How to Make Quantum-Classical Workflows Accessible

This invited talk by Jeanette Lorenz, head of the Department of Quantum Computing at the Fraunhofer Institute for Cognitive Systems IKS, argues that practical quantum advantage will only be reached through hybrid quantum-classical workflows made efficient and accessible to users. She reviews why near-term noisy devices and their algorithms have not delivered industrial value, why fault-tolerant quantum computing introduces new complications across the software stack, and why quantum hardware being orders of magnitude slower than classical hardware complicates simple speedup arguments. Surveying applications in molecular simulation, optimization, and machine learning, she shows that realistic use cases treat the quantum computer as a computational subroutine within predominantly classical HPC workflows, for example through problem decomposition in optimization or quantum-enhanced steps inside classical training loops. Using a capacitated vehicle routing example of emptying sensor-equipped waste containers, she walks through the many decisions a user faces, from problem encoding and algorithm selection to transpilation and runtime execution, and argues for abstractions and automation in the upper software stack, citing a decision-tree tool with a no-code web interface. She presents quantitative analyses of hardware noise, sampling overhead, and loss-landscape stability that indicate when a given algorithmic combination cannot yield advantage even on perfect hardware. The talk concludes with a taxonomy of hardware, system, software, and application-level benchmarking, stressing reproducibility and transparent reference implementations, with the QUARK modular benchmarking library and the Bench QC project as examples.