Commutator Studios: Shaping the Future of Quantum Error Management

Quantum computing is advancing rapidly, but performance remains fragile. Even the most promising algorithms are limited by noise, decoherence, and device-specific errors that undermine accuracy and make results unreliable.

While much of the industry is focused on building better hardware or implementing low-level error correction, there is a growing demand for tools that help developers understand and manage quantum errors at the application level. Just as classical software development evolved with better debuggers, optimizers, and compilers, quantum development now requires a platform layer to bridge the gap between theory and practical performance.

Commutator Studios was founded in 2023 by César Rodríguez Rosario (CEO) and Jan Knaup (CTO) to address exactly that challenge. The company is developing a hardware-agnostic Quantum Error Management Platform that enables developers to detect, mitigate, and optimize for quantum decoherence, resulting in faster runtimes and more accurate outcomes across a wide range of algorithms and hardware platforms.

What sets Commutator apart is its unique position in the stack. Rather than competing with hardware-focused error correction tools, it complements them by offering a developer-facing layer that abstracts away device-specific complexities. The company’s AI-augmented platform delivers results up to 20 times faster and more accurately.

In July 2025, Commutator Studios raised €1.5 million in a funding round led by Backtrace Capital and Preston-Werner Ventures, a renowned U.S. venture fund led by GitHub co-founder Tom Preston-Werner. They were joined by HBG and Jan Goetz, CEO and co-founder of IQM Quantum Computers.

Learn more about the future of quantum error management from our interview with Commutator Studios co-founder and CEO, César Rodríguez Rosario:

What Inspired You To Start Commutator Studios?

I have been working in quantum computing for nearly 25 years. I was fascinated early on by the physical limits of computation. Computer engineering felt too conventional, so I pursued a PhD in quantum physics at the University of Texas, Austin, under George Sudarshan. He was nominated eight times for the Nobel Prize and is recognized for his work on quantum optics and open quantum systems. Richard Feynman even mentioned him in his book “Surely You’re Joking, Mr. Feynman!”. 

My career took me from theoretical physics to Harvard, where I joined Alán Aspuru-Guzik’s group. Aspuru-Guzik is known as the inventor of quantum algorithms for chemistry. Many of today’s widely used algorithms emerged from that group’s work.

Editor’s note: A quantum algorithm sits just below the application layer in the quantum stack. It defines the sequence of quantum operations used to solve a problem and is executed through lower layers, such as circuits, compilers, and hardware.

Later, I worked at several institutions, including the Max Planck Institute, before joining the U.S. startup Strangeworks as Chief Science Officer. There, I evaluated quantum hardware and software, trained developers worldwide, and worked with customers across industries.

A turning point occurred when I was invited to speak as a keynote speaker for the IEEE (Institute of Electrical and Electronics Engineers). I spoke openly about the challenges quantum computing still faces, especially error management at the developer level, and framed them as opportunities for startups. The response was overwhelming.

At the same time, I saw how often clients relied on me personally to solve exactly these challenges, which was not scalable. I realized the better path was to create a product that could systematize and extend my expertise. That decision led me to leave my role and start Commutator Studios.

What Is Commutator Studios Building to Unlock Practical Quantum Computing?

Our model is simple: deliver the best possible quantum results. Our platform is built for developers and users of quantum applications, who today are not achieving the performance they should.

Editor’s note: Quantum applications are the end-user programs that run on quantum hardware. They sit at the top of the quantum stack and rely on lower layers, such as circuits, compilers, and hardware, to execute their functions.

I like to explain it with an analogy: Classical computers are like trains, while quantum computers are like airplanes. Airplanes are not just faster; they take you places trains cannot go. But just as flying also comes with turbulence and storms, the same physics that gives quantum computers their power also leads to decoherence, which manifests as quantum errors. These errors directly affect quantum circuits, the fundamental building blocks of quantum programs.

Editor’s note: Decoherence is the process by which a quantum system loses its quantum properties due to interaction with its environment. It is one of the main causes of errors in quantum computing.

Editor’s note: Quantum errors refer to inaccuracies that arise from hardware imperfections, noise, and decoherence. They can lead to incorrect results even when the quantum algorithm is theoretically sound.

Editor’s note: Quantum circuits are step-by-step instructions for quantum computers. They instruct the system, which applies operations (called gates) to the qubits in a specific order, much like a recipe in cooking.

Just as pilots rely on radar and weather models to choose optimal routes, quantum developers also need tools to understand and manage their work. Just as it is the pilot, not the airplane or the aircraft manufacturer, who needs access to these instruments, it is the quantum developer who requires novel tools to achieve optimal performance.

Commutator Studios develops a hardware-agnostic Quantum Error Management Platform that empowers developers to identify, mitigate, and compensate for errors. In short, we create the tools quantum developers need to program quantum computers more effectively.

Editor’s note: Quantum Error Management Platform refers to a layer of software tools designed to detect, mitigate, and optimize for quantum errors during application development and execution.

By combining proprietary algorithms with AI, our technology improves the performance and reliability of quantum software across all platforms, delivering results up to 20 times faster and with higher accuracy. This enables developers and end-users in industries such as chemistry, pharmaceuticals, finance, and logistics to derive real value from quantum computing.

At Which Layer of the Quantum Stack Does Commutator Studios Operate?

We operate at the quantum application level. Most companies working on error mitigation or correction are closely tied to the hardware, as they require special access to improve error rates at that layer. But some of those errors persist and reach the application level. Using techniques I first developed at Harvard, we can identify and address these errors.

If you can send a quantum application to run on a quantum computer, you can also run it through our system. We evaluate the noise of the targeted quantum computer, optimize the compilation process, and refine the application before submission. In practice, this means a cloud-based layer that improves quantum applications before they are executed on hardware.

Editor’s note: Noise in quantum computing refers to random or systematic disturbances that affect qubits and gates, leading to unreliable computation.

Editor’s note: Compilation in quantum computing is the process of translating a high-level quantum algorithm into a hardware-executable sequence of quantum gates, often optimized for a specific device.

So You’re Rather Providing a Layer That Complements Quantum Error Correction Than an Alternative, Right?

Exactly, our platform is complementary. We often use the lower-level tools to analyze errors and then guide users on the best options available. These are highly technical choices, and developers want to focus on their applications, rather than the errors or hardware. Our goal is to abstract away that complexity and recommend the optimal settings before the application is executed.

Editor’s note: Lower-level tools refer to software that operates close to the hardware, such as error correction codes, hardware-aware optimizers, or firmware-level calibration systems.

How Does Your Platform Remain Hardware Agnostic While Still Dealing With Device-Specific Error Noise Profiles?

To start with, we made a deliberate choice to focus on gate-based quantum computers. Most of these systems support OpenQASM, the industry standard quantum assembly language. Because almost every major gate-based system (IBM, IQM, and many others) supports OpenQASM, we can deploy our tools across various hardware platforms with minimal effort. Of course, not all quantum hardware uses this standard. For example, D-Wave’s quantum annealers do not, which is why we are de-emphasizing those for now.

Editor’s note: Gate-based quantum computers perform computations using sequences of quantum gates, similar to logic gates in classical computers. They contrast with other architectures, such as annealers.

Editor’s note: OpenQASM is a standard low-level programming language for gate-based quantum computers. It defines how quantum circuits are described and executed on compatible hardware.

Editor’s note: Quantum annealers are a type of quantum computer that solves optimization problems by gradually transforming the system into a low-energy state, rather than using discrete gates.

Our focus is on the largest category, where the majority of developers and users are currently active. This approach keeps us flexible. Some believe platforms like trapped ions, neutral atoms, or photonic systems may gain traction earlier, while others point to silicon or superconducting systems as stronger long-term candidates. By supporting all of them, we make sure we are well-positioned regardless of how the field evolves. In short, we are not concerned with picking a single winner. Our goal is to help every hardware platform perform optimally.

Editor’s note: Trapped-ion qubits are one of several quantum computing platforms. They use individual ions (charged atoms) trapped and controlled with lasers. This platform is known for high precision and long coherence times.

Editor’s note: Neutral atom qubits are another quantum platform. They use atoms held in place by laser fields or optical tweezers and are valued for their scalability and flexible architectures.

Editor’s note: Superconducting qubits are a widely used platform that encodes quantum information in superconducting circuits cooled to near absolute zero. This setup enables fast gate operations and relatively mature hardware.

Editor’s note: Silicon spin qubits form a quantum platform based on manipulating the spin of electrons or nuclei in silicon. They are compatible with conventional semiconductor processes and may allow integration with existing chip technology.

Editor’s note: Photonic qubits represent quantum information using particles of light (photons). This platform enables high-speed communication, operates at room temperature, and is compatible with existing fiber-optic infrastructure.

What Does Improvement Look Like in Practice for You, Beyond Standard Metrics Like Fidelity or Error Mitigation?

Metrics like fidelity and error mitigation are very low-level. We use them, but our goal is to abstract them away from the user.

Editor’s note: Fidelity is a metric that quantifies how close a quantum operation or result is to the ideal, error-free version. Higher fidelity means fewer errors.

Editor’s note: Error mitigation is a set of techniques used to reduce the impact of errors without fully correcting them. It is essential for today’s noisy quantum devices.

To understand the errors, you can rely on the reports from the hardware providers. But those parameters are often too limited to capture all the errors. Using techniques I developed at Harvard, we run code directly on the quantum hardware, allowing it to inspect itself effectively. This enables us to identify errors that providers typically do not report.

Editor’s note: Parameters, in this context, are numerical values used to describe a quantum system’s noise model or error profile. They are critical for simulation and optimization.

Of course, this inspection has a cost, as running additional code consumes quantum computing resources. That is why we focus on modeling only the errors that impact the specific code the user wants to run. By conditioning our analysis on the user’s circuit, we dramatically reduce the parameter space.

The result is that we can identify which errors matter most for performance and how they affect accuracy, meaning whether the algorithm delivers the correct result. We use a mix of circuit simulation and proprietary techniques to predict this impact. Instead of brute-forcing every possible error configuration, which would be prohibitively expensive, our approach finds the optimal compilation and error strategies much faster and at lower cost.

Editor’s note: Brute-forcing refers to exhaustively trying every possible combination (in this case, of quantum error configurations) to identify the best outcome. This becomes infeasible at scale due to the enormous number of possibilities.

From this data, we are also building an AI model that will guide developers in programming quantum computers with error considerations in mind. We are still in the early stages, but the direction is clear. As my co-founder, Jan, who is the AI expert, likes to joke, he is building an AI to replace me and make me useless.

What Are the Main Technical Bottlenecks You Encounter?

The first is access to quantum hardware. This access is essential because our approach requires running analyses directly on the hardware to study errors. To tackle that, we are part of the IBM Quantum Network and receive credits there, and we also have partnerships with other hardware providers. 

The second challenge, which we have not yet fully reached, is scaling the AI we are building. We believe that it will scale well due to the sparsity of the models we use, but this is an assumption we are still testing.

Lastly, we need more access to production-level quantum applications. Currently, we rely on benchmarking standards and open-source workloads to test and train our software. 

Editor’s note: Open-source workloads, in this case, are publicly available quantum programs and benchmarks used by researchers and developers to evaluate and improve quantum computing systems.

What Sets Commutator Studios Apart From Competing Approaches?

Most quantum companies working on errors operate very close to the hardware. We do not consider them competitors but partners. Their models differ: they often require special access to the hardware, and their tools necessitate a deep technical knowledge that extends beyond the typical focus of most application developers. We see them as part of the stack. 

In contrast, we are creating a new category of developer tool for quantum programming, designed to achieve optimal performance at the application level. A helpful analogy is the game engine industry. Game engines enable developers to create visually stunning games without worrying about optimizing for specific platforms, such as PlayStation, Xbox, or iPhone. The engine handles that complexity. Similarly, our platform enables quantum developers to focus on their applications while we optimize the hardware in the background.

What Advice Would You Give to Fellow Deep Tech Founders?

Fall in love with the problem, not the solution. I fell in love with this problem long before I knew how to solve it, and that has kept me going. Once you are committed to the problem, focus on the capabilities, insights, or “secret sauce” you have that gives you leverage. 

I always struggle when people say they are looking for an idea to start a company. Problems to solve are everywhere!