You type the problem. ROX OS does the rest.
No circuits. No SDKs. No vendor lock. You describe what you need: ROX OS reads every available machine, builds the job, predicts the result, runs it, and proves it.
The way in eight steps
Describe the problem
A molecule, a graph, a QUBO, a Pauli list, a circuit file, or a record workload. ROX OS answers with a plan card.
Where
Simulator, free. Or a connected machine: IBM Quantum on your instance, IQM through Braket.
Set up the job
Name, budget cap, duel or ROX only, accuracy target. ROX chooses the rest and says why: every choice a line you can change.
Forecast
One row per machine: queue, machine time, what the machine carries today with a recommendation, sealed price, budget check.
Seal
The hash sets. Any change to this plan breaks the seal. If the forecast misses, the invoice is void.
Run
In queue behind n jobs. Running, s seconds of machine time. Analyzing raw counts. The ring turns only while something real happens.
Result: the record
READ · PREDICT · OUTPERFORM · HARVEST · PROVE. The number, then the sentence, then the proof.
Documents
Certificate, audit PDF, result JSON, raw counts. Recompute from raw counts. Share the record with your organisation.
One kernel. Five subsystems.
| Module | What it does | In numbers |
|---|---|---|
| PERFORM the scheduler | More result per QPU-second. Execution layer: every run stabilized and outperforming the vendor path. | up to 78 % more effective fidelity, twice the state surviving deep circuits |
| FORESEE the predictor | Know a machine before you pay for it. Sealed forecasts before you pay: backend ranking, launch windows, expected depth, deviation alarms. | 1.6 points mean error on untouched chips |
| PRISMA the device reader | Every chip has a noise fingerprint. We read it from data you already paid for. Diagnostics, acceptance testing, drift monitoring, zero machine seconds. | sibling chips told apart to two hundredths |
| HARVEST the recovery layer | Your old data becomes worth more. Continuous re-extraction of value from every run you have already paid for. | 15 to 17 points of state quality mined retroactively, zero QPU cost |
| PROVE the audit log | Numbers with a chain of custody. Preregistered runs, official job IDs, forensic audits, raw counts you can recompute in five minutes. | 140+ preregistered runs |
No competitor covers even two of these subsystems.
Two examples
VQE. The ground state of a 12-spin ring in two topologies, 432 two-qubit gates. ROX OS lands closer to the exact value than the vendor stack on every rung, 2.1× closer at the first rung. The same ring runs on the simulator for free.
QAOA. MaxCut on a 16-qubit chain, depth ladder to 230. Signal where the vendor default reads the coin flip.
Where it earns
Every industry below runs on the same two problem engines: combinatorial optimization and ground-state search. Our record in exactly these two: 140+ preregistered runs against the vendors' own stacks, at depths of up to 6,900 entangling gates, where default paths return coin-flip noise.
Biotech & Pharma
Drug simulation is ground-state search. We compute ground-state energies on real hardware, exact to measurement uncertainty: the number every chemistry pipeline is built on.
Materials & Industrial Chemistry
The hardest materials problems are frustrated many-body systems. On those, more than twice the quantum state survives with us, and our lead grows the deeper you go.
Finance
Portfolio, risk, allocation: optimization and ground-state search, back to back. Signal where the vendor path has already flatlined.
Transport & Logistics
Routing and scheduling at industrial depth: 6,900 gates in, still resolving structure, thousands of gates past the point where the default gave up. And 4 billion operations orchestrated through a single job, zero failures.
Cybersecurity
Certified quantum entropy: provably true randomness, yield per machine-hour maximized, with a forensic audit trail regulators can cite.
Who it's for
Six audiences. One kernel. First pilot customers don't buy a tool: they shape ROX OS.
You pay QPU bills
Fewer shots for the same answer. Your circuits stay untouched. Onboarding pays for itself out of proven savings.
You run a quantum platform
Your users get more results per QPU-minute, without you writing a line. Stacks with error-suppression tools. Members who see results renew.
You build quantum computers
Your machines, measured at their best. "ROX-ready" becomes a sales argument. Our advantage grows with your chip size: your roadmap is our multiplier.
You run HPC / research infrastructure
More science per machine-hour. Court-proof audit reports your committees can cite. Results count both ways: that's why they're believed.
You lead a sovereign program
On-premise. Licensed. No data leaves the country. Capability you cannot import: a predictive theory of quantum execution, patent-filed.
You study the next computing substrate
Run real hardware and simulator from one account. Learn what a machine can carry before you pay for shots. Every result carries a job ID you can cite.
Apps and machines
Custom (graph, QUBO, Pauli list), Circuit, Molecule and Record are live. Portfolio and Routing are visible tabs, coming with the next phase.
Machines: IBM Quantum ibm_fez, ibm_marrakesh, ibm_kingston, ibm_boston, ibm_phoenix, ibm_pittsburgh on your instance; IQM Garnet and Emerald through Braket; IonQ, Rigetti, AQT and Quantinuum in preparation; the simulator, always, free.
We call the shot first. Then we prove it. Every number is produced under preregistration: endpoints fixed in writing before submission. Standing offer: recompute any figure with us, live. Checking is allowed.