NOWFLOW
Owns the workflow
The request, the agents and tools, conditions, approvals, handover to machines, and the debrief.
Product overview / for technical teams
NowFlow runs AI agent workflows. NeuraOS runs AI on robots and edge devices. QFlow Studio runs quantum experiments. This deck shows what each owns.
The question every buyer asks
Our answer is a clear split of who approves, which system runs the work, who keeps it safe, and what record remains. Each product owns one part.
So we start with the handover points, then the features.
Who owns what
They can be used together on one programme, but each is a separate product with its own job.
NOWFLOW
The request, the agents and tools, conditions, approvals, handover to machines, and the debrief.
NEURAOS
What runs on the robot or edge device, its trust chain, deterministic safety, and recovery.
QFLOW STUDIO
Circuit, code, route, approval, run, result, and the copy a reviewer gets.
How work moves through
This is the pattern all three products follow. It does not mean they ship today as one integrated system.
Write down the goal, its owner, the systems allowed, and the safe stop.
Connect the agents, tools, rules, approvals, interfaces, and recovery steps.
A named person releases one versioned package. The device can still refuse.
Keep the request, the decision, what went wrong, the recovery, and the proof of completion.
Product 01 / NowFlow, available today
NowFlow turns a request or an event into a visible graph. Each agent has a set scope, tools are approved, and every approval, checkpoint, and failure path is drawn in.
Available product. Hosting, identity, data retention, model providers, interfaces, security, and support are agreed per customer environment.
NowFlow / what the graph shows
Every router, condition, branch, loop, approval, error, and recovery step stays attached to the run, so you can see why it did what it did.
An operator request, a structured event, an API call, a sensor, or a fleet exception.
Review, correct, release, pause, override, and escalate are steps with a named owner.
Acknowledgements, progress, exceptions, recovery, and the closeout, all linked.
NowFlow / the seven stages
Simulation and review inform the decision. A named person still has to give permission.
Product 02 / NeuraOS, public reference build
NeuraOS is an operating-system foundation for robots and edge devices. AI runs on a trusted base, beside a safety layer that is tested separately.
Reference build rc.8, tested in simulation. Physical use is not yet cleared. Motion and emergency stop stay outside the AI.
NeuraOS / who can say no
Three things are kept separate on purpose: permission for a workload, the device's own safety, and the human in command.
Each workload has an identity, minimal rights, safe updates, and an audit trail.
Limits on space, time, sensor validity, and health, plus override and safe-stop.
AI may recommend and plan within limits. It cannot approve itself or widen the mission.
Product 03 / QFlow Studio, live web app
QFlow Studio is a web app for designing quantum circuits and learning. It keeps the generated code, the simulator or provider used, the approval, and the outputs.
Routes to real hardware are labelled beta, preflight, test, or planned. Specialized routes and qOS are still in development.
QFlow Studio / the record
The record links the question, the circuit, the code, the environment, the route, the outputs, the failures, the limits, and the verdict.
State the question, the route, the credentials, the device context, and the baseline.
Check the generated code and provider status before sending it to hardware or sharing.
Keep the counts, exports, limitations, reviewer notes, and the next decision.
Where each product stops
A programme can use one of them, all three, or mix in other systems. At every join, it must be clear who owns what and what gets recorded.
The workflow graph, agent and tool scope, approvals, handover to machines, exceptions, debrief.
What runs on the device, local inference, deterministic safety, the trust chain, recovery.
The quantum workflow, circuit and code, route status, run history, the reviewer's record.
Acceptance criteria, certification, the safety case, interfaces, operations, final authority.
Where to start
One small, real workflow is enough to test the fit.
How an evaluation runs
The first review shows control and a clean record on one small but real workflow.
Roles, interfaces, data, identity, retention, network, safety, and who decides.
Normal, refusal, degraded, recovery, and stop cases, plus the record required.
Watch the behaviour and the records, without widening what the system may do.
Widen the scope only when the record supports it.
Next step
We will tell you which product fits, map the interfaces, and design the smallest evaluation that ends with a record you can review.
neuraparse.com/products