• Misunderstanding of the non-emergency scope

• Incorrect expectations about automation replacing dispatch

• Staff confusion about conversational vs menu behavior

• Misinterpretation of safety bias

These are training or policy alignment issues.

• The system behaves outside the documented safety model

• Intent routing appears to bypass safety unexpectedly

• Guardrails fail to activate when clearly required

• Single safety escalation

• Policy disagreement about over-transfer

• Staff concerns about conservative behavior

• Misinterpretation of the safety transfer reason

• Repeated safety escalations without trigger language

• Escalation fails to trigger when obvious risk language is present

• Sudden spike in safety transfers without configuration changes

• Safety override appears inconsistent across similar calls

• Intent misclassification

• Overlapping intents

• Conditional logic errors

• Questions asked out of order

• Unnecessary transfers due to action selection

Review:

• Intent mapping

• Example utterances

• Action configuration

• Intent classification fails across multiple clearly similar calls

• Confidence scoring appears broken

• System ignores configured intent rules

• Changes do not persist after saving configuration

• Incorrect transfer number

• Missing fallback configuration

• Misaligned default action

• Unexpected transfer during business hours

Validate:

• Action settings

• Time-based routing

• Fallback logic

• Transfers route correctly in config but fail live

• Calls terminate unexpectedly without configured action

• Guardrails override properly configured action without visible trigger

• Incorrect business hour routing

• Holiday misrouting

• Feature flag accidentally enabled/disabled

• Unexpected automation after-hours

Validate:

• Schedule boundaries

• Time zone settings

• Feature flag states

• Routing ignores correct schedule settings

• Time blocks behave inconsistently

• Feature flag changes do not apply

• Multiple time-triggered failures across intents

• Minor transcription errors

• Occasional barge-in interruption

• Single LLM delay

• Thinking indicator confusion

Review:

• Transcript

• Timeline

• Retry events

• Repeated agent_failure transfers

• Widespread STT mishearing across many calls

• Multiple LLM timeouts in short window

• Barge-in instability across calls

• Provider outages suspected

• Single incorrect language detection

• Caller confusion during switch

• Occasional unsupported language transfer

Review:

• Language detection thresholds

• Switch cap settings

• Transcript language consistency

• Oscillation between languages repeatedly

• Auto-detection misfires across multiple calls

• Stack partially switches (STT changes but TTS does not)

• Language change tool fails to execute

• Single unexpected transfer

• Time-based routing confusion

• Intent-based transfer misunderstood

• Jurisdiction rule misapplied

Validate:

• Intent action

• Transfer number

• Time logic

• Location verification

• Repeated unknown transfer reasons

• Agent_failure spike

• Safety escalation spike without clear trigger

• Transfers routing to wrong number despite correct config

• Transfer loops not caused by telephony misconfiguration

• Single duration-cap transfer

• One-time timeout

• Minor retry event

Review:

• Timeline

• Duration thresholds

• Retry limits

• Frequent duration-cap transfers

• Retry loops

• Early timeouts across multiple calls

• Calls ending abruptly without clear lifecycle event

• Forwarding rule incorrect

• Return line misconfigured

• Caller ID mismatch

• Looping caused by PBX routing

• Adjacent PSAP number incorrect

Validate:

• Carrier forwarding

• Dedicated return line

• No forwarding on return line

• End-to-end test calls

• Clean telephony configuration but transfer fails

• Calls disappear mid-transfer

• No call reaches ANET despite correct forwarding

• Repeated fallback transfers without telephony explanation

Telephony must be validated internally before escalation.

• Missing required field

• Incorrect field mapping

• Misaligned incident category

• Draft fallback expected but misunderstood

Validate:

• Field mapping

• Required field list

• Workflow mode (Automated vs Suggested)

• CAD availability

• Repeated failed submissions with correct fields

• CAD submission succeeds but data corrupted

• Status in Triage does not match CAD record

• Integration works intermittently

• API authentication errors

Provide:

• Call IDs

• Timestamps

• Error messages

• CAD record examples

• Field mismatch

• Authentication credential expired

• Feature flag misalignment

• Integration returns structured errors repeatedly

• Data loss suspected

• External system rejects properly formatted payload

• Unexpected duplicate submissions

• Single anomaly

• One unusual transfer

• One low score

• Status misunderstood

Use:

• Transcript review

• Timeline review

• Score correlation

• Repeated anomalies of same type

• Dashboard status inconsistent with Triage

• Timeline missing expected events

• Data not populating

• Minor daily fluctuation

• One-day spike

• Seasonal volume shift

Compare:

• Weekly trends

• Prior periods

• Analytics data missing

• Metrics inconsistent with Triage data

• Large unexplained trend change without config changes

• Data aggregation delays persist

• Single low score

• Individual mis-tag

• Slight score fluctuation

Review:

• Transcript

• Transfer reason

• Intent clarity

• Score category appears broken across many calls

• Tags applied incorrectly across patterns

• Scoring logic not aligning with transcript

• No scoring generated

Escalate to Support when:

• The issue is repeated

• The issue crosses multiple calls

• The issue contradicts documented configuration

• The issue cannot be explained by telephony or configuration

• The issue suggests provider instability

• Data integrity may be affected

Do not escalate:

• Isolated calls

• Policy disagreements

• Training issues

• Expected safety behavior