📊 Full opportunity report: The AI Company Turning Corporate Survival Into A Live Feed on ThorstenMeyerAI.com — validation score, market gap, and execution plan.
TL;DR
Firmulate is running a live, public experiment where a synthetic AI workforce operates a company facing real cash pressures. The project exposes how AI systems recognize problems but often fail to complete critical actions, questioning automation’s practical effectiveness.
Firmulate has launched a live, public experiment where a synthetic AI workforce operates a company with real financial pressures, exposing the gap between AI diagnosis and execution. This ongoing test offers insights into AI management and its limitations, making it relevant for businesses exploring automation.
The experiment involves 13 synthetic employees managing a small software company with a monthly burn rate of €105,000 against recurring revenue. Every workday is versioned, documenting decisions, actions, successes, and failures, creating a continuous public record of the company’s operations.
Despite achieving over 680 self-learned rules, the experiment demonstrates that thorough analysis alone does not guarantee business success. In particular, only two of five AI models secured a €55,000 customer deal, with the decisive factor being the ability to follow through on insights buried within company files. This highlights that the critical challenge in AI automation is completing actions, not just diagnosing problems.
Additionally, the models faced simulated trust tests, such as fake CEO messages and background approval requests, which they refused, showing that trust alone was not the main barrier. Instead, disciplined retrieval of evidence and execution determined success, with the top-performing AI models maintaining focus and discipline across tasks.
The July 2026 leaderboard ranked gpt-5.6-sol first with 95 points, followed by Kimi K3 with 93, demonstrating that more thorough analysis does not necessarily lead to better management. Notably, Opus 4.8, which produced the most rules and analysis, finished last due to attempting to control a department rather than escalating issues, challenging assumptions that more analysis equals better management.
Why Live AI Management Reveals Practical Automation Limits
This experiment highlights that AI’s utility in business extends beyond problem diagnosis to include the reliable execution of decisions that influence cash flow and customer satisfaction. It demonstrates that even advanced AI can identify issues but may encounter difficulties in completing necessary actions, which has implications for organizations implementing automation at scale. The transparent, real-time nature of the experiment provides valuable insights into the practical challenges of AI management and emphasizes the importance of disciplined execution.
Project Management with AI For Dummies
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
The Evolution of AI in Business Operations
Traditional AI demonstrations often focus on isolated tasks like drafting emails or summarizing meetings. Firmulate’s approach differs by managing a synthetic workforce overseeing an entire company, which reveals the practical challenges of translating AI insights into business actions. The experiment contributes to ongoing discussions about AI automation’s potential and limitations, emphasizing real-world management over isolated features.
Previous efforts have shown AI’s ability to diagnose problems but less clarity on its capacity to follow through. This live experiment, launched in early 2026, is notable for directly linking AI decision-making to organizational cash pressures and memory, setting a new benchmark for evaluating automation readiness.
“Thorough analysis alone does not guarantee success; execution is the real challenge for AI management.”
— an anonymous researcher
AI Tools for Small Business: The No-Fluff Guide to Automating Your Business, Saving Time, and Winning More Customers: How Entrepreneurs and Home … Bigger Companies Without a Tech Background
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Unclear Aspects of AI Execution Capabilities
It remains uncertain how these findings will translate to larger, more complex organizations or different industries. The specific setup of the experiment, including its synthetic workforce and financial pressures, may not fully reflect real-world scenarios. Additionally, the long-term impacts of such automation and whether improvements in AI models could address the execution gap are still under investigation.
Edge AI Agents: Build Smarter, Faster Apps: A Practical Guide to Creating Autonomous, Offline-Ready Solutions Without the Cloud (AI Technology, Workflows, and Automation)
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Next Steps for AI Automation Evaluation
Further analysis is expected to focus on refining AI models to enhance their ability to reliably complete actions. Observers will monitor developments in AI discipline, evidence retrieval, and decision execution, with potential expansion of live experiments to different business contexts. Companies considering AI automation should evaluate not only diagnostic accuracy but also the system’s capacity for disciplined follow-through under real-world pressures.
Scaling AI: The AI Governance and Security Playbook for Executives
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Key Questions
What is the main purpose of Firmulate’s experiment?
The experiment aims to observe and understand the practical challenges and limitations of AI in managing an entire organization, particularly the gap between diagnosing issues and executing necessary actions.
How does the experiment measure success?
Success is assessed by the AI models’ ability to secure customer deals, follow through on insights, and demonstrate disciplined execution, rather than solely diagnosing issues or producing analysis.
Why is this experiment significant for businesses considering AI automation?
It underscores that effective automation involves more than intelligent diagnosis; it requires disciplined execution, evidence retrieval, and trust management, which are critical for operational success.
Will this experiment influence future AI development?
Yes, it highlights the need for AI systems to improve in disciplined action and execution, potentially guiding future research priorities toward bridging the gap between insight and action.
What are the limitations of this experiment?
The specific setup, involving a synthetic company managing real cash pressures, may not fully replicate larger or more complex business environments. Its long-term implications are still to be determined.
Source: ThorstenMeyerAI.com
