📊 Full opportunity report: AI’s Management Shortcomings Are Hidden Behind Successes on ThorstenMeyerAI.com — validation score, market gap, and execution plan.
TL;DR
A recent experiment shows AI models can diagnose crises and develop pitches but struggle to finalize work, highlighting hidden management flaws. Trust and discipline are key differentiators, not just analysis quality.
Recent experiments conducted by Firmulate have demonstrated that while AI models can accurately diagnose business crises and generate appropriate responses, they often fail to complete critical, trust-dependent tasks such as closing deals or executing authorized actions. This finding exposes a significant gap in AI’s management capabilities, with implications for enterprises relying on automation for operational decisions.
In a live test involving a small software company, five AI models faced real customer crises, manipulative attempts, and commercial opportunities. All models identified crises, resisted social engineering, and formulated pitches. However, only two models successfully signed a €55,000 deal, despite all understanding the situation equally well. The experiment revealed that the key difference was not understanding or analysis, but the discipline to follow through and finalize work under pressure.
The experiment used a company with 13 synthetic employees, real money mechanics, and a transparent cash countdown, making operational discipline visible. The models’ performance was ranked in the July 2026 Crucible League, with GPT-5.6-SOL leading at 95 points. Trust and execution discipline emerged as the decisive factors, with even thorough models like Opus 4.8 failing to complete the most critical task—closing the deal—when attempting to act beyond analysis.
Additionally, the models effectively recognized manipulation attempts, such as fake CEO messages, and refused social-engineering requests, demonstrating that safety awareness alone does not guarantee successful management. The core challenge identified was whether AI can translate correct analysis into trustworthy, completed work amid real-world pressures, a capability still largely unproven at scale.
This research highlights that AI’s ability to understand and analyze business crises does not automatically translate into effective management or operational success. Enterprises relying on AI for decision-making must consider that completion and discipline are critical factors, not just reasoning or safety features. The failure to finish tasks can lead to missed opportunities, despite correct analysis, exposing organizations to operational risks and trust issues. As AI adoption grows, understanding these management shortcomings will be essential for designing more reliable, disciplined systems.
AI for Project Managers: A Desk Reference & Field Guide: Use Artificial Intelligence to Streamline Workflows, Automate Tasks, and Make Smarter Decisions with Practical Tools and Ethical Insights
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Background of AI Management and Operational Testing
Over recent years, AI models have advanced rapidly in understanding complex business scenarios, with many demonstrations focusing on analysis, summarization, and safety. However, practical deployment in operational settings remains challenging, especially when AI must act autonomously or semi-autonomously. The Firmulate experiment builds on prior work by testing AI models in a live, decision-making environment that mimics real business pressures, including manipulation attempts and financial stakes. This approach aims to reveal whether AI can bridge the gap from understanding to action, a question increasingly relevant as organizations seek to automate more of their workflows.
“The models understood the crises and formulated responses, but the core issue was whether they could follow through and complete the work under pressure.”
— an anonymous researcher
The AI Sales Coach: Objection Handling, Closing, and Prospecting Reimagined (The Objection Handler's Library)
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Unresolved Questions About AI’s Operational Capabilities
It remains unclear how scalable these findings are across different industries and more complex operational environments. The experiment was conducted in a controlled, small-scale setting, and broader validation is needed to determine if similar management shortcomings will manifest in larger, real-world deployments. Additionally, the long-term impact of integrating AI systems that can diagnose but not reliably execute tasks is still uncertain, especially regarding trust, safety, and organizational resilience.
Pydantic AI for Automation Workflows: Build Typed, Reliable, and Production-Ready AI Automations in Python
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Next Steps for Evaluating and Improving AI Management
Organizations should consider conducting similar live tests within their own operations to assess AI’s discipline and completion capabilities before full deployment. Developers and vendors need to prioritize enhancing AI’s ability to follow through on decisions, especially in high-stakes contexts. Further research is expected to explore methods for embedding operational discipline into AI systems, potentially through improved training, oversight, or hybrid human-AI workflows. Monitoring how these systems perform in real-world scenarios will be crucial for building trustworthy automation.
AI discipline and trust training
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Key Questions
Why do AI models sometimes fail to complete tasks despite understanding them?
Understanding a problem does not guarantee that an AI will follow through with the necessary actions, especially under pressure or manipulation. Completion requires discipline, operational protocols, and sometimes human oversight to ensure tasks are finalized.
What does this mean for companies using AI for automation?
Companies should evaluate not only AI’s reasoning and safety features but also its ability to reliably complete tasks. Testing AI in live, decision-making environments can reveal hidden management gaps that might impact operational success.
Are these findings applicable across different industries?
The experiment was conducted in a specific context, so broader applicability remains uncertain. More testing across industries and operational scales is needed to confirm whether similar management shortcomings exist universally.
How can organizations improve AI’s ability to complete work?
Enhancing operational discipline, embedding verification steps, and combining AI with human oversight can help ensure that analysis translates into trustworthy, finished work. Ongoing monitoring and testing are also essential.
What should AI developers focus on next?
Developers should prioritize building AI systems that not only understand and analyze but also reliably act within operational protocols, especially in high-pressure or manipulative environments.
Source: ThorstenMeyerAI.com
