You will get AI customer support automation that triages and drafts replies in Zendesk
Project details
Your support inbox gets the same twenty questions every week, and someone retypes the answer every time.
This puts an AI copilot inside the helpdesk you already use. Every incoming ticket is read, classified, tagged and routed automatically. For tickets it recognises, it drafts a reply in your tone using your real help centre content and your past resolved tickets, so the answer is grounded in what your company actually says instead of invented.
Your team stays in control. Nothing sends on its own unless you ask for it. An agent opens the ticket, finds a draft already written with the source articles linked, edits a line and sends.
What you get: ticket classification and routing, retrieval over your knowledge base and ticket history, drafted replies inside the helpdesk, an escalation path for anything sensitive, and a short handover doc so you can change the rules yourself later.
Works with Zendesk, Intercom, Freshdesk, Front, or a plain Gmail or Outlook shared inbox. Built in Python or Node and deployed to your own infrastructure.
This puts an AI copilot inside the helpdesk you already use. Every incoming ticket is read, classified, tagged and routed automatically. For tickets it recognises, it drafts a reply in your tone using your real help centre content and your past resolved tickets, so the answer is grounded in what your company actually says instead of invented.
Your team stays in control. Nothing sends on its own unless you ask for it. An agent opens the ticket, finds a draft already written with the source articles linked, edits a line and sends.
What you get: ticket classification and routing, retrieval over your knowledge base and ticket history, drafted replies inside the helpdesk, an escalation path for anything sensitive, and a short handover doc so you can change the rules yourself later.
Works with Zendesk, Intercom, Freshdesk, Front, or a plain Gmail or Outlook shared inbox. Built in Python or Node and deployed to your own infrastructure.
Chatbot Type
Customer Service ChatbotDevelopment Technology
ChatGPT, Node.js, PythonMessaging Platform
Email, Slack, WebsiteWhat's included
| Service Tiers |
Starter
$700
|
Standard
$1,800
|
Advanced
$3,500
|
|---|---|---|---|
| Delivery Time | 7 days | 14 days | 21 days |
Number of Revisions | 1 | 2 | 3 |
Number of Conversation Steps | 5 | 12 | 25 |
Number of Messaging Platforms | 1 | 2 | 3 |
Action Plan | |||
API Integration | - | ||
Bug Fixes | - | ||
Chatbot Flow Design | |||
Conversation Script |
Frequently asked questions
About Fabian
Software Engineer | Web Apps, Automation & Custom Systems
Nittendorf, Germany - 6:57 am local time
Most of my work is systems, not screens: web applications, internal tools, data pipelines, and integrations between things that were never designed to talk to each other. Some of it involves AI. Most of it is just careful engineering.
WHAT I BUILD
Web and app development, front to back
Automation and integrations across APIs, databases and third party tools
Data work: collection, cleaning, moving, reporting
AI where it earns its place: assistants, document search, voice, agents
HOW I WORK
I take a small number of projects at a time. You get direct answers, including when something is not worth building. I write the code myself and it still works after handoff, which is the part most people skip.
Self hosted where privacy matters. German and English, based in Germany, working worldwide.
If you already know what you want, send the spec. If you do not, describe the problem and I will come back with the shape of a solution before you spend anything.
Steps for completing your project
After purchasing the project, send requirements so Fabian can start the project.
Delivery time starts when Fabian receives requirements from you.
Fabian works on your project following the steps below.
Revisions may occur after the delivery date.
Access and ticket audit
I connect to your helpdesk in read mode, pull a sample of resolved tickets and your help centre, and map which ticket types are worth automating first.
Triage and retrieval build
I build the classifier that tags and routes tickets, plus the retrieval layer over your knowledge base, then test both against tickets your team has already answered.