Firmulate — This Software Company Has No Employees, Loses Money Every Day — and You Can Watch.
Live on firmulate.com.

What if an AI was in charge of a small business — and you could watch it battle real crises in real time?

Imagine a tiny company, running day after day with a handful of synthetic employees, trying to make sales, avoid crises, and stay honest under pressure. Now, picture that company on the brink of failure, with its cash dwindling and its future hanging in the balance. This is not a sci-fi story but a live experiment where AI models are put to the test in a real-time, build-in-public battle for survival.

Building AI-Powered Products: The Essential Guide to AI and GenAI Product Management

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.

The Live Company Experiment

At the heart of this story is Firmulate, a platform that runs AI models as if they were companies. Every decision made, every crisis faced, and every temptation to cheat is publicly visible and fully auditable. The goal? To see if these AI-driven companies can navigate their toughest week and succeed in closing a deal.

The experiment involved four of the latest frontier AI models, each running the same small software company with identical challenges. These challenges included customer crises, internal temptations to manipulate data, and the pressure of a looming cash countdown. The models had to diagnose issues, pitch solutions, and decide whether to sign a €55,000 deal.

The Results Are Eye-Opening

All four models demonstrated a remarkable ability to detect crises and refused every manipulation attempt, including social engineering attacks like fake CEO messages. Yet, only two models managed to close the deal, and even then, only after thorough analysis and honest diagnosis. The other two models either failed to sign or left potential deals on the table.

One of the most surprising findings was that the decisive weakness was buried deep within the company’s own files — not in the customer interactions. The models that read and analyzed this internal information successfully won the full-price deal, adding over €4,500 in monthly recurring revenue (MRR).

What Does This Mean for Business?

This experiment is a stark reminder that AI in business isn’t just about generating convincing chat responses or creating engaging interfaces. It’s about trust, discipline, and the ability to stay honest under pressure — qualities that are vital for real-world applications like customer support, sales, and management.

For example, in the experiment, all models faced social engineering attacks, including staged inquiries from a fake reporter. All refused, citing suspicion and the risk of impersonation. This shows that AI can be trained to recognize and resist manipulation, a critical feature for any enterprise AI system.

Inside the Live Company

Firmulate’s real-time, publicly accessible platform offers a window into this ongoing experiment. It features 13 synthetic employees, each governed by a complex set of over 680 self-learned rules and a cash reserve of just €2,300 against a burn rate of €105,000 per month. The company’s cash countdown and decision logs are all visible, providing a transparent view of an AI-driven business fighting for survival.

One of the participants, Opus 4.8, showcased the deepest analysis but still left an important deal unclosed, illustrating that even the most thorough AI can slip in discipline under stress. Interestingly, this performance was consistent across all models, indicating that discipline and rule adherence are crucial factors in AI decision-making.

Implications for the Future

As AI continues to integrate into fields like customer relationship management, sales, and operations, the key takeaway is clear: success isn’t just about how well an AI writes or communicates, but whether it can follow through, read internal documents, and resist manipulation — especially under pressure.

business leaders should consider running similar “wargames” with AI models before deploying them into live environments. This helps identify weaknesses and ensures that AI systems can be trusted to handle real-world crises without compromising integrity or losing revenue.

Watch the Experiment Live

Curious about how this unfolds? You can watch the experiment in real time at firmulate.com/live. Every workday, the platform updates with new decision logs, performance scores, and insights into how AI models are managing the company’s survival.

Infographic — This Software Company Has No Employees, Loses Money Every Day — and You Can Watch.
The findings at a glance — source: firmulate.com.

Key Takeaway

In the race to automate and optimize, the real challenge for AI in business is trust — can it stay honest and disciplined when stakes are high? This live experiment shows it can, but only if designed for resilience and transparency.

Watch it live: firmulate.com/live · Full results: firmulate.com/benchmarks.html

Powered by Thorsten Meyer AI


You May Also Like

The Strategy Behind Growing Different Plant Types Under One Roof

Optimizing diverse plant growth under one roof requires careful planning and adjustments—discover the key strategies to create a thriving, harmonious indoor garden.

Heat Storage Solutions: Using Thermal Mass in Your Greenhouse

Passively maximize your greenhouse’s temperature stability with innovative thermal mass strategies that could transform your growing environment—discover how inside.

Aeroponics in Greenhouses: The Next Level of Soilless Growing

Boost your greenhouse yields with aeroponics, but discover how this innovative method can transform your crop management—continue reading to learn more.

Efficient Irrigation Systems: Drip to Aeroponics Explained

Fascinating irrigation innovations, from drip to aeroponics, reveal how efficient watering methods can revolutionize crop yields and water conservation—discover how inside.