Skip to main content

Platform

Run CDE on your data
without rebuilding around it.

CDE runs on the data you provide for the discovery problem at hand. Choose cloud-hosted, self-hosted, or air-gapped deployment based on data locality, compute policy, and how close the engine needs to sit to the workflow.

Contact Vareon Back to CDE
CDE engine deployment architecture

Your Data

Start from the data you already have

Unlike systems that require millions of samples, CDE discovers governing equations and causal structures from the datasets you already collect.

Your data stays yours. CDE trains per run on the data you provide. No pre-existing model or external dataset is required before discovery begins.

Why teams choose this deployment path

  • No external datasets required to start
  • Models build per-run on your data
  • Works with small datasets
  • Built-in simulation universes for testing and benchmarking

REST API

Upload episodes via POST /v1/data/upload with structured JSON payloads. Profile data, list datasets, and manage uploads programmatically.

Python SDK

client.upload_episodes(), client.profile_data(), client.generate_data() — typed methods with full IDE support.

MCP Integration

Upload, profile, and generate data through Model Context Protocol tools. Agents can provision data without writing HTTP calls.

Real-Time Streams

POST observations to /v1/streams/{id}/ingest as they arrive. Server-sent events for live status. Automatic trigger for discovery runs.

Data Model

Episode-based ingestion

All data enters CDE as episodes — structured observations over time. Only timestamps and observations are required. Spatial, relational, hierarchical, and causal modalities are automatically detected when present.

Core

  • Timestamps
  • Observations
  • Metadata

Spatial

  • Spatial coordinates
  • Spatial features

Relational

  • Graph structure
  • Edge attributes
  • Dynamic edges

Hierarchical

  • Hierarchy mappings

Causal

  • Actions
  • Interventions
  • Events

Quality

  • Missingness indicators
  • Regime labels

Supported input formats: JSON episodes (native), CSV, Parquet, and HDF5 (via SDK conversion). Molecular dynamics adapter payloads are automatically normalized.

Compute

Automatic resource planning

CDE estimates compute requirements from your data profile and discovery mode. Resource planning is automatic.

Cloud-Hosted

Available

Vareon provisions and manages GPU compute. Automatic hardware selection based on data profile and discovery mode. Pay for gpu_seconds consumed. No infrastructure to manage.

Self-Hosted

Available

Deploy CDE on your own infrastructure. License fee only — bring your own GPUs. Full control over data residency and network topology.

Air-Gapped

Available

Fully isolated deployment with offline license activation. No outbound network required. Suitable for classified environments, defense, and regulated industries.

Hardware selection

CPU or GPU, automatic

Runtime budget

Estimated per-run, capped

Concurrency

Multiple runs in parallel

Billing

Compute consumed

Bring Your Own Agent

Connect your own agent workflows

CDE's API surface is programmatic end to end. Bring your own research agent or orchestration layer and use CDE as the discovery engine behind it.

An external agent can formulate hypotheses, design experiments, launch runs, and review results through CDE's structured task system without replacing the discovery engine itself.

Typical agent capabilities

  • Plan research campaigns
  • Formulate hypotheses
  • Design and launch experiments
  • Execute discovery runs
  • Review and interpret results

Integration endpoint

POST /v1/agents/sessions/byo/scientist

{

"project_id": "...",

"objective": "Discover governing equations...",

"dataset_id": "...",

"provider_id": "..."

}

Your agent receives structured tasks, claims its work items, executes discovery runs through the full API, and submits results. CDE handles governance, provenance, and negative controls automatically.

Model Provider

Use your existing model provider

CDE's agent cognition layer supports any OpenAI-compatible model endpoint. Use your existing provider for planning, hypothesis generation, and result interpretation while CDE handles the actual discovery workflow.

Configure the provider once with a base URL, default model, and API key. CDE routes cognition calls to your endpoint using the standard chat completions protocol.

Configuration

  • Any OpenAI-compatible chat completions endpoint
  • Custom base_url and default_model
  • API key via environment variable
  • Multiple providers per tenant

How it works

  • POST /v1/providers to register your endpoint
  • Reference provider_id in agent sessions
  • Cognition calls route to your LLM
  • Standard chat/completions protocol

Put CDE where the data already lives.

Contact us to scope the right deployment path for your data, compute, and operational constraints.

Contact Vareon Explore CDE