Moonshot Hackathon 2026

Turn scattered science into executable causal worlds

CausalForge transforms fragmented research papers and hypotheses into manipulable causal world models. Simulate interventions, expose contradictions, and discover the next decisive experiment.

How it works

Four steps from literature to discovery

01

Ingest Science

Feed research papers, abstracts, or structured notes into CausalForge. Extract structured claims, variables, and mechanisms.

02

Build Causal World

Transform extracted claims into an explicit causal graph with variables, edges, signs, uncertainty scores, and evidence counts.

03

Simulate Interventions

Choose an intervention and observe predicted downstream effects. See how confidence propagates through the causal structure.

04

Discover Next Experiment

Identify the highest-information-gain experiment. Know exactly what to measure next and why it matters.

Technical Depth

Not a chatbot. Not a search engine.

Scientific infrastructure for the age of AI-native research

Core Engine

Explicit Causal Graphs

Real causal structure, not hidden in a prompt. Variables, edges, signs, strength, and uncertainty are all inspectable.

Scientific Rigor

Contradiction Detection

Automatically identifies conflicting claims, incompatible mechanisms, and evidence gaps across literature.

Prediction

Counterfactual Simulation

Choose interventions and observe modeled downstream effects with confidence propagation and sensitivity analysis.

Discovery

Experiment Ranking

Information-gain scoring to identify which experiment would most reduce uncertainty and change the field.

Trust

Evidence Inspector

Every edge and recommendation traces back to source documents, excerpts, and confidence levels.

Interface

Grounded Assistant

Ask CausalForge anything about the model. Answers come only from the graph, claims, and simulations — never hallucinated.

The Vision

Humanity stores science in documents. Documents are readable but not executable. This slows discovery and hides contradictions.

The future is not just AI answering questions about papers. The future is AI turning science into manipulable causal worlds. CausalForge is a first prototype of that future.

Enter the Causal World
CausalForge — Moonshot Hackathon 2026Built with NVIDIA NIM · MongoDB Atlas · Next.js