Use case developmentWorkshopVisualization widgetsObservability Chart

Observability Chart

The Observability Chart widget plots platform telemetry for a single resource over time, directly inside a Workshop application. You can use this widget to embed live operational metrics—such as execution counts, latency, and compute utilization—alongside the rest of your workflow, so that operators can spot trends, anomalies, and performance shifts for the entire workflow at a glance.

Module builders configuring an Observability Chart widget can:

Chart metrics for a function, action, or compute module.
Choose which metric to plot for the selected resource (for example, executions or P95 duration for a function).
Derive the visible time range from Workshop variables, and let users brush a range directly on the chart to update those variables.
Expand a chart for a closer look, or view it in Workflow Lineage to debug the underlying executions.

The screenshot below shows an operational dashboard built from several Observability Chart widgets: the execution counts and P95 duration for an Escalate Order action, alongside the same metrics for a triagePatient function. A single time range control at the top of the module drives all four charts.

An operational dashboard composed of four Observability Chart widgets, showing execution counts and P95 duration for an action and a function over the past seven days.

The metrics shown here are the same near real-time metrics surfaced elsewhere in the platform—see AIP observability metrics for more details. The Observability Chart widget lets you embed those metrics in a purpose-built Workshop application.

Configuration options

The widget setup panel is shown below, configured to plot executions for a triagePatient function over a variable-backed time range.

The Observability Chart widget setup panel, showing the resource type, resource, metric, and start and end time options.

Resource type

Select the type of resource whose metrics you want to plot. The remaining configuration options depend on this selection. Supported resource types are:

Function
Action
Compute module

Resource

Choose the specific resource—the function, action type, or compute module—to chart. Only one resource can be plotted per widget; add multiple Observability Chart widgets to compare several resources side by side.

Metric

Select which metric to plot for the chosen resource. The available metrics depend on the resource type:

Functions and actions:
- Executions: Success and failure counts over time.
- P95 Duration: The 95th-percentile execution duration over time.
Compute modules:
- CPU: CPU utilization over time.
- Memory: Memory usage over time.

Time range

The visible time window is controlled by two timestamp variables:

Start time
End time

Back these options with Workshop variables to control the range shown, for example, from a Date/Time Picker elsewhere in the module. These variables are also widget outputs. When a user selects a range directly on the chart, the widget writes the new bounds back to the Start time and End time variables. This lets you zoom the chart interactively and drive other widgets from the selected range.

Interactivity

Range selection: Dragging to select a range on the chart updates the Start time and End time variables, which zooms the chart and can filter or update any other widgets bound to those variables.
Show legend: Toggle a legend that labels each series in the chart.
Expand: Expand a chart into a larger dialog for closer inspection.
Run history: Select Run history to open the resource's run history in Workflow Lineage and investigate the individual executions, logs, and traces behind the metric.

Availability

Telemetry is only available while the underlying resource is reporting metrics. If the resource cannot be found, or a compute module is not currently running or has no active session, the widget displays a message in place of the chart rather than an empty plot.