Skip to main content

​
How It Works

When Squasher receives an error, it runs the error through an AI analysis pipeline. The error type, message, stack trace, and surrounding context are analyzed and a structured result is attached to the error group.
1

Error ingested

The SDK or log drain sends an error to Squasher.
2

Error grouped

Squasher generates a fingerprint from the error type, normalized message, and top in-app stack frame. The error is either added to an existing group or creates a new one.
3

AI analysis

Squasher’s triage engine analyzes the error and produces a structured result.
4

Results attached

The triage summary, severity, category, and suggested fix are saved to the error group and displayed in your dashboard.

​
Triage Output

Every triaged error includes:
FieldDescriptionExample
SummaryOne-line root cause explanation”API response returned null instead of expected array”
SeverityImpact assessmentcritical, high, medium, low
CategoryError classificationruntime, type, network, auth, config, unknown
Suggested FixActionable fix recommendation”Add null check: data?.items?.map(...)”
Likely RegressionWhether this is a new issuetrue / false

​
Cost

AI triage is included free on all plans — there is no per-seat charge for AI features. For comparison, Sentry’s AI feature (Seer) costs $40/seat/month.

​
Accuracy

Triage quality depends on the context available:
  • SDK errors provide the richest context: typed stack frames, breadcrumbs, user info, request details. Triage accuracy is highest.
  • Log drain errors provide message + severity + source. Triage is still useful but less specific without stack frames.

​
Disabling Triage

AI triage runs automatically on new error groups. To disable it for a specific project, go to Settings > AI Triage and toggle it off.