Quantum use-case discovery

Almost every company can find a slide that says quantum computing will transform its industry. Very few can name one of their own problems, describe how a quantum algorithm would attack it, and say which classical method it would have to beat. That gap is what use-case discovery closes.

The work is a series of workshops and follow-up analysis. We go through your operations with the people who run them, collect candidate problems, match each to a family of quantum algorithms, and then try hard to rule it out. What survives is a short list of quantum computing use cases that are specific to your company and honest about timing.

What counts as a quantum computing use case?

A quantum computing use case is a specific problem, with real data and a measurable outcome, where a quantum algorithm has a plausible path to beating the best classical method on the same problem. A vague goal such as “optimize logistics” does not qualify until it is narrowed to a decision you make.

By contrast, “Reduce empty kilometers in next-day routing for 400 trucks across three distribution centers” is a candidate.

Most candidates fall into one of four families.

Optimization

Routing, scheduling, portfolio construction, network design and other combinatorial problems. Quantum approaches include QAOA on gate-based machines and quantum annealing. This is the family businesses ask about most, and the one where the evidence of advantage is weakest so far. Classical solvers such as Gurobi, CPLEX or OR-Tools, plus good heuristics, are very strong on these problems.

Simulation of molecules and materials

Reaction energies, catalysts, battery materials, drug binding. Classical methods approximate the quantum behavior of electrons, and some systems (the nitrogen-fixing enzyme FeMoco is the textbook example) are hard to approximate well. This is where most researchers expect the first useful advantage, but it will likely need error-corrected machines.

Machine learning

Quantum kernels, quantum neural networks and related methods. Results so far are mixed. Some proposed speedups were later matched by classical algorithms, as happened when Ewin Tang showed in 2018 that a classical method could match a well-known quantum recommendation algorithm. Loading large classical datasets into a quantum computer is also slow, which cancels out many theoretical gains.

Sampling and Monte Carlo

Risk simulation, option pricing and probabilistic models. Quantum amplitude estimation offers a quadratic speedup in theory. The catch is that quadratic speedups need very large, very fast fault-tolerant machines to pay off; a 2021 analysis by Google researchers (Babbush and colleagues, PRX Quantum) argued they are unlikely to help on early error-corrected hardware.

How do you judge quantum advantage against classical baselines?

Quantum advantage is judged by comparing the quantum result with the best classical method someone has seriously tried on the same problem, at the same size, with the same time and cost budget. A comparison against a naive algorithm or a shrunken problem is a demo, not evidence of advantage.

That sounds obvious, and it is where most vendor claims fall apart. Benchmarks that ignore queue time and error mitigation overhead tell you very little either.

The public record shows how fast claims move. Google’s Willow chip, announced in December 2024, ran a random circuit sampling task in under five minutes that Google estimated would take a leading classical supercomputer about 10 septillion years, but the task has no commercial use. In July 2026, IBM and research partners including Algorithmiq, Qedma and the University of Chicago announced three candidate advantage demonstrations. The papers were arXiv preprints, and the Quantum Advantage Tracker, an open, community-led effort that IBM contributes to, exists to compare such claims with the best classical responses. The tracker keeps a separate list of superseded candidates, where improved classical methods closed the gap.

For your problems, we set the baseline first. If you don’t already have a strong classical solution, building one is often the best use of this year’s budget, and it is also the bar any future quantum approach must clear.

Filters that remove most candidates

We apply the same filters to every problem, and we write down which one removed it:

1. Does the problem have the structure a known quantum algorithm exploits, or does it only sound hard?
2. Is the current classical approach good, or is the real gap data quality, process or tooling?
3. How large does the problem need to be before quantum could help, and how many logical qubits and operations would that take?
4. Can the input data be loaded efficiently, and can the answer be read out in a useful form?
5. Would a better answer be worth enough money or risk reduction to justify the effort?
6. When could hardware of the required size plausibly exist, and what would change that estimate?

Most candidates stop at the first or second question. That is normal and useful. A problem that fails because your current solver is weak goes into a separate list of classical improvements.

How do the workshops find quantum computing use cases?

The use-case discovery workshops find candidates by asking business owners about decisions rather than technology, then reviewing the top candidates with your technical staff. Between sessions we map each problem to a quantum formulation, estimate the resources it would need and check published results, so the readout covers both the shortlist and the discards.

The first workshop is with business owners: operations, planning, risk, R&D, whoever owns decisions that involve large search spaces or physical simulation. We ask about decisions, not technology: what do you decide every day or every month, how long does it take, what data goes in, and what would a 2% better answer be worth?

The second is technical. With your data scientists, operations research staff or engineers, we look at how the top candidates are solved today, which tools they use and where they hit limits. Then we do most of the work between sessions: mapping each problem to a formulation (QUBO, Hamiltonian, circuit model), estimating resource requirements and checking the published literature for comparable results.

We finish with a readout that walks through the shortlist and the discards. Everything is available in English or Spanish.

What you do with the shortlist

Each shortlisted problem ends with one of three recommendations. Watch it, with named signals that should bring it back to the table. Improve the classical baseline now. Or run a scoped experiment, which is what our hybrid quantum proof of concept is for.

If you also need to handle the cryptographic side of quantum risk, the quantum readiness assessment covers both in one engagement. Teams that want to run these evaluations themselves later can start with our training for technical teams. For a sense of where the candidates usually are by sector, see logistics and transport or healthcare and pharma.

AndesQubit is opening engagements in stages. If you have a problem you suspect might be a quantum problem, describe it to us. We’ll tell you plainly whether it is worth a workshop.