Haiqu launches agentic quantum OS for research teams

Haiqu has launched an Agentic Quantum Operating System for enterprise and scientific quantum research and development. The New York-based software developer said the platform has already been tested on current quantum hardware.

The system combines agentic artificial intelligence with Haiqu's middleware to help researchers design applications, run experiments and repeat them with lower computing demands. It is aimed at quantum research teams facing high costs and long development cycles when moving from an idea to a working experiment.

Haiqu described the product as a full-stack platform that takes users from a business question or scientific research idea to an execution-ready quantum application plan. The software has three core components: agentic intelligence for application design, a software development kit for optimisation and error mitigation, and a runtime layer that manages execution.

Richard Givhan, Chief Executive Officer and Co-founder of Haiqu, said the main constraint in quantum research is often not machine access itself. "The bottleneck for quantum R&D teams is often not access to a QPU. It is the time and expertise required to identify the right problem, structure the work and get credible application prototypes," Givhan said. "With our first Agentic Operating System, we are giving R&D teams effective tools to achieve commercial applications as systems become more powerful."

Recent internal tests showed large differences in the cost and time required for some workloads, according to Haiqu. In one example, the company said a molecular dynamics simulation that had previously taken more than nine hours and cost about USD $30,000 was reproduced on its platform in about 30 seconds for roughly USD $25.

Haiqu said it found similar or better results in work involving optimisation algorithms, quantum machine learning models and probability distributions. It also said the software could generate advanced scientific workflows from scratch, including simulations of the single-impurity Anderson model, which is used to study strongly correlated electron systems.

Another test involved a software pipeline for neutron-scattering experiments on one-dimensional quantum magnets. Haiqu said the pipeline reproduced experimentally observed signatures of magnetic materials, illustrating how current quantum computers can support scientific simulation when paired with suitable software.

Early users

Some large consulting and professional services groups have been given early access to the system, including Capgemini and Deloitte. Haiqu did not disclose commercial terms or the scale of those deployments.

The launch comes as software groups try to make quantum computing more useful on today's hardware, which remains limited by error rates, qubit constraints and execution costs. Much of the industry's recent work has focused on narrowing the gap between theoretical algorithms and experiments that can be run repeatedly on available machines.

Banks, manufacturers, pharmaceutical groups and consulting firms have been exploring whether quantum systems could eventually help in areas such as optimisation, simulation and machine learning. For now, many projects remain at the research stage, with companies testing how to frame problems in ways existing hardware can handle.

Kristin Milchanowski, Chief AI and Quantum Officer at BMO and founding director of the BMO Institute for Applied Artificial Intelligence and Quantum, said tools in this area are relevant to the broader question of scaling the technology.

"As quantum hardware continues to evolve, foundational challenges such as data loading and efficient utilization of limited qubits remain critical hurdles," Milchanowski said. "Observing research into tools like Haiqu's middleware allows for a deeper understanding of how these bottlenecks might eventually be addressed. These early-stage, research-driven insights are vital for informing the long-term direction of the quantum landscape and understanding the future scalability of the technology."

Haiqu said its software is hardware-agnostic, meaning it can work across different quantum systems rather than relying on a single machine architecture. It also said its broader software stack can run applications with up to 100 times more operations on current devices than rival approaches, though it did not provide comparative details in the announcement.

The business is based in New York City and has staff across the US, Canada, Ukraine, the UK, the EU and Singapore.