Krenn·IQ provides two tools — Atlas (quantum compute-cost triage) and CAPAS (claim-admissibility gate) — as a hosted web application and as a downloadable open-source engine. The open-source components are licensed under the Apache License 2.0; see the LICENSE and NOTICE files in each repository.
Atlas produces an operational diagnostic: an estimate of the classical-simulation cost of a circuit and a recommended compute route, with calibrated confidence. It is not a proof of classical simulability, a proof of quantum advantage, or financial/investment advice. Verdicts marked with reduced confidence are explicitly flagged as such. Decisions you make using the tool — including spending on quantum hardware — are your own responsibility.
The hosted web is provided for evaluation and individual use, subject to fair-use rate limits that may change. Do not attempt to disrupt the service, circumvent rate limits at scale, or systematically scrape the API to reconstruct the engine. For production or integrated use, use the downloadable engine or an Enterprise agreement.
The service and software are provided “as is”, without warranty of any kind, to the maximum extent permitted by law. The operator is not liable for any direct or indirect damages, losses, or decisions arising from use of the tool. Hosted availability is best-effort (no SLA) unless a separate Enterprise agreement states otherwise.
The hosted web currently operates without user accounts and does not require you to provide personal data to run an analysis. When you submit a circuit (OpenQASM), it is processed in memory to compute the diagnosis and the result is returned to you. Standard, transient web-server metadata (e.g., IP address for rate-limiting and abuse prevention) may be processed but is not used to build user profiles.
Some optional features (e.g., natural-language interpretation in the Sponsor tier) send the structured numeric outputs of the engine to a third-party LLM provider (Anthropic) for interpretation. This is opt-in and clearly indicated in the UI before it happens; the free/local tier uses a local model with no external calls. The downloadable engine makes no external calls except an optional, documented license/calibration check.
Privacy questions: fomv9354@gmail.com.
We treat every submitted circuit as confidential intellectual property. It is processed ephemerally (in memory, for the duration of the request) and is not persisted to disk or any database by the hosted service.
For defense, pharmaceutical, financial, or any work where circuits cannot leave your network, use the downloadable engine: it runs entirely on your hardware, makes no analysis calls to our servers, and your circuits never transit the public internet. This is the recommended mode for data-residency-sensitive or classified workloads.
The hosted demo runs on cloud infrastructure in the United States (Azure, East US). It is intended for evaluation. For data-residency requirements (e.g., EU/GDPR, or in-country mandates), use the local/on-premise engine, or contact us about a regional or on-prem Enterprise deployment.
Atlas is calibrated against measurements taken on publicly accessible quantum hardware. Those calibration datasets are part of the product; your own circuits are never added to them.
This document states the project’s current policies in good faith and in plain language. It is an initial v1.0 and may be updated; material changes will be reflected in the version and date above. For a formal contractual agreement (Enterprise), a tailored MSA/DPA is provided on request.