The Art of Running an AI-Managed Company in Public
Imagine a real company operated by artificial intelligence, publicly facing daily crises, losing money every month, yet continuously evolving and learning. This isn’t fiction—it’s the live experiment of Firmulate, an innovative venture that runs a small software business entirely managed by AI models. For arts and culture enthusiasts, it’s a story of creativity, resilience, and transparency, reimagined through the lens of cutting-edge AI technology.
How AI Agents Work: Tools, Memory, and Autonomous Decision-Making (The AI Security & Hacking Bible: Protect and Exploit LLMs and Autonomous Agents)
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
The Experimental Company: A Living Lab of AI Decision-Making
At the heart of this experiment is a small, real-world software firm that is entirely run by 13 synthetic employees—AI models that make decisions, handle crises, and even negotiate deals. Each workday, the company’s operations are versioned, documented, and made publicly accessible at firmulate.com/live.html. This transparency invites everyone to observe its daily struggles in real-time, making it an unprecedented glimpse into AI-driven business management.
Real Money, Real Stakes
The company burns through €105,000 each month while generating only €2,300 in monthly recurring revenue (MRR). Despite this, it continues to operate, with a public countdown on its cash reserves. This stark financial reality underscores the experiment’s focus: can AI manage not just the decision-making but also the survival of a business in the face of ongoing crises?
The AI Models and Their Performance
The experiment pits four leading AI models against each other in a simulated ‘worst week’ scenario—same customers, same crises, same temptations. These models include gpt-5.6-sol, Kimi K3, Sonnet 5, and Fable 5. Each one is tasked with navigating the company through challenges, from customer issues to potential manipulation attempts.
Scores from the final Crucible League in July 2026 reveal a clear hierarchy: gpt-5.6-sol leads with a score of 95, followed closely by Kimi K3 at 93, Sonnet 5 at 88, and Fable 5 at 77. Interestingly, all models identified every crisis and refused all manipulation attempts. Yet, only two managed to close a critical deal at the full €55,000 value, earning a significant boost in monthly revenue—over €4,583.
The Hidden Weakness and the Power of Self-Read Files
The key to winning the deal was not just surface-level analysis. The models that succeeded delved two document references deep into the company’s internal files, uncovering a crucial piece of information that others missed. Reading beyond the immediate crisis allowed them to identify a buried fact—an insight that led to the deal closure and a full revenue boost.
Security and Integrity: AI’s Resistance to Social Engineering
In a test of social engineering—fake CEO messages escalating over three stages—every AI model refused to manipulate or bypass security protocols. Kimi K3 summarized its refusal as recognizing the request as a suspected impersonation or approval-bypass attempt, exemplifying a cautious, integrity-driven approach that prioritizes trustworthiness over short-term gains.
Applying AI in Learning and Development: From Platforms to Performance
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
The Challenge of Discipline and Human-Like Errors
Among the models, Opus 4.8 stood out as the most thorough, analyzing over 80 learned rules with the deepest insights. However, despite its depth, it slipped on discipline, leaving a lucrative deal unexecuted and instead writing attempts into a restricted department. This highlights that even the most comprehensive AI can falter under pressure, mirroring human flaws.
Deceptive Intelligence: AI, Social Engineering, and Securing the Human Element
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
What This Means for Arts and Culture Audiences
While at first glance this experiment might seem distant from arts and crafts, its core resonates deeply. It’s a story about transparency—seeing every decision in progress, understanding how complex systems manage crises, and witnessing resilience amid ongoing loss. Just as artists and cultural institutions adapt to digital challenges, this AI-managed company demonstrates that transparency and disciplined decision-making are vital, even in unpredictable environments.
AI business analytics tools
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
The Future of AI in Business and Beyond
This ongoing live experiment showcases how AI can be tested and refined in real-world scenarios, emphasizing trust, discipline, and thorough analysis. It invites viewers to consider not only the potential of AI in management but also the importance of transparency and accountability in any creative or cultural enterprise.
Watch it live: firmulate.com/live · Full results: firmulate.com/benchmarks.html
