Priority Scoring
Score message urgency as low | medium | high | critical. Runs a purpose-built transformer model in under 100ms. Costs $0.03 per score.
Two ways to use priority
1. Standalone: Call POST /api/ai/priority to score any message ($0.03).
2. Combined: Pass include_priority: true to the classify endpoint to get intent + priority in one call ($0.20 + $0.03).
Standalone Priority Scoring
Blanket Priority Rules
Blanket rules fire for any message matching a priority level, alongside intent-specific routing. One rule per priority level per org. Use these for cross-cutting actions like paging on-call for critical issues.
Per-intent priority filtering
In addition to blanket rules, you can add a priority_filter to any routing rule. When set, the rule only fires when the message's priority matches one of the filter values.
Example: Set priority_filter: ["high", "critical"] on a bug_report routing rule to only create Linear issues for urgent bugs.
See the Routing API for the priority_filter parameter.