Can AI Run a Company Without Employees — and Survive the Worst Week?

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

Imagine a small startup without any human employees, running entirely on artificial intelligence, yet facing the same crises and temptations as any traditional business. Now, picture watching that company struggle, fail, and even win deals — all in real time. This is not science fiction; it’s the groundbreaking experiment happening now, and it offers a fresh look at AI’s true potential in the business world.

Table of Contents

The Live Experiment: An AI-Run Business Under Pressure

At the heart of this bold experiment is a real-time, publicly accessible AI-driven company hosted at firmulate.com/live.html. This virtual enterprise is staffed by 13 synthetic employees, each governed by self-learned rules and decision protocols. Every workday, the company’s operations are versioned and transparent, exposing how AI manages crises, makes decisions, and attempts to close deals — all while burning €105,000 in cash against a modest €2,300 monthly recurring revenue (MRR).

What makes this experiment truly extraordinary is that four leading AI frontier models are subjected to the same challenging week in the company’s life, facing identical customers, crises, and temptations. The goal? To see how well these models handle real-world pressures and whether they can stick to honest, effective business practices.

The Results: Successes and Failures in AI Business Management

All four models identified every crisis, demonstrating excellent situational awareness.
All refused manipulation attempts, including social engineering tricks like fake CEO messages and reporter baiting.
Only two models managed to sign the €55,000 deal their own analysis had earned them — translating to a €4,583 boost in monthly recurring revenue.
The critical weakness was hidden deep within the company’s files, not in the customer interactions. The models that read these files thoroughly secured the deal at full price.
The most thorough model, Opus 4.8, analyzed deeply but slipped in discipline, leaving deals unexecuted and escalating conflicts into departmental silos.

The Broader Implications: What This Means for the Future of AI in Business

This experiment underscores a vital point: effective AI management isn’t just about writing convincing chat responses. It’s about whether AI can finish what it starts, read critical internal documents, resist manipulative tactics, and make disciplined decisions under pressure. For businesses integrating AI into CRM, support, or forecasting, these are the real tests of usefulness and trustworthiness.

The performance scores, from highest to lowest, highlight differing levels of robustness: GPT-5.6 scored 95, Kimi K3 scored 93, Sonnet 5 scored 88, and Fable 5 scored 77. Notably, Kimi K3, despite being a newcomer, displayed the cleanest discipline and successfully closed the deal, even without default effort parameters.

Built-in Public: Transparency and Continuous Progress

Every decision, every rule learned, and every version update is publicly recorded and accessible. The live site not only demonstrates ongoing results but invites observers to witness the evolution of AI decision-making in real-time. Visitors can also participate in a quiz, testing their ability to distinguish model decisions, and even run their own business wargames against exported data — all without risking real systems.

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

As an affiliate, we earn on qualifying purchases.

The Significance for the Future of Work

This experiment pushes the boundaries of transparency and accountability in AI deployment. It shows that AI can handle complex, messy situations and make disciplined choices, even without human oversight. For industries like electric vehicles or micromobility — where AI might soon optimize logistics, customer service, or maintenance — understanding these capabilities is crucial.

As the experiment continues and the league of models grows, it offers a glimpse into how AI might someday manage critical business functions while you watch, learn, and prepare for the next wave of automation.

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

This real-time AI business experiment reveals how AI models handle crises, make decisions, and stay honest under pressure — essential insights for future automation in every industry, including mobility and EVs.

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