The thin part of a supply chain is a shape, and it can be watched.

Single qualified sources, concentrated capacity, queue positions and weather exposure are all visible in the intelligence layer before they are visible in a schedule or a price. Monitoring is what turns them into something you hear about.

Risk is a property of the connections, not of a company.

A facility is not risky on its own. It becomes risky when it is the only path from a build to a component it needs, when the long-lead item it makes has a lead time longer than the float in the schedule, or when the capability behind it exists at three sites worldwide. Every one of those is a statement about relationships, which is why it can be computed rather than assessed.

This is the argument for resolving the sources first. A risk register maintained by hand goes stale between reviews and only ever contains what someone thought to write down. An exposure derived from the model is recomputed whenever any of its inputs move, and it finds the concentration nobody had thought to look for - which is exactly what the Console is for.

Most disruptions announce themselves in public, quietly.

A permit is amended. A vessel diverts. A queue position slips. A storm track shifts two degrees and now crosses a corridor that three builds depend on. None of this is secret, and none of it arrives as an alert - it arrives as a small change in a document or a feed nobody on your team is reading that hour.

Nuclir watches the sources it already ingests for exactly those changes, resolves them against the entities they affect, and recomputes the exposures downstream. What reaches you is not the document; it is the sentence that says a component you depend on now has one qualified source instead of two. Where a change matters for export exposure rather than capacity, it is flagged as that instead.

Six exposures the model can express.

Each of these is derived from relationships the intelligence layer already holds, which is what makes it recomputable rather than a field someone has to remember to update.

Single qualified source

A component with exactly one path to a supplier able to deliver it, and the builds standing behind that one path.

Lead-time exposure

Long-lead equipment whose procurement window is longer than the slack left in the programme that needs it.

Capacity concentration

Work that depends on capability held at very few sites, regardless of how many companies claim to offer it.

Power availability

Queue positions, curtailment and headroom against the load contracted for it, which is where a compute schedule actually breaks.

Geographic and export exposure

Where a dependency sits, and where that placement raises a control question before it raises a commercial one.

Weather and climate exposure

Storm tracks, heat and hydrology crossing the corridors, plants and campuses a position depends on.

Watch, resolve, recompute, tell you.

The loop is short on purpose. A finding that arrives a quarter after the change that caused it is a report, not a monitor.

Watch
Every source behind your assets, routes and counterparties is re-read on its own cadence, and diffed against what it said last time.
Resolve
A change is attached to the entity it is about, not to the document it appeared in - which is the only reason a filing under a subsidiary's name can raise a flag on the parent you depend on.
Recompute
Exposures derived from the layer are recalculated: source counts per component, capacity concentration, queue movement, route and weather dependency.
Notify
What crossed a threshold you care about reaches you with the source, the moment, and the assets it touches attached - by webhook or in the Console.

We report what the record says, not what we predict.

There is no forecast here, no risk score sold as an oracle, and no non-public information. Every finding traces to a source you could read yourself, with the moment it said so - the same provenance rule the rest of the system runs on. A monitoring product that cannot show its working is asking a committee to take schedule or capital risk on faith.

It also does not replace your own process. It tells you where to point it: which components have one path, which queue positions are moving, and which of your positions share a dependency you had not noticed they shared. The decision, and the audit trail behind it, stays yours - which is why it is available through the API as well as on screen.

Evidence that keeps pace with the decision.

Evaluate Nuclir against the systems, markets, and decisions that matter to your organization. The result is current intelligence with the context needed to use it responsibly.

Request access Speak with the team