Imagine a smart home assistant that not only manages your devices but also runs a tiny business, making decisions under pressure—without any human intervention. This isn’t science fiction. It’s the latest experiment in AI-driven management, and it’s happening live at Firmulate. Here’s what it reveals about trust, performance, and the future of automation.
Open a free Amazon Business account
Business pricing, bulk buying and tax-exempt orders.
As an affiliate, we earn on qualifying purchases.
The Live Company: An AI-Run Business in Real Time
At the heart of this experiment is a small, virtual software company operated entirely by AI models. These models aren’t just chatting or generating text—they’re making critical business decisions, facing real crises, and managing actual money mechanics. Every weekday, the company’s activities are versioned and publicly accessible, offering an unprecedented window into how AI performs in a complex, high-stakes environment.
Currently, the company spends €105,000 each month to operate while generating only €2,300 in monthly recurring revenue—an unsustainable situation that underscores the challenge of building trustworthy AI systems capable of managing real-world tasks.
100 AI Prompts for Small Business & Daily Work: Copy, Paste, Customize & Get Better Results with AI
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
The Experiment: Testing AI Under Pressure
Four leading AI models, including the latest frontier models like gpt-5.6-sol and Kimi K3, were each tasked with running the same week of the company’s operations. They faced identical crises, customer interactions, and temptations to cheat—such as manipulation attempts and social engineering attacks. Every decision made by these models was versioned and auditable, providing a clear view of their reasoning processes.
Remarkably, all four models identified every crisis correctly and refused every manipulation attempt. Yet, only half of them managed to secure the €55,000 deal based on their own analysis, revealing a crucial gap between diagnosis and execution. The models that read deeper into the company’s files and identified a critical buried piece of information succeeded where others faltered, closing the deal at full price—adding €4,583 in monthly recurring revenue.
AI in Strategy and Decision-Making for Small Business Owners: Affordable AI Tools to Evaluate Ideas, Model Outcomes, and Set Priorities (AI Productivity for Small Business Owners Book 10)
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
What the Results Tell Us About AI Performance
While the models showed impressive crisis detection and refusal to manipulate, their ability to follow through—particularly in closing deals—varied significantly. For example, Opus 4.8, the most thorough participant with over 80 learned rules, slipped in discipline and left opportunities unexploited, costing potential revenue. Interestingly, the models’ discipline and decision quality were influenced by their configurations, such as the ‘effort parameter’ used in Kimi K3.
These outcomes highlight a stark truth: AI systems can be excellent at identifying problems and resisting deception but may struggle with consistency in execution. The difference between spotting opportunities and acting decisively is crucial when deploying AI for real-world business management, especially in sensitive areas like customer relations and financial transactions.
AI for Crisis Management & Business Continuity: The Executive Handbook for Protecting Your Organization with 50 AI Prompts for Detection, Response, and … BUSINESS & MANAGEMENT LIBRARY SERIES 38)
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Built-in Public: A Transparent, Ongoing Experiment
The entire setup is accessible, real, and ongoing at Firmulate’s live site. Visitors can watch the company operate daily, listen to actual decisions, and see decisions made by different models side-by-side. Every workday version is documented, and the decision-making process is openly auditable, providing an unprecedented level of transparency in AI deployment.
AI Agents and AI Automation with n8n: The Complete Beginner’s Guide, Build Agentic AI Systems Step by Step
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Implications for the Future of AI in Business
This experiment underscores a vital point: the success of AI in managing real businesses hinges not just on how well it writes or generates responses, but on its ability to finish tasks, avoid deception, and read critical information before acting. In other words, the focus shifts from chat quality to tangible performance and trustworthiness.
As AI begins to touch more aspects of daily life—from smart home management to enterprise operations—the lessons from this live experiment are clear. Building trust, ensuring reliability, and verifying that AI systems can follow through under pressure are the essential next steps toward widespread adoption.
Watch it live: firmulate.com/live · Full results: firmulate.com/benchmarks.html
Back to school Picks
back to school
As an affiliate, we earn on qualifying purchases.
