Researcher simulates Iran-Strait of Hormuz crisis using 6 AI agents, showing how LLMs can model real-world geopolitical scenarios

Vincenzo Manto published an interactive simulation on Google Colab that models the Hormuz shipping crisis (triggered by Iran seizing 2 ships) by deploying 6 separate AI agents—each representing a different decision-maker (Iran, US, oil traders, etc.)—that negotiate and respond to each other's moves in real time.

Unlike static reports, the simulation lets readers adjust starting conditions and watch how each AI agent's decisions cascade—showing, for example, how one nation's naval blockade triggers price spikes for traders, which then pressure governments to act. This reveals feedback loops that traditional written analysis misses.

For business professionals tracking geopolitics, this offers a new way to stress-test supply-chain risks and commodity price exposure without waiting for real events to unfold. For students and analysts, it demonstrates how AI agents can model complex multi-party conflicts where no single actor controls the outcome.