Less “what if.”
More “which one?”
Which van takes the delivery? Who works which shift? What fits in the shipment? Start with a decision your team already makes — then see the data, the constraints and the plan.
What decision are you trying to make?
Pick a familiar problem. Each example shows what goes in, what comes out and where the demo stops.
Small, transparent demos. The seven optimization examples use synthetic inputs from our runnable Lab library. The plans below are classical references or clearly labeled previews — not live QPU results, customer outcomes or claims of quantum advantage.
State the decision.
List the options, the limits and what makes one plan better than another.
Start with a known answer.
Small examples let us check every option. In Lab, compare quantum samples with that classical reference and check the constraints.
Change an assumption.
Rerun the example on the local simulator. Hardware is an optional next step, subject to circuit compatibility, availability and a separate credit quote.
Six deliveries. Two vans. Who takes what?
A dispatcher needs to share today's deliveries between two drivers without overloading either van or sending them across town unnecessarily.
You bring
Six delivery locations, a depot and a four-parcel limit per van.
You get
A stop list for each driver and a depot-to-depot route for each van.
Library → search Vehicle routing with QAOA → Open a copy in a new project → Run file.
- Van 1
- A · B · C
- Van 2
- D · E · F
- Capacity check
- 3 / 4 parcels each
- Route distance
- ≈ 18.7 km combined
Classical reference for the supplied coordinates, not a live quantum result. Distances are straight-line model distances.
Where quantum fits Technical details
QAOA proposes a two-way assignment on 6 qubits. Classical route enumeration orders each van's stops. The notebook compares the clustering proxy with actual modeled route distance and rejects overloaded vans.
Who should work on which shift?
An operations lead has pairs of people who should be scheduled apart. Some clashes matter more than others. Which split avoids the most conflict?
You bring
People A–F and a list of pairwise scheduling clashes, each with an importance weight.
You get
Two named teams, a score for the clashes avoided and a list of the clashes that remain.
Library → search Shift scheduling with QAOA → Open a copy in a new project → Run file.
- Shift 1
- A · D · E
- Shift 2
- B · C · F
- Clash weight avoided
- 15 of 18
- Still unresolved
- 3 weighted points
Exact classical reference for the demo's clash graph. Quantum samples are checked against this reference.
Where quantum fits Technical details
The weighted MaxCut model uses 6 qubits. The notebook checks all 64 assignments to establish the exact reference, then scores the trained QAOA samples.
Three jobs. One machine. What goes first?
A production planner has a four-hour window and three jobs with different durations and priorities. The machine cannot do two jobs at once.
You bring
Job A: 1 hour, priority 3. Job B: 2 hours, priority 1. Job C: 1 hour, priority 2.
You get
A start time for every job, with no overlap and a priority-weighted completion score.
Library → search Job scheduling with QAOA → Open a copy in a new project → Run file.
- 00:00 → 01:00
- Job A · priority 3
- 01:00 → 02:00
- Job C · priority 2
- 02:00 → 04:00
- Job B · priority 1
- Completion score
- 11 · lower is better
Exact reference schedule: 3 × 1 + 2 × 2 + 1 × 4 = 11. Times are relative to the start of the planning window.
Where quantum fits Technical details
An 11-qubit time-indexed model penalizes overlaps and missing or repeated jobs. Only 6 of its 2,048 bitstrings are legal schedules; the notebook reports feasibility as well as score.
Which generators cover a 95 MW demand?
An energy planner needs a combination of generators that meets demand without turning on unnecessarily expensive units.
You bring
Six generators, their fixed output and modeled operating cost; one demand target.
You get
An on/off list, total output, demand gap and a modeled operating cost.
Library → search Unit commitment with QAOA → Open a copy in a new project → Run file.
- Switch on
- Hydro 25 · wind 40 · solar 30 MW
- Leave off
- Two gas units · coal
- Supply / demand
- 95 / 95 MW
- Modeled cost
- 27 units
Exact classical reference for the toy inputs. Cost units are synthetic, not currency or an energy-price forecast.
Where quantum fits Technical details
QAOA uses 6 qubits to sample generator subsets. The objective combines a squared demand mismatch penalty with operating cost; all 64 subsets provide a classical reference. A penalty alone does not guarantee demand is met.
Five parcels. A 15 kg limit. What makes the cut?
A shipment has more cargo than capacity. Picking the most valuable single parcel first is not always the best way to fill the load.
You bring
A weight and value score for each parcel, plus the available capacity.
You get
A loading list that stays within capacity, total weight, total value and what stays behind.
Library → search Knapsack with QAOA → Open a copy in a new project → Run file.
- Load
- P1 · P2 · P5
- Leave behind
- P3 · P4
- Weight / capacity
- 15 / 15 kg
- Total value
- 35 points
Exact classical reference for the 15 kg fixture. Value is an arbitrary priority score, not a monetary estimate.
Can you pack a better load?
Choose parcels, change the limit, then compare. This widget checks all 32 combinations in your browser — classical calculation, not a quantum run.
Where quantum fits Technical details
Five selection bits and four slack bits encode the 15 kg fixture on 9 qubits. The notebook checks feasibility independently of the penalty score and compares QAOA samples with all 32 possible loads.
Which systems should we patch first?
An IT team wants the smallest patch set that touches every connection in a simplified system map. This makes the coverage trade-off visible before discussing a larger maintenance plan.
You bring
Eight systems and ten connections. A connection is covered when at least one of its endpoints is selected.
You get
A minimum-size patch list and an explicit list of connections still uncovered.
Library → search Patch management as minimum vertex cover → Open a copy in a new project → Run file.
- Patch set · 1 / 2
- web-frontend · vpn-gateway
- Patch set · 2 / 2
- database · file-server
- Systems selected
- 4 of 8
- Connections covered
- 10 of 10
One exact minimum vertex cover for the supplied graph. Covering a graph edge is not proof that an attack path is eliminated.
Where quantum fits Technical details
An 8-qubit QAOA model penalizes edges whose two endpoints are both unpatched. Exact enumeration establishes the smallest covering set; sampled solutions are checked for uncovered edges.
Choose three assets — with the trade-off visible.
An analyst wants to compare a small selection of assets rather than rank each one independently. Assets that move together can change the appeal of a combination.
You bring
Six synthetic asset categories, expected-return inputs, a covariance matrix and a requirement to select exactly three.
You get
A three-asset selection, a risk-adjusted objective score and a comparison with the exact reference.
Library → search Portfolio selection as a QUBO → Open a copy in a new project → Run file.
- Selected in this fixture
- Gold · tech · energy
- Not selected
- Bonds · telco · retail
- Selection constraint
- 3 of 6 categories
- Objective score · not a return
- 0.178 · risk penalty 1.0
Exact classical reference for synthetic inputs, not an investment recommendation. The objective score is not a percentage return; no live prices or promised returns.
Where quantum fits Technical details
Six qubits encode asset selection. The notebook reports how often the exactly-three constraint is met, then compares legal quantum samples with the classical reference. A high penalty score is not automatically a valid portfolio.
One project room for a distributed team.
Picture an engineering team working with external advisers: project discussion, documents and meetings need one shared place. Quantum Safe Rooms is a separate collaboration product, not a quantum optimization job.
You bring
A project room, invited participants and their devices.
You get
A workspace for the project, with the protection of each channel explained separately.
- Chat room keys
- ML-KEM-768 wrapping when set up
- Device sign-in
- ML-DSA-65 once registered
- Files
- Server-side AES-256-GCM at rest
- Meetings
- Gated entry · not media E2EE
Example workflow, not a customer deployment claim. Quantum Safe Rooms runs on separate infrastructure and does not spend RoRo credits.
What is protected Technical details
ML-KEM protects chat room-key distribution, not all file or meeting keys. These are classical post-quantum algorithms, not quantum key distribution. Browser key generation uses the platform CSPRNG; no certified quantum entropy claim is made.
There is more in the Lab.
Watch a key exchange.
Search the Library for BB84 key exchange to explore how an eavesdropper changes the error rate. A simulated protocol lesson, not a deployed communications link.
Compare two classifiers.
Search for Tumour screening: quantum kernel vs RBF for a research comparison on a public dataset. Not a diagnostic service or clinical validation.
Open a working example.
The Library includes source, inputs and explanations you can inspect and modify.
Browse RoRo LabChoose where to test the quantum step.
Explore simulators and quantum hardware through one submit_run() interface. These are optional execution targets, not evidence that the reference plans above were run on every machine. Check circuit compatibility and the quote before submitting.
Catalog snapshot from the live platform, August 2026. Availability changes hour to hour; a machine that is queued or in maintenance shows as busy. Orgs can carry negotiated pricing — the quote at submit is what binds.
Bring a decision. Change one constraint.
Start with a small, anonymized example: your options, a hard limit and the result you want to improve. We can walk through the model, compare a reference plan and discuss what a realistic pilot would need.
