Agent-Based Modeling

Feed the system a text — a regulation, a news archive, notes about people in your life — and it extracts the relevant actors, builds personas, maps the network, simulates a conversation among them on a question you pose, and produces a report. You can then chat with any single agent or poll the whole group.

Cases tested

• MiCA regulation — 49 actors (regulators, central-bank heads, journalists, owners), modeled with separate public vs. private positions.
• Personal career pivot — 20 real people from Max's network; some replies were startlingly in-character (mother responded about spring weather, exactly as she would).
• Iran–US war — ~1,000 actors; too expensive to simulate. Scaling limit is in the tens-hundreds, not thousands.

Business applications

• Policy lobbying — model an EU directive, map coalition options (Convector / chemical-ban directive is a live candidate).
• Leadership 360 — project "who a team would be together in 5 years" from cross-evaluations.
• Strategic simulation — a leader brings a decision; run the likely conversation among stakeholders.

Honest caveat

This is not prediction — it's structured scenario-building. Before any predictive claim, we need backtesting on known past events.

Full write-up: Agent-Based Modeling