Hi All -
Looking for some assistance here. I want to create a workflow that:
1, Triggers every 6 hours to get 200 tickets in Zendesk. (Done)
- Tells AI to review the context, word choice, tone, etc. of the 200 tickets and nark each one with a Zendesk Standard field priority. (bonus points if it can also learn as time goes on and use old tickets for priority consideration when reviewing future ones)
Where I need help is mainly around which AI node to select to get me started…hoping that for whichever one it is, I can explain to the AI what I’m wanting it to do via a prompt of my making.
- Maybe some finer details like setting the automation to only pick up tickets that are of Open status and do NOT have a priority marking already but this is really secondary to the AI node.
Please and thank you for any help with this matter!!
You can totally do this in n8n by looping over the 200 tickets and sending the relevant fields into an AI node that returns a normalized priority label you then map back into the Zendesk priority field. I work at eesel AI and for our Zendesk AI ChatGPT agents we follow a similar pattern where the AI classifies each ticket by urgency and then updates priority or tags accordingly, so your idea is very doable in practice. AI ChatGPT Agents by eesel App Integration with Zendesk Support
In n8n I’d start with something like the OpenAI or “AI Agent” node, define a strict schema in the prompt (eg “respond with only one of: low, normal, high, urgent”), and then use an IF or Switch node to convert that into the correct Zendesk priority value. You can later fine tune it by feeding resolved tickets plus their final priority back into the same flow as extra context so it “learns” from past decisions without needing actual model fine tuning.