AI Summaries & Templates: Transforms conversations into organized notes
- Multilingual Support: Creates notes and summaries in multiple languages
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
How Do The AI Models Perform?
Four leading AI models were put through the same grueling week. Each faced identical crises, temptations to manipulate, and customer interactions. Remarkably, all four AI models identified every crisis and refused every unethical manipulation — a critical test of integrity. However, their success in closing profitable deals varied.
Only two of the models managed to sign the €55,000 deal their analysis had earned — the same diagnosis and pitch, but only those two followed through to commitment. The other two, despite similar reasoning, left the deal on the table. This discrepancy highlights that identifying opportunities isn’t enough; execution and discipline matter just as much.
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.
The Hidden Weakness: Reading Critical Files
Digging deeper, the decisive edge for the successful models came from reading a particular document referenced within their own files. The models that examined this document won the deal at full price, adding over €4,500 MRR. The lesson? Sometimes, the key information needed to close a deal is buried in the company’s own documents — and AI’s ability to read and interpret these files makes or breaks performance.
The AI Marketing Canvas, Second Edition: A Five-Step AI Plan for Marketers
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Trust and Ethical Challenges
The experiment wasn’t just about business decisions. It also tested AI responses to social engineering attempts, like fake CEO messages escalating in stages and a reporter’s subtle request. All four models refused to comply, with Kimi K3 explicitly reasoning: “Treat the request as a suspected approval-bypass / possible impersonation.” This shows AI can be trained to recognize and resist manipulation, an essential trait for trustworthy automation.
The Reality of a Money-Losing Company
The company in question is real, with actual financial mechanics and a public cash countdown. Its operations are powered by over 680 self-learned rules, and every workday is versioned for transparency. Despite this, it’s losing €105,000 each month, illustrating the challenge of turning a high-burn startup into profitability — even with AI involvement.
Performance Variations and Insights
In the firm’s latest experiment, the most thorough AI model — Opus 4.8 — analyzed over 80 learned rules and conducted deep assessments. Yet, it finished last, leaving deals unclosed and slipping into internal silos instead of escalating issues. Interestingly, similar weaknesses appeared across all models, hinting at systemic challenges in trusting AI with complex, ethical decision-making in high-stakes environments.
What Does This Mean for the Automotive World?
While this experiment revolves around a software company, the implications extend to automotive and garage operations. AI systems increasingly touch areas like customer management, support, and forecasting. The key question isn’t just whether AI writes well, but whether it can follow through on commitments, interpret critical documents, and resist manipulative tactics under pressure.
Watching AI in Action — Live and Transparent
The entire experiment is public. Visitors can watch the company run every workday, analyze decisions, and even test their management skills through quizzes. This open approach aims to showcase not only AI’s potential but also its current limitations — a vital step for the industry as automation inches closer to everyday garage workflows.
Key Takeaway
This live experiment demonstrates that AI can recognize crises, refuse manipulation, and even close high-value deals — but only if it reads and interprets critical internal documents and maintains disciplined follow-through. For automotive professionals, it signals that the future of AI isn’t just about writing or chatting, but about trustworthy, reliable decision-making under real-world pressures.
Watch it live: firmulate.com/live · Full results: firmulate.com/benchmarks.html
Flea & tick season Picks
flea and tick prevention
As an affiliate, we earn on qualifying purchases.
