Frequently asked questions

NowFlow governs agentic work across people, approved tools, systems, and machine interfaces. QFlow Studio connects quantum workflow design, route choice, generated source, runs, and reviewer-safe evidence. NeuraOS places approved AI and robotics workloads close to hardware under human authority and an independent safety envelope. They can be composed for a bounded programme, but none is a universal dependency for the others.

NowFlow follows a command, build, validate, operate, and observe sequence. Teams capture the intent and authority boundary, compose the visible graph, test normal and failure paths, release consequential action through a named human, and retain checkpoints, decisions, tool calls, exceptions, and outcomes as run evidence.

A workflow can use configured tools, approved APIs, data tables, knowledge sources, system events, and authorized robot, fleet, mission-system, or edge interfaces. Every connection remains subject to the credentials, identity, network, retention, model-provider, deployment, and support boundaries agreed for the target environment.

Each agent can carry one purpose, instructions, model, scoped tools, knowledge references, and a reusable versioned place in the graph. Agents may prepare, recommend, decompose, and route inside those limits; named people retain review, correction, release, pause, override, and escalation responsibilities.

No. NowFlow should not be described as a beta product. The product and sign-in surface are available; production topology, identity, retention, model-provider, support, security, and machine-interface commitments still require a bounded review for the target environment.

No. NowFlow can govern the higher-level task, approval, handoff, exception, and evidence route. Robot controllers, flight or motion control, emergency protection, certified safety functions, command authority, and weapons-release decisions stay outside its authority, and device-local safety can still refuse or pause an approved task.

QFlow Studio uses a design, route, code, and evidence flow. It keeps the brief and visual circuit connected to a local simulator or status-labelled provider route, generated Qiskit, Cirq, or OpenQASM, a human submit decision, run state, results, exports, and a reviewer-safe evidence record.

No universal provider entitlement is implied. Local simulation is available, while credentialed routes carry explicit beta, validation, preflight, credential-test, or planned labels. Provider accounts, credits, pricing, regions, queues, devices, and service terms remain controlled by each independent provider and should be checked in the live status matrix.

No. AI assistance may draft a circuit, source, route note, or evidence context, but a person remains responsible for verification, credentialed submission, and sharing. Credentials and unrelated private workspace data stay outside reviewer-safe evidence records.

NeuraOS is a governed edge operating platform for robotics and defence systems. It places approved perception, sensor-fusion, planning, coordination, and inference workloads close to hardware while keeping signed release identity, accountable human authority, deterministic command arbitration, independent safety, recovery, and evidence explicit.

NeuraOS work is scoped, prepared, validated, authorized, and operated. Intended and prohibited use, data and software lineage, threat paths, nominal and adversarial tests, signed release identity, residual risk, recovery, monitoring, and evidence expiry are reviewed before and after a workload reaches the edge.

No. The public repository is a proprietary product-capability, architecture, governance, technology, and visual record. It currently distributes no NeuraOS source code or binary release. Named third-party components keep their own licences and source obligations.

NODERIQ is a separate applied research and productisation programme for verifiable robotic workflows, shared context, heterogeneous-device coordination, edge resilience, human authority, and physical completion evidence. It is not presented as a released product, completed customer deployment, certified capability, or universal fleet platform.

Start with one operating question and one bounded path. Name the owner, input, expected output, human authority, system interfaces, local safe state, failure case, acceptance evidence, and stop or scale decision. Then select only the product, service, or research component required for that path.

Public pages describe product architecture, intended controls, and review routes, not a universal production topology or certification. Identity, network, hosting, model providers, credentials, data residency, retention, audit, support, safety, and acceptance requirements are verified for the actual environment before a production commitment.