Reading the substances register

The substances register is four views of the same data, in the order someone actually works in:

Tab	The question it answers
Summary	What needs attention, and where do I go?
Inventory	What do we hold, and where?
Readiness	Is our data good enough to assess?
Controls	What do the regulations require of us?

Readiness and Controls appear for New Zealand organisations only. The threshold engine encodes the NZ Hazardous Substances Regulations, and hazardous-substances law differs in structure between countries — not just in wording — so applying New Zealand thresholds elsewhere would be worse than showing nothing. The inventory itself works everywhere.

Summary — the register in five numbers

Summary is deliberately not an introduction page. Static explanatory text is read once by whoever set the system up and skipped by everyone after, and it drifts: thresholds change, the copy doesn’t, and a page explaining the rules becomes a page misdescribing them.

Instead it teaches through your own register:

3 substances can’t be assessed — their classification records a hazard class family rather than a category, so no quantity threshold matches them.

That sentence explains what Readiness is, why it exists and what to do, at the moment it’s relevant — and it removes itself when the count reaches zero.

Every block that has nothing to say disappears. A healthy register is a short page. If you see only “19 substances across 3 sites” and a line saying nothing is outstanding, that is the system working, not the system failing to load.

The one block that doesn’t disappear

Confidence stays until every threshold table has been checked against the Regulations:

2 of 4 threshold tables verified against the Regulations.

The provenance of our data differs per control. Signage and emergency response plan figures were read from the consolidated Regulations; others came from WorkSafe guidance and have not yet been line-checked. Presenting both with equal authority would be a claim we can’t support, so the page says which is which and asks you to treat the rest as indicative.

Readiness — whether your data can be assessed at all

This tab exists because of a failure that hides itself.

The inventory records 5.1.1. The schedules set thresholds against 5.1.1A, 5.1.1B and 5.1.1C — which require an emergency response plan at 50 L, 500 L and 5,000 L respectively. A hundredfold spread, and nothing in the record says which.

So every threshold check silently misses, and the controls report comes back empty — which reads exactly like good news. Readiness is what makes that visible: it lists the substances whose classification is too coarse to assess, and shows what each candidate category would require, so choosing between them shows its consequence rather than reading as a tidiness request.

A substance with no classification at all is a related but different problem: it contributes to no threshold whatsoever. That one shows on Summary and colours the substance amber on the register.

Both answers live in section 2 of the Safety Data Sheet.

Not every unmatched classification lands here. Classes 6.3, 6.4, 6.5, 6.8 and 6.9 carry no threshold in these Regulations at all — flagging those would be a false alarm, so they are left alone.

Controls — what your holdings require

See The controls report for exactly what is and isn’t calculated. Two things worth knowing here:

Controls are counted per place, not per substance. The thresholds apply to how much is held in one location, so the report groups your holdings by site and location label and assesses each group separately.

Any control can become an Action. Each triggered control carries a Raise action button, which opens the action wizard pre-filled with the control, the place it fired, the quantity that triggered it and the provision it comes from. A control nobody is assigned to is a control nobody is doing — this is how a finding becomes someone’s job with a due date against it.

What SteadyBot may and may not say

Summary carries a What to do first panel. It is worth being precise about the division of labour, because it’s a hard line:

	Rules engine	SteadyBot
Which controls apply	✅	❌ never
Threshold figures	✅	❌ never
Regulation citations	✅	❌ never
Whether a substance can be assessed	✅	❌ never
What to do first	❌	✅
Grouping work sensibly	❌	✅
Plain-English framing	❌	✅

SteadyBot is given the engine’s findings — each already carrying its control, threshold and citation — and never your raw quantities. It cannot compute a threshold, because the numbers to compute one from never reach it. Its job is to order the work and group what can be done in one go: “three of these sheets come from the same supplier, so one email covers it” is a judgement it is good at and a rules table is not.

It can decline. If the register is too incomplete to prioritise, the honest answer is “there isn’t enough here to advise on yet, start by classifying the six substances on the Readiness tab” — and that is what you’ll get, rather than manufactured insight.

Nothing is generated until you ask. Every figure on Summary is computed as the page loads and is always current. The plan is not: it costs money and takes seconds, so it is drafted when someone presses the button, saved for everyone in the organisation, and stamped with when it was written and which threshold dataset it was written against. If the numbers have moved since, press the button again.

Each line can become an Action, the same as a triggered control — so a plan turns into assigned work rather than something to read and forget.