Observability

Zero measures your application without you installing an agent, configuring a collector, or learning a new tool. What it gives you are four readings.

📈Traffic and consumption

Requests, latency, error rate, CPU, and memory over time.

Open →📜Logs

Three sources the screen never conflates: application, build, and platform.

Open →🕓Activity

The organization's operation history.

Open →⚖️Resources and limits

What was reserved, the ceiling, and observed usage.

Open →

The distinction that runs through everything

Three kinds of silence reach these screens, and they call for opposite reactions:

What you read	What it means	What to do
Collection not configured	The platform is not measuring this	Waiting will not help
Collecting, no traffic	It is measuring and nobody visited	Widen the period
Collecting, with traffic	There is something to read	Read it

A single "no data" for all three would teach you to ignore the screen — and this is exactly where you come when something is wrong.

For the same reason, absence never becomes zero:

error rate with no requests ....... not measured, not 0%
percentile with no samples ........ not measured, not 0 ms
limit not declared ................ no limit, not 0

Automatic refresh

The traffic and consumption screens refresh themselves every 30 seconds. No faster: the finest measurement granularity is one minute, and asking more often would return the same answer.

There is a pause button for anyone comparing two moments who does not want the ground moving.

What the platform does not hand over

Internal monitoring dashboards and infrastructure credentials are not part of the product. Those tools exist and are operated by NNumbers — what reaches you is the finished reading, scoped to your organization and with secrets protected.

Next steps

• Traffic and consumption
• Logs
• What exists today