# Zendesk — AI agent team ideas URL: https://openenvelope.org/tools/zendesk Zendesk is the central record for every customer issue. multi-agent teams built on top of it can triage tickets the moment they arrive, flag SLA risks before they breach, and keep your knowledge base from going stale — without adding headcount. Here are six teams worth building. ## Ticket triage team Classifies every incoming ticket by issue type, priority, and product area, assigns it to the right agent or team, and adds the relevant tags. Agents start every ticket already knowing what they're dealing with. Agents: - Ticket Classifier: Reads the ticket and determines type and priority - Assignment Router: Identifies the right agent or team - Tag Applier: Labels the ticket for reporting and routing ## Knowledge base maintainer After every resolved ticket, checks whether the resolution is documented in your KB. When it isn't, drafts a new article for review. Over time your KB reflects the issues customers actually have. Agents: - Resolution Analyst: Extracts the fix from the resolved ticket - KB Checker: Searches the KB for a matching article - Article Drafter: Writes a new article when the gap is confirmed ## SLA watchdog Monitors all open tickets against their SLA deadlines, sends escalation warnings to agents and managers at configurable thresholds, and reassigns tickets that are at risk of breach. You stop finding out about SLA failures after the fact. Agents: - SLA Monitor: Tracks time remaining on all open tickets - Escalation Notifier: Alerts agents and managers at the right thresholds - Reassignment Handler: Moves tickets at critical risk to available agents ## CSAT follow-up team After ticket resolution, follows up if no CSAT score is submitted, then synthesises negative feedback weekly and surfaces the most common themes. Your team stops guessing what customers are unhappy about. Agents: - CSAT Chaser: Sends a follow-up when no score is submitted - Feedback Collector: Aggregates negative scores and comments - Theme Analyser: Groups feedback into actionable patterns ## Repeat issue detector Spots clusters of tickets describing the same underlying problem, groups them, and notifies the product team with a structured summary. Bugs and broken experiences get escalated before they become a flood. Agents: - Cluster Scanner: Groups tickets by semantic similarity - Issue Synthesiser: Writes a clear description of the underlying problem - Product Notifier: Routes the summary to the right team ## Incident outreach team When a known incident is open, finds all customers in Zendesk likely to be affected, drafts a personalised status update for each, and queues them for agent review before sending. Proactive communication happens without manual list-building. Agents: - Affected Customer Finder: Identifies who is likely impacted by the incident - Update Drafter: Writes a personalised status message per customer - Send Gate: Human checkpoint before messages go out ## Related roles [support](https://openenvelope.org/roles/support), [operations](https://openenvelope.org/roles/operations) ## Related tools [slack](https://openenvelope.org/tools/slack), [intercom](https://openenvelope.org/tools/intercom), [freshdesk](https://openenvelope.org/tools/freshdesk) Browse all tools: https://openenvelope.org/tools