Oransim Causal Marketing Twin

This is a causal simulation engine for marketing teams who want to model campaign outcomes before spending budget. You get a 64-node causal graph that lets you predict ROI across creative/KOL/platform combinations, run counterfactual swaps mid-campaign (swap a KOL on day 3, see the 14-day trajectory change), and audit exactly which nodes drove each prediction. The backend uses a Causal Neural Hawkes Process with TARNet counterfactual heads and LLM-backed consumer agent personas. It's open source, runs locally in mock mode without API keys, and includes Python SDK access if you want to skip the HTTP layer. Useful if you're tired of A/B testing in production and want to explore the decision tree beforehand.