Support, risk and quality

When a highlight is tagged “security concern,” create a Linear ticket flagged urgent. Attach the verbatim highlight, affected account, and related context. Notify the security lead in #security immediately.

Every day, scan new support data for GDPR, CCPA, data deletion, data export, or right-to-be-forgotten language. For each match, create a tracked record with account, request type, verbatim quote, and response deadline. Notify the legal team.

Every Monday, pull support response time data from Hex (or from support data ingested into Dovetail) for the past 7 days. Cross-reference against customer sentiment in Dovetail. Flag accounts where slow response correlates with negative sentiment. Post to support leadership in #support-leads with the pattern and suggested triage.

When a highlight is tagged “legal concern,” create a doc in the pre-configured Legal and Compliance project (access on that project is owned and set by humans, not the agent). Include verbatim quote, account, context, and any related prior mentions. Notify the legal team in #legal-alerts.

Every night, scan highlights tagged in the past 24 hours. Flag any tag applied outside the approved taxonomy. Post a review list to the research ops lead in the morning. Flag only — never auto-retag.

Every Monday, find Linear tickets closed as “Fixed” in the past 7 days. Match each to original customer reports in Dovetail. Draft a personalized Gmail notification for each affected customer. Route to the account owner for review before sending.

When @mentioned with a bug description, search Dovetail and Linear for prior reports. Return: is this known (yes/no), Linear ticket link, current status, and workaround if any. If unknown, offer to create a new tracked report.

When a highlight tag spikes 3x baseline in 24 hours, pull Hex data for related feature usage in the same window. If usage also spiked, reports may reflect attention. If usage is normal but reports spiked, it’s a real quality issue. Post the interpretation to the PM.