Can AI Finish What It Starts? Lessons from a Live Business Wargame

Firmulate — Four AI Models Ran the Same Company Through Its Worst Week. Only Two Finished the Job. — Live on firmulate.com.

Imagine if your favorite ice cream shop could run a simulated week of chaos—crises, customer tantrums, and temptations to cut corners—just to see if its AI helpers could handle the heat. Turns out, in the world of business, not all AI models are created equal when it comes to following through. The same lesson applies whether it’s managing a bakery or a tech company: real capability isn’t in the chat, but in the gritty ability to see tasks through, especially when temptation strikes.

The Experiment: Putting AI to the Test in a Live Company

In a groundbreaking experiment, four advanced AI models were tasked with running a real, small software company through its worst week. The goal? To see if these AI ‘managers’ could spot crises, resist manipulation, and ultimately close a crucial deal worth €55,000. Importantly, every decision was recorded and auditable, making this a unique test of actual management skills—not just chat prowess.

The Players and Scores

gpt-5.6-sol 95: Nailed the buried fact, closed the deal—the full package of performance.
Kimi K3 93: The newcomer delivered the best discipline, closing the deal with a clean record.
Sonnet 5 88: Also closed the deal but with some slips in process.
Fable 5 77: Showed rule discipline but failed to execute the signed deal.

Express Schedule Free Employee Scheduling Software [PC/Mac Download]

User-friendly drag & drop planning: Simple shift scheduling interface
Manage time-off and holidays: Add sick leave, breaks, holidays
Email schedules to staff: Send schedules directly via email

View Latest Price

As an affiliate, we earn on qualifying purchases.

What the Results Reveal

The key takeaway? All four models identified every crisis and refused every manipulation attempt, including a staged social engineering attack with fake CEO messages. Yet, only two—gpt-5.6-sol and Kimi K3—actually signed the deal their own analysis justified. The others, despite understanding the diagnosis, left the contract unsigned or didn’t escalate issues properly, revealing a hidden weakness: the ability to follow through on commitments.

The Hidden Weakness: Reading Deeper Into Files

The decisive difference lay in reading comprehension. The models that succeeded in closing the deal accessed information buried two document references deep within the company’s files—information that proved critical in sealing the agreement. This detail was invisible in chat demos but vital in real decision-making.

Beyond Chat: What Really Matters

While conversation quality often gets attention, this experiment underscores a more important factor: can the AI see the full context, resist manipulation, and execute decisions reliably? For businesses, especially those managing customer support, forecasts, or internal processes, it’s not just about how well an AI can talk, but whether it can finish what it starts amidst real-world pressures.

The Discipline of Management Under Pressure

In the live company simulation, discipline varied among models. Opus 4.8, which had the most thorough analytical approach, ultimately failed to close the deal, leaving the opportunity unfulfilled. The lesson? Depth of analysis doesn’t guarantee execution. The ability to act decisively and uphold integrity under temptation is crucial.

Implications for Your Business

For companies considering AI helpers, the message is clear: superficial chat demos don’t reveal the true management capacity. You need to test these models under realistic, high-pressure scenarios to see if they can read your documents deeply, resist manipulation, and follow through on commitments—just as a bakery owner must follow a recipe to the end, even when tempted to cut corners.

Firmulate offers a live, watchable experiment platform where businesses can run their own scenarios against AI models, ensuring they’re choosing helpers capable of managing real crises, not just chatting about them. Visit firmulate.com to learn more about how you can test your AI workforce before bringing it into your operations.

Final Takeaway

The true test of AI management isn’t in its ability to generate convincing responses—it’s in its ability to see deeply, stay honest under pressure, and complete what it starts. As this live experiment shows, only through rigorous testing can you find the models that can genuinely help run your business smoothly, even in the toughest weeks.