Web Analytics Made Easy - Statcounter
Skip to content

Process Units

Process units are reusable definitions for the workflows you want to monitor. They can represent an AI agent, tool path, onboarding flow, support escalation, claims process, order flow, or any operational sequence.

What Process Units Do

A process unit groups related events, traces, metrics, evals, and business fields into one monitored operational object. It gives Anosys the context needed to evaluate health, detect bottlenecks, and route alerts.

When To Use Them

Use process units when you care about an outcome, not just a single trace or metric. Examples include “ticket booking agent,” “fraud review,” “checkout recovery,” “Claude Code errors,” or “Kubernetes incident triage.”

Prerequisites

  1. Ingest events with stable workflow identifiers.
  2. Decide the start and end condition for the process.
  3. Define health fields such as SLA, success state, quality score, cost, owner, and customer impact.

Step-By-Step Setup

  1. Open Process Units.
  2. Create a unit and choose a type.
  3. Add filters that identify the workflow.
  4. Define success, failure, warning, and timeout conditions.
  5. Add params such as team, SLA, customer tier, model, or route.
  6. Connect alerts or dashboards.
  7. Test with historical events.

Example Definition

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
name: support-escalation-flow
type: workflow
match:
  workflow: support-escalation
success_when:
  status: resolved
warn_when:
  duration_minutes: "> 15"
alert_when:
  eval_score: "< 0.80"
owner: support-ai

What Appears In The Console

The console shows process-unit lists, status, last updated time, parameters, alerts, related traces, copied definitions, edits, and run history.

Common Mistakes

  • Defining units around implementation details instead of business outcomes.
  • Missing start/end states.
  • Tracking too many fields before the process definition is stable.
  • Not connecting units to alerts or dashboards.