Skip to main content

Live tracing

The live tracing view shows runs that are happening right now alongside runs that have already completed, with the same UI for both.

Run counts at a glance​

The tracing dashboard surfaces run counts for the selected date range:

  • Total runs within the range.
  • In-Progress runs.
  • Completed runs.
  • Failed runs.

These counts update as runs move through their lifecycle.

Run status​

Every run has one of three statuses:

  • Completed
  • Failed
  • In-Progress

For In-Progress runs, the live view shows which parts of the application have completed, which have failed, and which are currently executing.

Future executions that have not yet started are not included in the In-Progress view — only steps that are currently running.

Trace flow visualization (tree view)​

For each in-progress run, a tree view represents the execution flow:

  • Nodes that have completed.
  • Nodes that have failed.
  • Nodes that are currently in-progress.

This is the right view to open when something appears to be stuck — you can see exactly which step is the current bottleneck.

Run Source column​

Every run has a Source column showing how it was triggered:

  • UI — invoked from inside SimplAI Studio (e.g., Playground / Test).
  • API — invoked over the REST API.
  • Embed — invoked via an embed widget on a hosted page.

Source is useful for filtering production traffic vs. developer / QA traffic.

Status filters​

Filter the table by status:

  • Completed
  • Failed
  • In-Progress

You can combine the status filter with date range, version, and users for tighter scoping.

Identifying who triggered a run​

Each row shows the user's email address (not the numerical user ID) — easier to scan and share across team conversations.

CSV download​

Everything visible in the table can be downloaded as CSV. The CSV maintains the same structure and columns as the on-screen table — copy it into a spreadsheet to triage, slice, or share with stakeholders.

What "trace" includes per run​

Opening a run row shows the run detail view with the trace — the same structure across the platform:

  • Step / span hierarchy
  • Inputs and outputs per step
  • Step durations and statuses
  • Errors when they occurred

For voice agents specifically, the trace also includes audio playback synced to the spans — see Voice Agent run history.

Ingestion usage in the trace​

Each run that touches a Knowledge Base shows ingestion usage attribution at the trace level — you can see how much KB ingestion was consumed by this specific run, which is the right place to spot KB cost surprises before they show up on the bill. Aggregated KB ingestion stats also appear on Analytics.