Sidekick and a Support Agent: A Weekly Shopify Workflow

If you want the basic differences between the two tools first, read our Shopify Sidekick vs AI support agent comparison, listed at the end of this article. This guide is about the workflow that connects them.

Want your tickets answered while you fix the causes behind them? Start the 14-day trial.

The ticket-to-fix loop in 4 steps

Using both tools side by side doesn't help much on its own. They pay off when what one learns reaches the other.

• Tickets show the pattern. In your ticket center, you see that many customers ask whether a jacket runs small.
• Sidekick fixes the cause. You ask Sidekick to add a fit note to the product description. Shopify's Sidekick documentation (checked October 6, 2026) says it "presents changes for your review before applying them", so you read the edit before it goes live.
• The knowledge base gets the update. You add the same fit note to the support agent's knowledge base. Its answers now match the product page.
• Next week's tickets confirm it. You check whether sizing questions about that jacket go down. If they don't, the fix wasn't the right one, so you go around the loop again.

Here's a second example with discounts. You decide that customers whose parcel is late by a set number of days get a small apology code. You ask Sidekick to create the discount in the admin. Then you write the policy the support agent follows and decide whether it may send that code. Sidekick builds the tool once. The support agent uses it ticket by ticket.

Who owns each step: split by who starts the task

The rule is simple. When a customer starts the task, the support agent handles it. When the task changes how your store works, Sidekick handles it and you approve the change.

• Reply to "Where is my order?": support agent. The customer starts it and wants an answer fast.
• Approve a refund that fits your policy: support agent, within its permissions. Rule based, comes up all day.
• Cancel an unshipped order on request: support agent, within its permissions. Urgent, must happen before fulfillment.
• Send a goodwill discount code: support agent, within its permissions. Small, rule based, needs a quick answer.
• Handle an angry customer or a chargeback threat: your team. Needs judgment and the right tone.
• Create the discount the agent hands out: Sidekick. Store setup, done once.
• Fix a product description that causes sizing questions: Sidekick. Product edit in the admin.
• Rewrite your returns policy page: Sidekick, reviewed by you. Content in the admin.
• Find which products bring the most complaints: both. Tickets show the problem, Sidekick digs into orders and products.
• Pull a weekly sales or refunds report: Sidekick. Analysis for the merchant.

For a longer list of which ticket types suit automation, see 7 tickets to automate and 5 to keep, listed at the end of this article.

Ready to turn your tickets into a list of store fixes? Start the 14-day trial.

Guardrails on each side of the loop

The two tools carry different risks, so they need different checks.

Sidekick: changes you review before they apply. Since Sidekick shows you changes before applying them, your review is the main check. Two habits help:

• Read every proposed change before you approve it, especially edits to prices and discounts.
• Keep each policy written in one place, so your store pages and your support rules always say the same thing.

Support agent: rules you write in advance. Nobody reviews each customer reply before it goes out, so the rules have to exist before the first ticket. For each action, decide:

• Allowed or not: can the agent refund at all, or only reply?
• Your policy: write the rules you want followed, in your own words. For example, "refunds only within 30 days of delivery" or "cancel only if the order hasn't been sent to fulfillment."
• What your team keeps: list the cases a person should always handle, such as chargeback threats or requests that bend your policy. Don't give the agent permission for those.

In Relay, you choose what it can do, from replies to refunds: refunds based on your policies, discounts within the limits you set, exchanges for eligible items, address updates, cancellations and order edits. Anything complex or sensitive goes to your team. Our guide to AI for Shopify returns and refunds, listed at the end of this article, shows how to write policies for actions that involve money.

The weekly routine

Here's what a normal week looks like for a small team running the loop:

• Every day: the support agent answers incoming tickets, acts within its permissions and passes the rest to your team. You handle the tickets it passed on.
• Once a week, 1: open your support statistics and look at resolution rate and time saved.
• Once a week, 2: read the tickets your team handled that week. Write down the most common cause in one line, such as "unclear return window" or "no fit info on jackets."
• Same session: fix that top cause in the store with Sidekick. It could be a product page, a shipping note or a policy page.
• Before you close the session: update the support agent's knowledge base or policies so it reflects the change.
• The following week: check whether tickets about that cause went down before you pick the next one.

One cause per week is enough. A small fix you check beats five fixes you can't measure.

Before the support agent goes live, run through the Shopify AI support agent setup checklist, listed at the end of this article. It shows you how to take real past tickets from your inbox and check how the agent would handle them.

Signs the loop is broken

• You paste customer emails into Sidekick every morning. You're using an admin tool for a support job.
• The agent quotes a return window that doesn't match your policy page. Sidekick changed the page and the knowledge base never got the update.
• The same question keeps coming in for months. What your tickets show isn't reaching the person who uses Sidekick.
• The support agent can edit products or prices. Store changes need your review, so keep them on the Sidekick side.

FAQ

How often should I run the loop? Weekly works for most stores. During a seasonal peak like Black Friday, take a quick look every day at the tickets passed to your team. Then wait until things calm down to make bigger store changes.

Which tool should handle refunds? When a customer asks for a refund in a ticket, the support agent handles it, within the permissions you give it and the policy you write. Anything outside them goes to your team. Sidekick helps you look at refunds afterward, for example to see which products get refunded most.

How do I keep Sidekick and the support agent consistent? Keep one written version of each policy. Each time you change a policy or product page with Sidekick, update the support agent's knowledge base that same day. The weekly ticket review will show you any gap you missed.

What if a fix doesn't reduce tickets? Then the fix didn't address the real cause. Reread a few of the tickets word for word. Customers often ask about something next to what you changed, such as delivery times when you edited the shipping price note. Pick the next cause and run the loop again.

Let Sidekick fix the store and let Relay answer your customers. Start the 14-day trial.