Quickstart: AI-Powered Support in 5 Steps
From configuring your routing strategy to enabling zero-touch Auto-Resolution in under 10 minutes.
5-Step Support Setup
Configure Channels
Navigate to Helpdesk > Settings > Channels. Enable email (IMAP/POP), live chat widget (embed code), and web portal. Connect the Connect module for phone/SMS ticket creation. Set up incoming email rules for auto-categorization.
Set Up Routing Strategy
Navigate to Helpdesk > Routing. Choose your strategy: `round_robin` (simple), `skill_based` (match agent proficiency), `load_balanced` (lowest workload), or `skill_then_load` (hybrid). Configure category-level overrides (e.g., "Billing → billing_team, skill-based").
Define SLA Policies
Navigate to Helpdesk > SLA. Set business hours (9 AM–5 PM, timezone-aware), add holidays, and configure response/resolution targets by priority. Default: Critical (30min response, 1hr resolution), High (1hr/4hr), Medium (4hr/24hr), Low (8hr/48hr). Enable Predictive SLA for proactive breach prevention.
Configure AI Copilot
Navigate to Helpdesk > AI Settings. Enable: conversation summarization, next-action suggestions, draft responses (RAG), and similar ticket discovery. Configure Auto-Resolution: enable the feature, set confidence threshold (default 0.85), and add sentiment gating (default skip if <30). Enable Smart Reply Guard to suppress thank yous and OOO replies.
Build Your Knowledge Base
Navigate to Helpdesk > Knowledge Base. Create categories, write articles with rich text (TipTap), and publish. Run `generate_embeddings` to enable semantic search. Set up proactive suggestions (surfacing KB articles while customers type). Configure article versioning: every edit creates a version; restore any prior version.
typescript
// Example: List recent urgent tickets via the helpdesk MCP toolconst tickets = await aeion.helpdesk({ action: "list_tickets", filters: { priority: "urgent", status: ["new", "open"] }, limit: 25,});// tickets.docs[0].subject === "Login button broken on safari"
// Example: Get full ticket detailconst ticket = await aeion.helpdesk({action: "get_ticket",id: "tkt_abc123",});
// Example: Create a ticket from a webhook handlerconst created = await aeion.helpdesk({action: "create_ticket",data: {subject: "Payment retry failed",description: "Customer's 3rd auto-retry failed",priority: "high",category: "billing",},});
// Example: Assign a ticket to an agent (uses round_robin / skill_based / load_balanced// based on your Routing settings; or pass agentId to bypass)await aeion.helpdesk({action: "assign_ticket",ticketId: "tkt_abc123",agentId: "agent_456",});
// Example: Semantic search across the knowledge baseconst matches = await aeion.helpdesk({action: "search_kb",search: "reset password without email",});// matches.docs[0].title === "How to reset your password"
// Example: Read aggregate stats for the dashboard widgetconst stats = await aeion.helpdesk({ action: "get_stats" });// stats.newCount, stats.openCount, stats.urgentCount, ...Three safety layers: (1) Sentiment gating—tickets with negative sentiment (<30 score) are skipped. (2) Active incident detection—if an incident is active within 5 hours, auto-resolve is skipped to prevent OOO replies during outages. (3) Confidence threshold of 0.85 requires high-confidence KB matches; vague matches are passed to agents.
When a ticket is closed and the customer replies "Thank you!" or an Out-Of-Office is received, Smart Reply Guard detects the intent and suppresses the ticket re-open. The message is logged but the SLA clock doesn't restart. This prevents a flood of false re-opens from bulk thank-you emails after a product launch.
Yes. Each category can have custom routing rules that override the global strategy. For example, "Billing" category → `skill_based` with billing specialists. "Technical Issues" → `load_balanced` to prevent any single agent from being overwhelmed. Rules are checked in priority order.
Every time an article is edited, a new version snapshot is created automatically. Versions track the version number, what changed (content update, title change, etc.), a change summary, diff stats (characters added/removed), and the full content at that point in time. Restoring an older version creates a new version referencing the one it was restored from — nothing is overwritten.
Yes. Screen share sessions run over WebRTC. Remote control is off by default and only turns on after the agent explicitly requests it and the customer consents. Every control action (key press, mouse move, clipboard) is logged for compliance, and the customer can revoke access at any time.
It analyzes historical ticket data — resolution time by category, priority, and agent skill — to calculate a risk score for each open ticket. Tickets with high risk scores are flagged as "at risk" before the actual SLA breach, so a supervisor can reassign proactively.
Helpdesk shares the Aeion OS database natively. The ticket sidebar queries CRM (contacts, deals, activities) and Commerce (orders, subscriptions, invoices) directly — no API sync needed. Agents see the customer's CRM lifetime value, active subscriptions, and order status without leaving the ticket.
Yes. The Copilot runs on the platform's AI service, which is configured in platform settings — any AI provider the platform supports (OpenAI, Anthropic, etc.) can power it. If AI is unavailable, it falls back to basic rule-based summaries instead of failing.
Ready to empower your support team?
Configure channels, routing, SLAs, AI Copilot, and your knowledge base — most teams are live in under an hour.