Imagine watching four different actors attempt to save a struggling business — but only one actually manages to close the deal. In the world of AI, performance isn’t just about chatter; it’s about execution, integrity, and the ability to finish what it starts. That’s the surprising lesson from a real-world AI experiment that pits top models against each other in a simulated crisis — and the results could change how we evaluate AI’s true business value.
The Experiment: Putting AI Models to the Test
In a bold, transparent test, four leading AI models were tasked with running a small software company through its toughest week. This wasn’t a mere chat demo — it was a full-fledged, auditable simulation involving real money mechanics, customer crises, and the temptations to cut corners. The goal? To see which AI could diagnose problems, resist manipulation, and ultimately close a €55,000 deal earned through its own analysis.
Same Crisis, Different Outcomes
All four models identified every crisis and refused every attempt at manipulation, including fake CEO messages and reporters’ tricks. Yet only two models actually signed the deal. The others, despite understanding the issues, left the money on the table, unable or unwilling to follow through with the closing process.
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 Made the Difference?
The key was not just what the models read or how they diagnosed the problems — it was how they acted on this information. The decisive advantage belonged to models that read deeper into the company’s files, uncovering critical buried facts that weren’t obvious from surface-level analysis. These models, including one that scored 95 out of 100 in the final league, secured the deal by leveraging hidden knowledge, proving that true business effectiveness requires more than just good chat capabilities.
Beyond the Chat: Trust and Discipline Under Pressure
Interestingly, the model that was most thorough in analysis — Opus 4.8 — finished last in execution. Despite its depth, it slipped in discipline, leaving the signed deal unexecuted and writing attempts into a locked department instead of escalating. Meanwhile, Kimi K3, a newcomer with a clean discipline record, successfully closed the deal at full price. This highlights how crucial execution discipline and decision-making under pressure are — qualities that are invisible in typical chat-based evaluations.
Project Management with AI For Dummies
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
The Real-World Application
What does this mean for businesses considering AI augmentation? The experiment underscores a vital truth: the ability to produce impressive chat demos is not indicative of an AI’s capacity to finish real work, especially under pressure. Performance in a controlled conversation differs vastly from execution in a complex, money-driven environment.
The League Table and Implications
- gpt-5.6-sol scored highest (95), found critical buried facts, and closed the deal.
- Kimi K3 scored 93, also closing at full price with strong discipline.
- Sonnet 5 scored 88, closing the deal but with some process slips.
- Fable 5 scored 77, maintained discipline but failed to execute the deal.
- The baseline, representing partial progress, scored just 26.
This ranking reveals that the true measure of AI’s business competence lies in its ability to follow through, read deeper into data, and resist manipulations — qualities that are impossible to judge in static demos.
Databricks Data Intelligence Platform: Powering the Agentic Era
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
The Takeaway
For organizations, the message is clear: testing AI in real decision environments is essential. Chat demos can mislead; real performance hinges on whether AI systems can stay honest, read deeply, and execute reliably when it matters most. An AI that can close a deal under pressure isn’t just impressive — it’s indispensable.
To explore what your own AI workforce might truly accomplish, visit Firmulate and see the live experiment in action. Witness how real companies—like yours—perform when faced with their toughest week.
Watch it live: firmulate.com/live · Full results: firmulate.com/benchmarks.html
4 Disciplines of Execution
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
