Customer Question Helper
Sort incoming support tickets and get draft replies built only from your own help articles, with sources.
Your first result: draft replies with sources, saved as internal notes in your helpdesk.
$49
One-time · instant download
Checkout, receipt and download are handled by Gumroad. Opens in a new tab.

What you get
What's inside the kit
Automatic ticket sorting
Tags each ticket by category and urgency, and flags the ones that need a person.
Answers from your own help articles
Drafts can only use your approved help content, and show which article each part came from.
Built-in draft checks
Catches missing sources and unexpected amounts or links before a draft is saved.
Approval rules for refunds and credits
Refunds and credits need approval above your limits, with daily caps and no bulk actions.
Sample help center to start from
See the format, then swap in your own articles.
Setup guides for common helpdesks
How to connect your helpdesk, keep access minimal and handle privacy.
How to put it to work
How to get started
Follow these steps in order. After you buy, the same steps appear as a checklist in your kit viewer, with links to the exact guide for each step.
- 1
Add your help articles
20–40 minPut your approved help articles in the kit's help-articles folder as plain text files, then run one command so the helper can search them.
- 2
Load the automations into n8n
5 minLoad the ready-made 'failure alerts' and 'sort and draft' automations into your automation app.
- 3
Add credentials and settings
15 minConnect three accounts (the incoming-ticket link, your helpdesk and the AI service) and fill in the settings box.
- 4
Connect your helpdesk
15–30 minHave new tickets sent to the workflow, and map drafts to 'internal note / draft' in your helpdesk.
- 5
Set privacy settings
10 minTurn on automatic clean-up of old run history, mention your AI service in your privacy notice and decide how long to keep data.
- 6
Turn it on and review
ongoingActivate alerts, then triage. Review drafts for a few weeks and tune your articles.
Is it right for me?
Who it's for
A great fit if…
- Small support teams (1–5 people), including e-commerce
- Teams with a helpdesk or shared inbox and some written help content
Probably not for you if…
- You want replies sent automatically (there's no send step)
Questions
Frequently asked questions
Does it send replies to customers?
No. The workflow has no send step. Drafts are saved as internal notes for your team to review and send.
Can the drafts be wrong?
Yes. Review every draft and its sources before sending.
What data goes to the AI provider?
Redacted ticket text goes to the AI provider you choose. Mention this in your privacy notice.
Will this get me more customers, sales or time back?
We don't promise any result. The kit gives you tools and a safe starting setup; what happens depends on your business and how you use it.
Good to know before you buy
- Drafts can still be wrong. Review every AI draft and its citations before sending.
- No promise of accurate answers, faster resolution, saved time or any other outcome.
- Redacted ticket text is sent to the LLM provider you choose; disclose this in your privacy notice.
- Not legal advice.
Technical details (for the technically curious)
Technical words on this page, explained
- AI agent:
- An AI assistant that can take steps for you (like running a command or calling a tool), not just chat.
- LLM / AI model:
- The AI "brain" behind tools like ChatGPT or Claude. You usually rent it from a provider with an account key.
- API / API key:
- An API is how one program talks to another. An API key is a private password that lets your setup use a service.
- RAG:
- Looking things up in your own documents first, so the AI answers from your content instead of guessing.
- n8n:
- A free automation app (like Zapier) that runs step-by-step automations. You import our ready-made ones.
Formerly / technical name: Support Triage + RAG Draft-Reply Agent
Full description
Classifies each ticket (category, urgency, escalation), retrieves passages from your own help content, and drafts a reply that may use only those passages, with citations. Drafts are validated and saved as internal drafts; the workflow has no send step.
Full parts list
- n8n triage workflow + failure alerts
- Offline retrieval (no accounts needed) + optional embeddings index builder
- Citation and amount/link validation
- Sample help-center KB
- Draft-reply prompt
- Docs: helpdesk variants, least privilege, privacy
Requirements
- n8n
- Helpdesk or shared inbox
- OpenAI-compatible LLM API key
Tested on
- Node 20.19.2 / npm 11.21.0
- n8n 2.45.5 on Node 24.22.2 (local, no account)
- Linux kernel 6.12.94+
Tested locally with mocks; live services not yet tested. (2026-10-28)
What's in the download (42 files)
- 📁 .github/ (1 file)
- 📁 config/ (1 file)
- 📁 docs/ (6 files)
- 📁 kb/ (6 files)
- 📁 lib/ (2 files)
- 📁 n8n/ (2 files)
- 📁 prompts/ (1 file)
- 📁 scripts/ (6 files)
- 📁 tests/ (10 files)
- 📄 .gitignore
- 📄 CHANGELOG.md
- 📄 LICENSE
- 📄 package.json
- 📄 README.md
- 📄 REVIEW_STATUS.md
- 📄 SETUP.md
Version and changes
v0.2.0 · tested 2026-10-28
- 0.2.0 · 2026-10-30: Safe-actions approval gate: refund/credit thresholds, daily caps, no bulk actions, fail-closed.
- Unreleased · 2026-10-29: Compliance review: draft paid EULA, disclaimer, trademarks, privacy/security fixes and tests.
- 0.1.0 · 2026-10-28: Initial release: classification, offline retrieval, cited validated drafts (no send path), alerts, real n8n e2e harness.