Signals converge.
Read one at a time, they hide it.
A delivery velocity decline of eight percent is normal variation. The same decline alongside rising technical debt, falling platform reuse and rising integration complexity is a structural emergency. Individual thresholds do not capture that. This does.
Open the dashboardA reading, not a record.
It asks you for nothing.
This is the only tool in the operating system that collects no data of its own. It has no form and nothing to fill in. It reads what two tools you already use have recorded — the Signal Response Summary for the six signals, and the Productive Capacity Ratio for the economics layer — and reports what they say when they are read together.
Its scope is deliberately narrow: the continuous, quantitative, capital-bearing measures. The mandate is trigger-based and rarely moves. Bridges are project-scoped and conditional. Escalations are a process with their own record. Each of those is complete where it lives, and consolidating them here would produce a longer page rather than a sharper reading.
Two of its findings exist nowhere else, and they are the reason it is worth opening at all. One spans both source tools — poor AI performance, rising technical debt and a falling productive capacity ratio together mean AI investment is not creating leverage, and no single tool's surface can see all three. The other spans time: a signal that has stayed confirmed-adverse across two quarterly cycles has stopped being a signal and become the operating model.
The named combinations
Six combinations of signals that the manual names explicitly, each with the diagnosis it assigns — platform failure rather than delivery failure, a hidden maintenance load, a data or workflow problem rather than a platform one. When one is present, you see which states matched it and why.
The basis for every reading
Each matched condition shows what it matched on, labeled by where the reading comes from: grounded in the manual's own threshold, or inferred by this dashboard where the manual is silent. The inferred ones say so on the card, name the silence, and state the reading chosen — so you can disagree with it.
The six signals, at their true status
Each signal's current standing, escalated across your recorded history rather than read off the latest quarter alone — because a signal's persistence is part of its status. A signal you have not recorded reads as no data, never as steady.
The economics layer
The productive capacity ratio and its quarter-over-quarter movement, alongside foundational investment protection — what was confirmed, what actually happened, and how many initiatives were blocked by platform state.
The facts an escalation would rest on
Which signals are at act-now, the foundational variance, the blocked-initiative count. Shown as inputs, not scored into a risk rating — the manual assigns no weights, and inventing some would present an opinion with the authority of a calculation.
What it cannot tell you
Where a reading is not yet determinable, it says so and says why, rather than reporting the finding as absent. Those two are different answers, and the difference is usually the one that matters.
It will not tell you
the system is healthy.
There is no overall health score, no risk rating and no single “next action” verdict. Not because they would be hard to compute — they are trivial to compute — but because every one of them would require weights the manual does not give, and a number derived from invented weights is an opinion wearing the costume of a calculation.
The manual itself calls its six combinations examples of the diagnostic logic, not an exhaustive list. So when none of them is present, this dashboard says exactly that — no named combination is currently detected — and counts the individual signals that are adverse. It does not say you are fine. It cannot know that, and the place where a dashboard is most tempting to bluff is the place everyone looks first.
The standard it holds to is simple: never emit a verdict whose basis you cannot see and argue with. That trades decisiveness for trustworthiness on purpose. An instrument whose only value is that it can be trusted does not get to be punchy at the cost of being defensible.
Before the monthly cadence. The monthly session is where individual observations become structural diagnoses. Opening this first tells you which diagnoses the signals are already pointing at, and which are not yet determinable.
Before a quarterly capital conversation. The escalation facts and the economics layer are the inputs that conversation runs on — the reds, the foundational variance, the blocked initiatives — and each combination carries the cadence the manual assigns it, so nothing arrives at the quarterly session as a surprise.
After recording a quarter in either source tool. A new reading can complete a combination that was previously undeterminable, or push a confirmed signal past the point where it stops being a signal.
It is not a weekly surface. The signals it reads move on quarterly records, and checking it more often than they change tells you nothing new.
You have already recorded it.
This is what it says together.
Nothing to fill in. If you have recorded quarters in the Signal Response Summary or the Productive Capacity Ratio, the reading is there now. If you have not, the dashboard will say so plainly and name what to record first.
Open the dashboard