Smart Auto-Response Based on Ticket Content
Respond to routine support requests instantly, even after hours.
Nora reads incoming Zendesk tickets, identifies common issues (password reset, billing, returns), and sends templated responses immediately while queuing for agent follow-up.
When a new ticket arrives in Zendesk, Nora reads the subject and message content. It classifies the issue type based on the text. If the request matches a common category like password resets, billing inquiries, or returns, Nora immediately adds a public reply with the relevant information and next steps. It then appends an internal note with the classification and assigns the ticket to the appropriate agent group for follow-up.
Support agents no longer spend their first hour clearing out routine overnight requests. They log in to a queue where basic inquiries already have an initial response. Agents review the internal notes, handle the complex cases that require human judgment, and adjust priority levels if a customer replies needing further assistance.
Runs real-time as tickets arrive using Zendesk, run by your Customer Support.
How it works
- Detect new incoming tickets — Nora polls the Zendesk API to identify newly created support tickets. It retrieves the ticket ID, requester details, subject line, and the full body text of the initial message sent by the customer.
- Classify the request content — Nora analyzes the subject and message text to determine the issue category. It checks the content against known frequent request types such as password resets, billing questions, or return authorizations.
- Flag ambiguous ticket requests — If the ticket contains multiple unrelated questions or the issue type is unclear, Nora skips the automatic public reply. It adds an internal note indicating the classification failed and leaves the ticket unassigned in the main queue.
- Send the public reply — For clearly identified common issues, Nora updates the ticket with a public reply. It inserts the correct response text for that category, addressing the customer's specific question while setting expectations for any necessary agent follow-up.
- Update tags and internal notes — Nora applies relevant tags to the ticket based on its classification. It adds an internal note summarizing the action taken and assigns the ticket to the designated support group for review.
What this replaces
Support agents start their shifts by reading through the backlog of tickets that accumulated overnight. They open each ticket, read the request, identify the issue, apply the appropriate macro or copy-paste a standard response, and update the ticket status. They then manually assign the ticket to the correct department or agent.
- Read incoming ticket subjects and body text.
- Identify the category of the customer request.
- Select and apply the matching response macro.
- Add tags and assign the ticket to a group.
Rough estimate rather than a measured figure: about 45 minutes a day, on the basis of 30 overnight tickets taking 90 seconds each to read, categorize, and reply.
What it needs to run
Accounts and plans
- A Zendesk account with API access enabled.
- An active Zendesk admin or agent user account to generate the API token.
Access it asks for
- Zendesk API Token to authenticate requests.
- Read access to tickets and users to view incoming requests and sender details.
- Write access to tickets to add public replies, internal notes, tags, and assignments.
What Nora reads
- New ticket IDs, subjects, and body content.
- Requester email and organization details.
What Nora writes
- Public replies on tickets.
- Internal notes on tickets.
- Ticket tags and assignee/group fields.
When it goes wrong
- Zendesk API rate limit exceeded during processing If the account hits the 100 requests per minute limit and receives a 429 error, Nora pauses execution. It waits for the API-specified retry period before attempting to process the remaining tickets.
- Ticket contains highly sensitive personal information If the ticket includes credit card numbers or highly sensitive personal information, Nora stops processing. It leaves the ticket untouched and alerts an admin in WhatsApp to handle the secure data manually.
- Customer replies immediately to the auto-response Nora does not process replies to existing tickets by default. The ticket remains in the assigned agent's queue, and the new customer message updates the SLA tracking normally.
- Zendesk API token expires or is revoked Nora fails to authenticate and immediately stops trying to read tickets. It sends a message to the designated chat channel requesting a new Base64-encoded API token.
Who this is for
Worth it if
- Support teams at 10-50 person e-commerce companies receiving 50+ routine inquiries a day.
- Organizations with significant after-hours ticket volume from global customers.
- Teams that heavily rely on Zendesk macros for their initial responses.
Not worth it if
- Teams handling highly technical, bespoke support requests where no two tickets are the same.
- Low-volume support operations receiving fewer than 10 tickets a day.
- Companies using a shared inbox rather than a structured ticketing system like Zendesk.
A worked example
A 30-person software company receives frequent export and invoice requests overnight. Nora is configured to identify these topics, reply with the relevant knowledge base link, and assign them to the Tier 1 support group.
Public reply added to Ticket #49281: Hi Sarah, Thanks for reaching out. You can export your monthly activity data by navigating to Settings > Data Export and clicking 'Generate CSV'. I have also assigned this ticket to our billing team to ensure your recent invoice is attached to your account. They will review this and follow up shortly. Best, Support Team Internal note added: Nora classification: Data Export / Billing. Sent standard export instructions. Assigned to Tier 1 Billing group.
Frequently asked questions
Can I change the text of the automatic responses?
You tell Nora in plain language what you want the response to say. If you need the billing response to include a new link to your payment portal, you send a message in your chat channel asking Nora to update the billing reply template. The change applies to the next matching ticket.
What happens if Nora misinterprets the customer's request?
If Nora sends a password reset link but the customer actually wanted to change their billing email, the customer will reply stating the confusion. This reply goes straight to the human agent's queue. Agents can see the initial auto-response in the ticket thread and correct the issue manually.
Does Nora close the ticket after replying?
No. Nora leaves the ticket open and assigns it to the relevant agent or group. The automation is designed to provide an immediate initial response, not to resolve the ticket entirely. A human agent always has the final review before the ticket status changes to solved.
Will Nora respond to every single ticket that comes in?
Nora only sends a public reply if the ticket content clearly matches one of your defined common categories. If a ticket is complex, vague, or covers a topic Nora does not recognize, it simply adds an internal note stating it could not classify the issue and leaves it for your team.
Connects to
Related automations
- Daily Standup Digest to Slack Channel — Compile team updates into a single daily message so you never miss critical blockers.
- Auto-Respond to FAQ in Slack — Answer common questions instantly in Slack threads so your team stays unblocked.
- Extract Action Items from Slack Threads to Task Board — Turn meeting discussions into tracked tasks automatically.
- Smart DM Follow-Up Reminders — Never lose track of messages awaiting response.
- Alert When SLA Targets are Missed — Catch SLA breaches before they happen—not after.
- Follow Up on Low CSAT Scores Automatically — Turn unhappy customers into advocates by addressing issues fast.
- Auto-Suggest Knowledge Base Articles to Customers — Answer with the guide that already exists, before an agent picks the ticket up.
- Batch Process Routine Ticket Actions — Handle 50+ routine tickets in 5 minutes instead of an hour.