Summary

Oransim provides a causal digital twin for marketing, enabling enterprise growth teams to predict campaign ROI and optimize decisions before spending. It offers transparent, auditable causal simulation and LLM-driven insights for pre-launch ranking, mid-campaign adjustments, and post-mortem analysis. The OSS repo allows auditing the core engine; enterprise licensing is required for production data access.

How It Works

The system uses a 64-node Structural Causal Model (SCM) with per-arm counterfactual heads (TARNet, Dragonnet). It simulates consumers via IPF and LLM-backed "soul personas" for qualitative feedback. Temporal diffusion uses a Causal Neural Hawkes Process. The OSS version uses a demo corpus and baseline models; enterprise leverages licensed real-world data and advanced pretrained models.

Quick Start & Requirements

Clone (git clone https://github.com/OranAi-Ltd/oransim.git), cd oransim, and install with pip install -e '.[dev]'. Run backend: python -m uvicorn oransim.api:app --port 8001 &. Run frontend: python -m http.server 8090 --directory frontend. For ML features, pip install 'oransim[ml]' (PyTorch required). API mode needs an LLM API key. See docs/en/quickstart.md.

Highlighted Details

• Enables rapid (60-second) simulation of campaign combinations (creative,