You paste your store policies once, drop in the day's customer messages, and AI drafts a reply for each one. It tags the message, grounds the answer in your policy, and flags the action to take. You keep the final read and the send.
Run a store long enough and you notice something: the inbox repeats. Where is my order. Can I exchange this for a medium. Do you ship to Canada. The answers barely change.
But you still pay for each one. You re-read the message, dig up your own return window, type the same friendly paragraph, and send. Forty times. By lunch you've answered tickets and built nothing.
The fix isn't a clunky chatbot that annoys customers. It's a drafter that does the boring 80 percent — read, match policy, write the reply — and leaves the judgment to you. This article shows how.
Why does the same support question eat so much time?
Because every answer is cheap on its own and expensive in bulk. One reply takes ninety seconds. You don't have one. You have forty, every single day.
And each one has hidden cost. You re-read the message to be sure you understood it. You recall your exact return window so you don't promise the wrong thing. You soften the tone so it sounds like your brand, not a robot. Then you do it again for the next near-identical message.
None of that is hard. All of it is repetitive. That's the worst kind of work — too easy to feel important, too constant to ignore. It's also exactly the kind of work a prompt is good at.
The goal isn't to remove you. It's to stop making you the typist. Let the tool produce the obvious draft, and spend your attention on the three tickets that actually need a human.
What does a good support reply actually need?
It needs the right answer, grounded in your real policy, in your real voice. Strip a support reply down and it's four things, in order:
- The customer's actual question. Not what you assume — what they asked. A “where's my order” and a “this arrived broken” need different openings.
- The matching policy. Your return window, your shipping times, your exchange rule. The answer has to come from your store, not a generic script that promises something you don't offer.
- Your tone. Warm and casual, or crisp and premium. The same facts land differently depending on voice, and the voice should be yours.
- The next step. A reply that ends with nothing leaves the customer (and you) stuck. “Here's your tracking,” “send a photo and we'll replace it,” “I've started your refund.”
Three of those four are pattern work — read the message, pull the right policy, write it in a set tone. AI does that all day without getting tired or curt. The one part that stays yours is the call on the edge cases. That's what this tool is built around.
How does the tool draft a whole inbox at once?
It takes your policies plus a batch of messages and runs each message through the same steps in one pass. You set it up once, then feed it the day's inbox:
- Paste your policies. Shipping times, return/refund/exchange rules, sizing notes, and your brand tone. This is the knowledge every reply is grounded in.
- Drop in the messages. Copy today's customer emails or DMs, one per block. Whatever landed in the inbox.
- Tag and draft. It labels each message by type (order status, return, sizing, damaged, cancel, general), then writes a reply that uses your policy and your tone.
- Flag the action. Each card says what to actually do — send tracking, approve a return, request a photo, escalate — and scores how confident the draft is so you know which to double-check.
You go from a wall of unread messages to a deck of finished, on-brand drafts. You skim each one, fix the rare wording, and send. The inbox that used to eat your morning takes a coffee's worth of time.
Isn't this just a chatbot? Why draft instead of auto-reply?
No — and the difference is the whole point. A chatbot answers the customer directly, with no one checking. That's where stores get burned: it invents a policy, promises a refund you didn't approve, or talks like a machine in front of your best customer.
A drafter keeps you in the seat. It writes the reply; you approve it. The customer still hears from a human who actually runs the store. You get the speed of automation without handing a stranger your brand voice and your refund button.
It also fails safely. When a message is ambiguous or high-stakes — an angry customer, a weird edge case, a chargeback threat — the tool flags it low-confidence instead of confidently making something up. You see the flag and handle that one yourself.
So you get the best of both. The repetitive 80 percent is drafted and ready in seconds. The sensitive 20 percent is handed back to you, clearly marked, before anything goes out.
How do you keep the replies sounding like your brand?
You define the voice once, up front, and the tool holds it for every message. Most “AI support” sounds the same because no one told it who the store is. You fix that in two lines.
Give it a tone and a couple of rules: “warm and a little playful, never corporate, always thank them for reaching out, sign off as ‘the team at Wildgrain.’” Now every draft inherits that, whether it's a refund or a sizing question.
Add your non-negotiables too. Never promise a delivery date you can't hit. Never argue about a return inside the window. Always offer an exchange before a refund. These become guardrails the drafts respect, so the tool can't write something off-brand or off-policy.
The result reads like you on a good day — patient, consistent, on message — even on the fortieth ticket when a tired human would start getting short.
What does the real output look like?
Here's the actual output from the sample run. Two real messages, two different types, both grounded in the store's policy:
Type: Order status (WISMO) · Action: send tracking, reassure Hi Jenna — thanks for checking in! Orders usually ship in 1-2 business days and take 3-5 to arrive, so #1043 is right on schedule. Here's your tracking: [LINK]. It should land by Thursday — shout if it doesn't and I'll chase it down personally. — the team at Wildgrain
Why it works: it uses the store's real shipping window, gives the next step (tracking), and sets a clear expectation.
Type: Return / exchange · Action: approve exchange, send steps Hi Marcus — totally fine, happy to swap it! Our exchanges are free within 30 days, and you're well inside that. Reply with the size you'd like instead and I'll get the new one moving today — just pop the medium back in the mailer with the prepaid label inside. — the team at Wildgrain
Why it works: it quotes the exact 30-day exchange policy, offers the swap before a refund, and ends with one clear ask.
Two messages, two correct answers, zero retyping. Run it on the forty that landed overnight and the math changes your whole morning.
Where do you still beat the tool?
You win on judgment, relationships, and policy changes — and the tool is built to protect all three. It drafts; you decide.
First, the hard cases. A furious customer, a genuine mistake on your end, a loyal buyer who deserves a freebie. The tool flags these as low-confidence so you catch them and write the human reply yourself.
Second, the exceptions. Sometimes you bend the rule — extend a return, eat the shipping, comp an order to keep a fan. That's a business call, not a script. The tool gives you a clean starting draft; you decide when to go off-policy.
Third, the policy itself. What your return window is, when you refund versus replace, how generous you want to be — that's strategy, and it's yours. The tool just applies the rules you set, fast and consistently.
How do you turn this into a service you can sell?
Once the tool clears your own inbox, you can run it for other stores for money. Every Shopify owner with more than a few orders a day has the exact problem you just solved.
The offer is simple: you take their support inbox off their plate. You paste their policies into the tool once, then each day you draft the replies and either send them or hand them back for one-click approval. You're selling time back to a busy founder.
Stores pay for this. A part-time support VA runs $300 to $1,500 a month, and you can do the same volume in a fraction of the hours because the tool does the drafting. Start with one store you already buy from, prove a week, then ask for a referral.
It costs just $9 to start — one prompt that pays for itself on the first store. You run it in the browser with your own API key. Land two or three stores and it's a real side income built on a skill you now have.
How do you run it yourself?
You paste one prompt into Claude Code and it builds the tool for you. It opens as a dark dashboard, pre-filled with the sample store and messages above, so it works on the very first run.
It also has a Settings panel for your own API key, so you can run it on your real inbox every morning — again and again, on whatever lands that day.
Grab it below — drop your email and the prompt is on the very next page. Paste it in, swap in your policies and your messages, and let it draft the inbox.
Can you turn this into a side hustle?
Yes — think of it as a skill you just acquired in one paste. Skills can be sold, and this one sells by the deliverable.
The play is simple. Shopify and DTC store owners drowning in tickets already want done-for-you customer-support reply drafting — they just dread producing it. You produce it in minutes, deliver a clean result, and bill for the outcome. Going rates run $300 to $1,500 a month per store.
The best part is the cost to start: $9 to start — one prompt that pays for itself on the first job. The tool does the heavy lifting in minutes, so your margin is high and you can take on more clients without more hours. To get your first client, reach out to a few Shopify and DTC store owners drowning in tickets you already know. Do one for free, show them the result, and ask who else needs it.
FAQ
Will the replies actually sound like my store?
Yes — you set the tone and your rules up front (how warm, how you sign off, what you never promise). Every draft inherits that voice, and you still do the final read before anything sends, so it ends up sounding like you.
Does it send messages automatically?
No, and that's on purpose. It drafts replies for you to approve. You stay in control of what actually goes to the customer, so it never invents a policy or promises a refund you didn't okay.
Do I need to be technical to use it?
No. You paste one prompt into Claude Code and it builds the whole tool, pre-loaded with a working example. Then you paste in your own policies and messages.
Can I reuse it on tomorrow's inbox?
Yes — that's the whole point. Enter your API key once and re-run it every day on the new batch of messages. It's a reusable app, not a one-time output.