OranAi-Ltd/oransim

Causal Digital Twin for Marketing at Scale · Predict any marketing decision before you spend a dollar.

What it solves

Oransim is a causal simulation engine designed for enterprise marketing teams to predict the ROI of advertising campaigns before they are launched or while they are active. It replaces expensive, time-consuming A/B testing and retrospective analysis with high-fidelity simulations that can rank creative combinations, simulate mid-campaign interventions (like swapping influencers), and perform post-mortem counterfactual analysis to determine what would have happened under different conditions.

How it works

The system combines a 64-node causal graph with an agent-based society of virtual consumers. It uses a "do()" operator for counterfactual reasoning, allowing users to intervene in the simulation. The architecture integrates several components:

  • World Model: Predicts factual and counterfactual outcomes using either a LightGBM baseline or a Causal Transformer.
  • Agent Layer: Simulates a population of virtual consumers with LLM-backed "soul personas" that react to actual creative content.
  • Diffusion: Uses a Causal Neural Hawkes Process to forecast the 14-day rollout and opinion propagation across the agent population.
  • Platform Adapters: Pluggable modules that allow the engine to run on data from various platforms like XHS (RedNote), TikTok, and Instagram.

Who it’s for

It is primarily built for enterprise CMOs and growth teams in sectors such as beauty, FMCG, consumer electronics, and DTC brands who need to data-driven predictions for budget allocation and creative strategy.

Highlights

  • Counterfactual Reasoning: Ability to simulate "what-if" scenarios (e.g., swapping a KOL on day 3) in seconds.
  • LLM-Backed Personas: Virtual consumers that read and react to actual ad creatives.
  • Transparent Causal Logic: An open-source engine that allows users to audit the reasoning path from prediction to causal graph.
  • Pluggable Data Providers: Support for CSV, JSONL, and OpenAPI, allowing users to bring their own data into the simulation framework.

Related

  • Project
    mckinsey/causalnex

    A toolkit for causal reasoning and "what-if" analysis using Bayesian Networks to identify causal relationships and estimate the impact of interventions.

  • Project
    MiroShark/MiroShark

    A multi-agent simulation platform that creates grounded AI personas to simulate social and market reactions to specific scenarios across platforms like X and Reddit.

  • Project
    AIScientists-Dev/WorldSeed

    A world engine for emergent multi-agent outcomes that allows users to define rules and roles in YAML to simulate complex interactions, research communities, or games.

  • Project
    camel-ai/oasis

    A scalable social media simulator using LLM agents to mimic up to one million users for studying social phenomena like information spread and group polarization.

  • Dispatch
    vLLM Semantic Router: Enhancing Model Performance via Micro-Agent Collaboration

    vLLM introduces the Semantic Router, a serving-layer primitive that transforms a single model API call into a bounded collaboration of micro-agents to outperform frontier models on complex benchmarks.