Support tickets: a help desk with an AI assistant

Support Tickets gives your shoppers a real help desk instead of a contact form that dumps into a shared inbox. Customers open tickets from your storefront, you work them from a filtered dashboard queue, and every reply emails the customer automatically. The AI Support Desk Agent sits alongside you as a pre-triage assistant — it drafts, you decide, and it never talks to your customers on its own. ## How customers open tickets 1. Shoppers sign in on your storefront and go to **Support** at /<storeSlug>/support. 2. They choose a **category** and **priority**, enter a **subject** (up to 200 characters) and a **message** (up to 5,000), and submit. 3. The ticket and its opening message are created, and the customer can follow the whole thread from their account area under **Support**, including your replies. ## How to work the inbox 1. Open **Support** in the dashboard at /dashboard/support. 2. Filter the queue by **status** (new, open, pending, resolved, closed), **priority** (low, medium, high, urgent), or **category** (order, product, shipping, payment, account, other), or search subjects and messages directly. 3. Open a ticket to read the full thread — customer messages and staff replies in one place. 4. Write a **reply** and send. The message is appended to the thread and **emailed to the customer** automatically, so the conversation reaches them whether or not they are logged in. 5. Update the ticket's **status**, **priority**, and **category** as you learn more. Closing a ticket stamps it closed; reopen rather than starting a duplicate thread when the customer writes back. 6. Assign tickets to teammates so ownership is visible at a glance in the queue. ## The AI Support Desk Agent When the AI workforce is enabled, the **Support Desk Agent** triages every new ticket: 1. It reads the thread and proposes a **priority and category** — so a broken-package message stops sitting as low-priority "other". 2. It **drafts a helpful reply** based on the ticket, your catalog, and store context. Drafts land in the ticket marked as agent drafts, never sent to the customer directly — you review, edit, and send with one click. 3. It flags **refund-risk conversations** to the Operations Manager and hands repeated questions to the Docs & FAQ Manager as FAQ drafts, so the same question stops consuming your week. 4. Boundaries are hard-coded: the agent never closes tickets and never messages customers itself. The merchant always sends. ## What happens next Tickets resolved well become repeat customers, and tickets resolved repeatedly become FAQs — mine your closed tickets monthly and feed the patterns into your FAQ page or the docs area. Priority and category history also tells you where the operation actually hurts: a spike in shipping tickets is a carrier conversation, not a support problem. ## Tips 1. Work the queue oldest-first within each priority; new tickets feel urgent, but aging ones churn. 2. Use **pending** status when you owe the customer an update after checking with a supplier — it keeps the thread honest without closing it. 3. Let the agent's triage stand as the default and correct it when wrong; the proposals save the most time on obvious tickets. ## FAQ **Do customers need an account to open a ticket?** Yes — tickets are opened by signed-in customers so the thread is attributable and follow-ups reach the right person. That is also what keeps spam out of your queue. **What happens when I reply?** The reply is stored on the thread and emailed to the customer in your store's branding. Nothing double-sends: the email and the portal show the same message. **Can the AI agent close tickets or send replies by itself?** No. It proposes triage and drafts replies only; sending and closing are always human actions.