ProductBlogAboutSign inRequest access

The Orchestration Stack

From intent to execution.
Four layers. Zero friction.

Marqov connects what you want to compute with where it runs. A unified stack that handles everything from natural language to quantum hardware.

1Agent
2Capsule
3Platform
4Mesh

Scroll to explore layers

1Layer One

Agent

The intelligent interface between you and your compute

Describe what you want to compute in natural language. The Agent understands your intent, configures the right parameters, and creates executable workloads—no YAML required.

  • Natural language workload design
  • Real-time execution monitoring
  • AI-powered optimization suggestions
  • Results analysis and visualization
Marqov Agent
D
Run VQE optimization on 8 qubits for H2 molecule. Target IBM, optimize for accuracy.
M
I'll create a VQE Capsule for H2 ground state energy estimation.
vqe-h2-optimizationDraft
BackendIBM Quantum · ibm_brisbane
Qubits8 (4 active)
OptimizerCOBYLA · 200 iterations
Error mitigationZNE + DD enabled
2Layer Two

Capsule

Reproducible, portable compute packages

A Capsule is everything needed to run your job: code, dependencies, environment, and hardware requirements. Like Docker, but purpose-built for quantum-classical workloads.

  • Self-contained and immutable
  • Versioned and reproducible
  • Shareable across teams
  • Hardware-agnostic specification
capsule.yaml
1234567891011121314151617181920
name: vqe-h2-optimization
version: 1.2.0
author: research-team

compute:
  type: hybrid
  quantum:
    qubits: 8
    provider: [ibm, aws-braket]
    features: [error-mitigation]
  classical:
    gpu: optional

environment:
  python: "3.11"
  packages:
    - qiskit>=1.0
    - qiskit-aer
    - scipy

entrypoint: main.py
3Layer Three

Platform

Your control center for compute operations

The Platform is where Capsules live. Store, version, validate, and schedule your workloads. Track execution history, manage team access, and monitor costs—all from one place.

  • Capsule registry and versioning
  • Job queue and intelligent scheduling
  • Team collaboration with RBAC
  • Cost tracking and budget controls
Marqov Platform — illustrative
WorkloadsCapsulesAnalyticsSettings
247
Total Capsules
↑ 12 this week
12
Running Now
3
Backends in use
QPU · GPU · simulator
$1,247
Cost (MTD)
62% of $2,000 budget
NameBackendStatusDuration
vqe-h2-optIBM Quantum Running2m 34s
tensor-simGPU Cluster Running14m 12s
qaoa-maxcutAWS Braket Complete6m 41s

Illustrative interface — not live data.

4Layer Four

Mesh

One route to the hardware you choose

The Mesh abstracts hardware complexity. You choose a compatible backend; the Mesh normalizes, converts and submits to that backend — and never silently reroutes your work somewhere else.

  • Multi-cloud, multi-provider
  • Compatibility checked before any provider is contacted
  • Requested, recorded and billed target always agree
  • Transparent per-backend pricing
Your Capsule
Mesh
IBM QuantumBest match
127 qubits·2 in queue
AWS Braket
IonQ · 32 qubits·8 in queue
Azure Quantum
Quantinuum·23 in queue
GPU Cluster
A100 · 8 nodes·Available

Ready to orchestrate?

From quantum experiments to GPU workloads—one stack, unified control.