Qoveris by Core Matrix
Qoveris by Core Matrix

From a business problem to an explainable quantum model

Hybrid Quantum Decision Intelligence. Qoveris turns real decisions into rigorous quantum-ready models across every problem class quantum computing can attack — optimization, simulation, machine learning, Monte-Carlo estimation and search — always next to a classical benchmark.

Truth over hype.

Sign in Open Portfolio Studio Try the Quantum Agent

No quantum circuit required. Works offline with bundled demo data.

problem classes
5
solver backends
10
benchmarked
100%
hosted
EU

01 / How it works

From business problem to explainable quantum model

Domain-first by design. You bring the decision; Qoveris handles the modeling, the solving, and the explanation — and shows its work at every step.

01

Describe or upload

Start from a real decision — a CSV of assets, a spreadsheet, a graph, or plain language. You never write a quantum circuit; the domain comes first.

02

Automatic modeling

We translate the problem into a QUBO, then map it to an Ising Hamiltonian. Every coefficient traces back to a business input you can inspect.

03

Solve, classical + quantum

Run exact and heuristic classical solvers alongside QAOA simulation — and optional real quantum hardware. Results land side by side.

04

Explainable report

A business, math, and technical write-up with the classical-vs-quantum benchmark, optimality gap, stated limits, and a reproducible seed.

Circuit-first tools start with the qubits. Qoveris starts with the problem — a CSV, a spreadsheet, a graph, or a sentence — and never asks you to think in gates.

02 / Problem classes

Every problem class quantum computing can attack

Not just optimization. Five classes, one honest method: run the quantum algorithm on the built-in simulator and report it next to its classical baseline.

Optimization

Combinatorial decisions — portfolios, routing, scheduling — as QUBO / Ising models.

Quantum: QAOA · quantum annealing

Benchmarked against: exact brute force, simulated annealing, OR-Tools

Open the Portfolio Studio

Simulation

Ground-state energies of molecules and spin systems — H₂ and Ising chains today.

Quantum: VQE (variational quantum eigensolver)

Benchmarked against: exact diagonalization

Run VQE in the Lab

Machine learning

Quantum-kernel classification on nonlinear datasets, honestly benchmarked.

Quantum: quantum kernel methods

Benchmarked against: classical RBF-kernel classifier

Train a quantum kernel

Monte-Carlo estimation

Probability estimation with quadratically fewer oracle queries than sampling.

Quantum: quantum amplitude estimation (QAE)

Benchmarked against: classical Monte-Carlo sampling

Estimate in the Lab

Search

Find marked items in an unstructured space in roughly √N oracle queries.

Quantum: Grover’s algorithm

Benchmarked against: classical linear scan (~N/2 queries)

Search in the Lab

03 / Optimization library

One platform, a growing library of business problems

Finance-first, and multi-vertical — spanning operations, graph, energy and logistics. Each problem ships with automatic modeling and the full solver spectrum.

Portfolio selection

Binary, cardinality-constrained mean-variance: decide which assets to hold under a target count.

Finance-first

Continuous portfolio

Optimal weights — how much of each asset to hold — solved as a classical convex QP.

Finance-first

Knapsack / project selection

Pick the highest-value set of items or projects that fits within a fixed budget.

MaxCut

Partition a graph to maximise the weight of edges crossing the cut — a canonical combinatorial benchmark.

Scheduling

Assign tasks to time slots while avoiding conflicts and respecting capacity.

Unit commitment

Decide which generators to switch on to meet demand at minimum running cost.

Energy

Set cover / coverage planning

Choose the minimum-cost set of sites so every zone is covered at least once.

Logistics

TSP routing

Order stops into the shortest cyclic tour for a single vehicle.

Logistics
More on the wayNew verticals are added continuously.

04 / Automatic modeling

Every coefficient, explained

For optimization, Qoveris turns each problem into a QUBO automatically, then maps it to an Ising Hamiltonian; the other classes get their native quantum formulation — molecular Hamiltonians, amplitudes, feature maps. Nothing is a black box: returns, risk, correlations, budgets and constraints each map to terms you can read and verify.

  • Constraints become penalty terms with transparent weights.
  • Inspect the QUBO matrix and the Ising mapping side by side.
  • The same model feeds every solver, so comparisons are apples to apples.
QUBO → Ising

Illustrative QUBO matrix. In the studio, hover any cell to see which business input produced it.

05 / Solver spectrum

Classical, simulated, and real quantum

Three tiers, one model. Every quantum result is reported next to a classical baseline — no quantum advantage theatre.

Tier 1 · Classical

Trusted baselines

The ground truth every quantum run is measured against — exact where the problem size allows.

  • Exact brute forceProvably optimal for small instances.
  • Simulated annealingFast metaheuristic for larger sizes.
  • OR-Tools (CP-SAT)Exact constraint / integer solver.
  • CVXPYConvex quadratic programming for continuous weights.

Tier 2 · Simulated quantum

The full problem-class lab

Run every quantum method on a built-in state-vector simulator — no hardware queue, fully reproducible.

  • QAOA + VQEVariational circuits for optimization and simulation.
  • QAE + quantum kernels + GroverEstimation, machine learning and search.
  • Built-in simulatorDeterministic, seeded, offline-capable.

Tier 3 · Real hardware

Quantum, opt-in

When you want to test on physical devices, route the same model to real backends — always reported next to the classical baseline.

  • D-WaveQuantum annealing.
  • IBM QuantumGate-model QAOA.
  • AWS BraketMulti-vendor access.
“I have 40 stocks and want the best 6 to hold, balancing return and risk.”
Agent: Classified as a cardinality-constrained portfolio. Building the QUBO, running exact + QAOA + annealing, comparing against the classical optimum…

Result: deterministic engines computed the numbers; the agent explains them.

06 / Quantum Agent

Describe the problem in plain language

The Quantum Agent is a domain-first assistant: it classifies your problem, builds the model, runs the solvers, and explains the outcome in words you can take to a stakeholder.

Guardrail: the AI proposes and explains, but the deterministic engines are the only mathematical authority. It never invents numbers.

Open the Quantum Agent

07 / Reports & trust

Results you can defend

Explainable by construction, reproducible by design, and honest about quantum's current limits.

Three levels of depth

Each report reads at a business, math, or technical level — the same result, the right detail for the audience.

Benchmark, always

Classical-vs-quantum comparison and optimality gaps are part of every report, not an optional extra.

Reproducible

Every run is seeded and reproducible end to end. Re-run it and get the same numbers.

Limits stated plainly

Where a method falls short or a result is heuristic, the report says so. Truth over hype.

Exportable

Publication-ready output you can export and share — from dataset to write-up in seconds.

EU-hosted & sovereign

Hosted entirely in the EU (europe-west), with reproducibility and data residency built in.

08 / Workspaces & persistence

Save, reload, and share your work

Sign in with Google or email. Organise work in multi-tenant workspaces, save experiments, reload them later, and share them with your team — all in one account.

Accounts

Google or email sign-in.

Workspaces

Multi-tenant, team-scoped.

Saved experiments

Reload any run exactly as it was.

Sharing

Share within a workspace.

09 / Developer platform

A REST API and a Python SDK

Everything in the studio is available programmatically. Drive optimization from your own pipelines and get the same benchmarked, reproducible results.

  • POST /portfolio/run-full
  • POST /problem/run-full
  • POST /portfolio/continuous
  • POST /simulation/vqe · /montecarlo/qae
  • POST /qml/classify · /search/grover
# Python client SDK
from qoveris import Client

client = Client(api_key="...")

# Optimization — with classical baselines
result = client.portfolio.run_full(
    assets=my_assets,        # returns, risk, correlations
    cardinality=5,           # hold exactly 5 names
    solvers=["exact", "qaoa", "annealing"],
)
print(result.benchmark)      # classical vs quantum, optimality gap

# Any other problem class — same honesty rule
vqe = client.lab.vqe(system="h2")       # vs exact diagonalization
qae = client.lab.qae(target=0.25)       # vs classical Monte-Carlo
print(vqe.error_vs_exact, qae.query_advantage_factor)

10 / Trust & sovereignty

Reproducible by design, EU by default

Reproducible

Seeded runs, versioned models, identical results on re-run.

EU-hosted

Cloud Run + App Hosting in europe-west; Firestore in europe-west8 (Milan).

Honest about limits

We don't claim quantum advantage today. We benchmark it, and we say so.

See it run end-to-end in under a minute

Load a portfolio, build the QUBO, compare solvers, and generate a report — live. Works offline with bundled demo data.

Launch the Portfolio Studio Talk to the Quantum Agent