Causal Dynamics Engine

Built today

A causal dynamics engine
for science and engineering.

The Universal Discovery Engine

CDE discovers causal structure, governing equations, and directed mechanisms from observational and experimental data. Identify what drives what, what changes under intervention, and which experiment resolves ambiguity fastest.

Go from raw data to causal understanding your team can review, test, and defend.

Schedule a Demo Explore Use Cases

CDE — Causal Dynamics Engine, pipeline visualization

What CDE Gives Teams

Causal structure from the data you already have.

CDE discovers directed causal mechanisms, governing equations, and interpretable dynamics from observational and experimental data. Method selection happens internally. Output is causal claims your team can review, test, and act on.

Access CDE through API, SDK, CLI, MCP, or guided workflows. Every path leads to the same result: a repeatable process from raw observations to governed, reviewable findings.

See integration options

Discovers causal structure

Traces directed influence across variables, produces causal graphs with probabilities and identifiability scores, and shows what drives what.

Discovers governing equations

Extracts the equations behind a system, from compact laws to regime-dependent dynamics, grounded in causal relationships rather than curve fitting.

Profiles and routes automatically

Characterizes your data, selects the right internal methods, and starts causal discovery from the actual structure of the problem.

Validates before promoting

Runs negative controls, falsification tests, and held-out checks before any causal claim reaches your team.

Keeps the evidence

Stores inputs, configuration, causal beliefs, and outcome history together. Results can be replayed, reviewed, and defended at any time.

Input Data

Use the data you already have

CDE works with the data you already collect: measurements, traces, and structured observations across formats and modalities.

Time-series

Sensor readings, experimental traces, financial ticks, and any temporally ordered observations with regular or irregular sampling.

Spatial fields

2D/3D scalar and vector fields from simulations, imaging, or environmental monitoring on grids or unstructured meshes.

Geometric

Point clouds, meshes, manifolds, and shape data where the geometry itself encodes physical or biological structure.

Hierarchical

Multi-scale and nested data: molecular–cellular–tissue, component–subsystem–system, or any level-separated observation structure.

Relational

Graphs, networks, and interaction matrices: protein interactions, supply chains, circuit topologies, social dynamics, or causal diagrams.

Tabular

Feature-observation matrices from experiments, surveys, or databases. CDE discovers governing relationships across columns.

Multi-modal

Combined modalities: time-series with images, spectra with metadata, text annotations with measurements. Fused through explicit interfaces.

Workflow

From raw data to reviewable findings

Every discovery run follows a consistent sequence. Your team can see what happened, why a method was chosen, and what evidence supports the result.

Data ingestion

Observational streams, experiments, and simulation exports enter CDE with provenance tracking. Lineage records every source, time range, and preprocessing assumption before discovery begins.

Profiling

CDE characterizes the data and determines which discovery modes apply. The profile feeds metadata forward to validation.

Discovery

Based on the profile and your configuration, CDE selects the right approach automatically. Routing decisions are recorded in run metadata so reviewers can see why a strategy was chosen.

Causal structure

CDE discovers directed causal structure, governing equations, and system mechanisms from the data. Artifacts stay linked to configuration snapshots for replay and comparison.

Validation

Results are checked against held-out data, negative controls, and domain-specific sanity tests before promotion. Failed runs include full context on why they failed.

Claims

Validated structure is emitted as typed scientific claims with evidence links and governance state. Claims are the interchange format between CDE, your team, and your agents.

Evidence ledger

Every run appends a versioned record of inputs, configuration, outputs, and claim lineage. Trace forward from raw inputs or backward from any promoted claim.

Causal Dynamics

Discover causal structure from raw data

CDE answers what drives what, what changes under intervention, and which experiment would resolve ambiguity fastest. It selects the right approach automatically. Output is always causal.

Core Capability

Causal dynamics†

CDE maps directed influence between entities, produces causal graphs with confidence estimates, and designs targeted experiments to resolve ambiguous edges. Intervention reasoning is built in.

All four discovery modes feed into this objective: discovering the causal mechanisms that govern the system under study.

Deep dive into CDE

How It Works

CDE selects and combines methods internally. Your team interacts with CDE, not with individual techniques.

Compact laws

When the system admits closed-form governing relationships, CDE discovers expressions your team can read, compare against theory, and use in downstream systems.

Complex systems

When the dynamics are too rich for compact laws, CDE captures the structure while quantifying uncertainty and providing causal analysis.

Interpretable structure

CDE produces governing relationships you can verify, share, and challenge — regardless of the complexity of the underlying system.

Architecture

Composable and domain-agnostic

CDE adapts to your domain and data type. Every run is versioned and reproducible.

New domains work without rewriting the pipeline. Temporal data, spatial fields, relational graphs, multi-modal observations: CDE assembles the right configuration automatically.

Simulation universes

Built-in worlds for validation

CDE ships with simulation universes for validating causal discovery and benchmarking configurations. Each universe has known governing equations and causal structure, so you can verify correct mechanism discovery before running on proprietary data.

Spring

Pendulum

Lorenz

Lotka-Volterra

Van der Pol

Duffing

Brusselator

Glycolysis

FitzHugh-Nagumo

Kuramoto

Hodgkin-Huxley

CSTR

Wave

Heat

Burgers

Navier-Stokes

Tokamak Plasma

Battery Cell

Use these universes for regression testing, mode comparison, and operator training on failure modes before touching production systems.

Scientific Output

Typed scientific claims, not free text

Every discovery run produces typed, machine-readable scientific claims. Each claim carries metadata, confidence scoring, provenance, and governance status. Structured knowledge you can audit, compare, and reproduce.

Governing equationsCausal graphsConservation lawsSystem structureRegime boundariesDecompositionsSymmetriesInvariantsScope annotationsUncertainty estimatesExperiment recommendations

What CDE discovers

Governing equations with fit quality and scope annotations
Directed causal graphs with confidence estimates
Conservation laws with stability analysis
Symmetries and invariant properties
Regime boundaries and transition analysis
Competing explanations with evidence for each
Experiment recommendations to resolve remaining ambiguity

Evidence Ledger

Every run writes a hashed, versioned record of everything that happened: a structured evidence entry that supports audit, reproduction, and peer review.

Data Provenance

Input fingerprints
Configuration record
Source lineage

Run Metadata

Version identifiers
Environment record
Timestamps

Results

Primary metrics
Claims and evidence
Confidence estimates

Governance

Validation results
Promotion status
Replay recipe

Review And Proof

A discovery is only useful if the team can check it

When CDE surfaces a candidate law or causal path, every piece of validation and evidence stays attached. Your team can inspect the controls, trace the lineage, and decide whether to act.

A single Truth Dial controls the rigor-speed tradeoff across the pipeline. Set it for fast exploration, internal validation, or external-grade review depending on where you are in the research process.

Negative controls run automatically at every tier. Claims that pass move forward. Claims that fail stay visible with the reason attached.

Explore

Fast iteration. Claims are tagged as hypotheses. Use this for initial data exploration and rapid ideation.

Validate

Negative controls are applied automatically. Claims that pass are promoted to provisional status.

Publish

Full validation suite with deterministic replay. Generates a complete replay recipe suitable for external review and regulatory submission.

Why CDE

Why teams choose CDE

Teams choose CDE when they need to move from observations to causal understanding: directed mechanisms, governing equations, and reviewable claims they can defend.

Search and summarization tools surface what is already known.

CDE discovers new causal structure from your own data, including directed mechanisms and governing equations.

Generic prediction pipelines optimize fit without explaining the system.

CDE produces causal graphs and governing equations your team can inspect, test, and intervene on.

Point tools cover one method or one scientific niche at a time.

CDE selects methods internally across domains in a single causal dynamics engine.

Ad hoc notebooks make review and reproducibility hard to scale.

CDE ties causal claims to inputs, validation, and evidence. Results can be defended.

Industries

Where CDE fits

Built for teams that need causal understanding from complex data: directed mechanisms, governing equations, and causal models they can present to scientists, operators, or reviewers.

Life Sciences & Healthcare

Pharmaceutical R&D

Identify molecular interaction laws, binding dynamics, and pharmacokinetic equations directly from experimental assay data.

	Point Tools	CDE
Approach	Single method, manually configured	Automated method selection and composition
Output	Equations or predictions	Typed scientific claims: causal graphs, governing equations, conservation laws, and more
Causality	Correlations	Directed causal graphs with intervention reasoning and experiment design
Validation	Manual	Automated negative controls at every tier
Pipeline	Notebooks and scripts	Automated end-to-end: ingest through governed output
Model construction	You choose and configure	Engine selects from your data automatically
Reproducibility	Seed-dependent, manual tracking	Deterministic replay with evidence ledger
Governance	None	Truth Dial tiers with provenance tracking

A causal dynamics enginefor science and engineering.

Causal structure from the data you already have.

Discovers causal structure

Discovers governing equations

Profiles and routes automatically

Validates before promoting

Keeps the evidence

Use the data you already have

Time-series

Spatial fields

Geometric

Hierarchical

Relational

Tabular

Multi-modal

From raw data to reviewable findings

Data ingestion

Profiling

Discovery

Causal structure

Validation

Claims

Evidence ledger

Discover causal structure from raw data

Causal dynamics†

Compact laws

Complex systems

Interpretable structure

Composable and domain-agnostic

Built-in worlds for validation

Typed scientific claims, not free text

What CDE discovers

Evidence Ledger

A discovery is only useful if the team can check it

Explore

Validate

Publish

Why teams choose CDE

Where CDE fits

Life Sciences & Healthcare

Pharmaceutical R&D

Biotechnology

Clinical Research

Genomics & Proteomics

Neuroscience

Epidemiology & Public Health

Energy & Resources

Oil & Gas

Renewable Energy

Nuclear & Fusion Energy

Power Systems & Grid

Mining & Resource Extraction

Advanced Technology

Semiconductor & Electronics

Robotics & Autonomous Systems

Quantum Computing Research

AI & Machine Learning Research

Engineering & Manufacturing

Aerospace & Defense

Automotive & Mobility

Advanced Manufacturing

Civil & Structural Engineering

Materials & Chemistry

Materials Science

Chemical Engineering

Polymer Science

Nanotechnology

Climate & Environment

Climate Science

Oceanography

Environmental Monitoring

Ecology & Conservation

Finance & Economics

Quantitative Finance

Risk Modeling

Economic Forecasting

Cross-Industry Applications

Agriculture & Food Science

Telecommunications

Supply Chain & Logistics

A causal dynamics engine
for science and engineering.