Support Ticket Response Templates (and How to Draft Replies with AI)
Ten ready-to-use support ticket response templates for the situations every help desk handles, plus a Spojit workflow that picks the right template for each new ticket in Front, fills in the details and leaves a draft reply for your agent to check and send.
What This Integration Does
Good support templates give every customer a clear, consistent, friendly answer, and save agents from writing the same reply forty times a day. The ten templates below cover the most common ticket types. Copy them into your help desk's saved replies as they are, or use the workflow in the second half of this article: when a new conversation arrives in Front, an AI step reads it, chooses the matching template, fills in the order number, dates and names from the message (and from your store if you connect it), and saves the result as a draft reply on the conversation. The agent opens the conversation, reads the draft, edits if needed and sends. Nothing goes to a customer without a person approving it.
The workflow runs every few minutes over open, unassigned conversations, and tags each one it drafts for, so a conversation never gets two drafts.
The Templates
Replace the parts in square brackets. Keep the tone, adjust the wording to your brand.
1. Acknowledgement (first response)
Hi [first name],
Thanks for getting in touch. I've got your message about [topic] and I'm looking into it now.
You'll hear back from me by [time or day]. If anything changes in the meantime, just reply here.
[agent name]2. Where is my order
Hi [first name],
Your order [order number] shipped on [ship date] with [carrier]. You can track it here: [tracking link].
The latest update shows [status], and it's expected to arrive by [estimated date].
If it hasn't arrived by then, reply to this email and I'll chase it with the carrier for you.
[agent name]3. Shipping delay
Hi [first name],
I'm sorry, your order [order number] is running late. [Short reason, for example: the carrier has a backlog in your area.]
The new expected delivery date is [date]. [Optional: I've refunded your shipping cost as an apology.]
I'll keep an eye on it and let you know if anything changes.
[agent name]4. Refund approved
Hi [first name],
I've processed a refund of [amount] for order [order number]. It goes back to your original payment method,
and most banks show it within 3 to 5 business days.
You don't need to do anything else. Thanks for your patience.
[agent name]5. Return instructions
Hi [first name],
No problem, here's how to return [item]:
1. Pack the item in its original packaging if you still have it.
2. Attach the return label: [label link].
3. Drop it at any [carrier] location by [date].
Once it reaches us, we'll process your [refund or exchange] within [number] business days.
[agent name]6. Bug or technical problem
Hi [first name],
Thanks for reporting this, and sorry for the trouble. To help our team reproduce it, could you send:
- the steps you took just before it happened
- the device and browser (or app version) you're using
- a screenshot or the exact error message, if you have one
I've logged it as [ticket reference] and I'll update you as soon as we know more.
[agent name]7. Escalation to a specialist
Hi [first name],
I want to make sure you get the right answer, so I've passed your request to [team], who handle [topic].
[Specialist name] will reply by [time or day]. You don't need to repeat anything; they have the full history.
[agent name]8. Unhappy customer
Hi [first name],
I'm sorry. You expected [what they expected] and got [what happened], and that's not good enough.
Here's what I've done: [action taken]. Here's what happens next: [next step and when].
If this doesn't put things right, reply and I'll make it my priority.
[agent name]9. Account access or password
Hi [first name],
You can reset your password here: [reset link]. The link works for [time period].
If the email doesn't arrive within a few minutes, check your spam folder or reply and I'll help.
For your security, we'll never ask for your password by email.
[agent name]10. Closing the ticket
Hi [first name],
I'm glad that's sorted. I'll close this conversation now, but if anything else comes up,
just reply to this email and it will come straight back to us.
Thanks for being a [brand] customer.
[agent name]Prerequisites
- A Front connection in Spojit with an API token that can read conversations, create drafts and manage tags.
- A Front tag such as
ai-drafted, and its tag ID (from thelist-tagstool). - The teammate ID drafts should be authored as (from
list-teammates), usually a shared support teammate. - The templates above saved somewhere the workflow can read: paste them into the prompt, or upload them to a Spojit knowledge collection.
- Optional: a Shopify or WooCommerce connection so drafts can include real order status.
Step 1: Schedule Trigger
Add a Trigger node with type Schedule and cron */5 * * * *.
Step 2: Find New Conversations
Add a Front step with list-conversations and a search query q such as is:open is:unassigned. Name the output convos. Add a Loop in ForEach mode over the results, with a Condition inside that skips conversations whose tags already include ai-drafted.
Step 3: Read the Message
Inside the loop, add a Front step with get-conversation on {{ item.id }} to get the subject and the latest message, and the customer's name and email. If you connected a store, add a lookup here (for example Shopify list-orders with query email:{{ customer_email }}) so the AI has the real order number and tracking.
Step 4: Pick the Template and Fill It In
Add a Connector node in Agent Mode on the Text Tools connector. An empty tool selection does not mean "no tools" (it gives the agent every tool on the connector), so use Text Tools, whose tools only reshape text and cannot change any system, so this step only writes the draft. If your templates live in a knowledge collection, add a Knowledge node before it that retrieves them and pass its output into the prompt. Prompt:
You draft replies for our support team. Choose the ONE template below that fits the
customer's message and fill in the brackets using only facts from the message and the
order data. If a fact is missing, keep the bracket so the agent fills it in.
Never promise refunds, credits or dates that are not in the data.
Templates:
[paste the ten templates]
Customer message: {{ message.text }}
Customer name: {{ customer_name }}
Order data: {{ orders }}Add a Response Schema with template (the template number), confidence (high, medium or low) and reply (the filled-in text). Name the output draft.
Step 5: Save a Draft Reply and Tag the Conversation
Add a Front step with raw-api-request to create a draft reply on the conversation: POST to /conversations/{{ item.id }}/drafts with a body containing author_id (your teammate ID) and body ({{ draft.reply }}, with line breaks converted to <br>). Then a Front add-tag step with the ai-drafted tag ID. Agents see the draft in the reply box when they open the conversation.
Tips
- Low confidence, no draft: skip drafting when
confidenceislowand tag the conversationneeds-humaninstead. A missing draft is better than a wrong one. - Knowledge base answers: for product questions, add a Knowledge node that searches your help articles and include the passages in the prompt, so drafts answer with your real policy.
- Measure it: compare drafts with what agents actually sent for a week. Templates that agents always rewrite need rewording.
- Same templates everywhere: keep one master copy of the templates and paste it into Front's message templates as well, so agents and the AI use identical wording.
Common Pitfalls
- Sending instead of drafting: posting to the conversation's messages endpoint sends immediately. Use the drafts endpoint so a person always reviews.
- Invented facts: without the "only facts from the data" rule, a model may guess a tracking number or delivery date. Keep the rule and the order lookup.
- Double drafts: without the tag check, every run drafts again for the same unassigned conversation.
- Tone drift: keep your brand voice in the prompt ("warm, plain English, no exclamation marks") so every draft sounds like your team.
Testing
Send three test emails to your Front inbox: a "where is my order", a refund request and a bug report. Run the workflow manually and open each conversation in Front: each should have a draft based on the right template and the ai-drafted tag. Adjust the prompt until all three are right, then turn on the schedule.