QCentroid has launched QCentroid Missions, a new capability within its QuantumOps platform designed to evaluate computational solutions, classical, quantum-inspired, or quantum, against an organization’s existing baseline. The system uses expert agents to benchmark solver paths across multiple approaches, returning validated evidence to inform production decisions, and is being demonstrated this week at NVIDIA GTC Berlin, QCentroid says.

“With Missions, teams no longer have to manually drive every stage of an experiment,” says Carlos Kuchkovsky, CEO and co-founder of QCentroid, “They define what they want to improve, how far the system is allowed to go and what counts as proof.” Unlike many quantum efforts, the platform can recommend retaining the current classical solution if it remains optimal based on the evidence.

QuantumOps Agents Automate Solver Evaluation Across Computing Backends

This capability moves beyond isolated quantum experimentation by integrating diverse computing backends into a unified evaluation framework, allowing teams to select resources as part of application engineering. The system’s design acknowledges that optimal solutions aren’t always quantum. It can recommend retaining existing classical methods if evidence indicates they remain superior, a pragmatic approach uncommon in early-stage quantum development. The platform’s automation extends to defining specific criteria that must be met during experimentation before any solver evolution is considered.

Teams establish these gates to ensure results align with pre-defined business needs, which connects quantum exploration to practical application. The system then uses expert agents to autonomously cycle through discovery, feasibility analysis, solver design, implementation, and benchmarking, all within the boundaries set by the user, according to QCentroid. Each Mission incorporates a measurable goal, computational limits encompassing budget and time, and the aforementioned validation gates, creating a self-contained experimentation loop.

This allows for a more efficient use of resources, as the platform handles the iterative process of testing and refinement. QCentroid Missions builds on CUDA-Q to help enterprise teams prepare for a future where “useful workloads in future quantum supercomputers will draw on CPUs, GPUs and QPUs,” according to the company. The new capability will be available to QCentroid Platform customers starting in the fourth quarter of 2026, expanding the platform’s role in governing experimentation across heterogeneous computing environments.

Useful workloads in future quantum supercomputers will draw on CPUs, GPUs and QPUs.
Sam Stanwyck, director of quantum product at NVIDIA

NVIDIA CUDA-Q Integration Validates Production-Ready Quantum Solutions

QCentroid Missions establishes a benchmarking contract, evaluating potential solver evolutions against an organization’s existing computational baseline before implementation, a departure from approaches that begin with a specific quantum algorithm. The system’s design prioritizes outcomes, such as reduced execution time or improved solution quality, initiating experimentation with a measurable business objective rather than a predetermined technology, the company says. Teams define parameters for improvement, resource limits and acceptable evidence levels, allowing automated agents to manage the experimental cycle and deliver data-driven recommendations.

NVIDIA’s CUDA-Q integration accelerates circuit simulation within QCentroid Missions, using NVIDIA accelerated computing through the QCentroid Launchpad. This allows for a unified experiment lifecycle spanning classical, quantum-inspired and quantum approaches, with quantum processing unit (QPU) execution reserved for scenarios where the evidence supports its use.