👉 n8n No Limits Access :

https://www.hostg.xyz/SHJ1L

n8n VPS Hosting | Secure and scalable automation

coupon -15% : DRFIRAS15

👉 n8n Course (in French)

n8n Course in French: Automation & AI Agents

coupon -20% : LANCEMENT : HERE

<aside> 🎯

What you will build: an n8n workflow that reads every new Gmail email and asks Jev (TypeSafe AI's decision model) which team it belongs to, whether it is urgent, whether it needs a reply and whether the customer is upset. The workflow files the email under the right label, prepares a reply draft (never sent automatically) and keeps a record of every decision.

</aside>

The flow at a glance

Gmail (new email)
   └─> Configuration (your settings)
        └─> Extract email fields (sender, subject, body)
             └─> Jev · Evaluate  : urgency (score) + needs a reply? (yes/no) + upset? (yes/no)
                  └─> Jev · Route : Sales | Support | Billing | Partnerships | Newsletters | Spam
                       │            + Fallback output if confidence < 80 % -> "To review" label
                       └─> Gmail label (Jev/...)
                            └─> Collect routing decision
                                 └─> If "Needs a reply?"
                                      ├─ YES -> Write reply draft (LLM) -> Save reply as Gmail draft -> Log
                                      └─ NO  -> Log Jev decision (Data Table)

<aside> 💡

The golden rule of this workflow: Jev decides, the LLM writes. Jev writes nothing: it picks from your options and returns a confidence level. The LLM (OpenAI) only steps in to write the draft, and only when Jev said a reply was needed.