n8n-nodes-jev
n8n Jev Decision node that sends confident items onward automatically and routes uncertain ones to a person.
Overview
This n8n community node adds a Jev Decision step that answers a question about each workflow item and sorts it into two outputs: Confident, handled automatically, and Needs review, sent to a person. You set how sure Jev must be, with 0.8 suggested as a starting point.
The README walks through a fictional support team routing emails by team, urgency and refund request, and reports that in the author's test on 200 support emails Jev handled 59% on its own and got 97.5% of those right. Example workflows are included in a workflows folder. It is written for non-specialist operators who want automated triage with a human safety net.
What the decision model decides
| Type | Question | Answers |
|---|---|---|
| Choice | Which team should handle this email? | billing · technical · account access · sales |
| Yes / no | Is this email urgent? | yes · no |
| Yes / no | Does it ask for a refund? | yes · no |
How it works
- Install the node and add a TypeSafe API key credential.
- Add a Jev Decision step after the trigger.
- Write the question and possible answers, and set a confidence threshold.
- Connect Confident to automation and Needs review to a human queue.
Strengths
- Built-in human-in-the-loop split.
- Example workflows ready to import.
- Author-reported 97.5% accuracy on the auto-handled share in their test.
Caveats
- Results come from the author's own 200-email test.
- Requires a TypeSafe key.
- Licence not stated.
Getting started
Install the node via n8n Community nodes settings, add your TypeSafe API key, and import an example from the workflows/ folder.At a glance
- Type
- Project
- Status
- beta
- Language
- TypeScript
- Runtime
- n8n community node
- Has code
- yes
Use cases and integrations
How curated lists describe it
n8n node that routes items only when the AI is confident and sends the rest to…
wh000wh000/awesome-jev-live
Frequently asked questions
What happens to items Jev is unsure about?
They go to the Needs review output for a person to handle.
What threshold should I use?
The README suggests starting at 0.8.
How well did it work in testing?
The author reports 59% handled automatically with 97.5% of those correct on 200 emails.
Related projects
Cloudflare Flue agent example that adds a validated route_with_jev tool for typed support routing through AI Gateway.
ServiceprototypeTypeScriptAsync LangGraph example where a Jev Choice labels each email as invoice or general and routes it to a handler.
AppprototypePythonSmall Go agent-loop library with an optional goodall/typesafe package that uses Jev as a tool or hook beside chat models.
LibrarybetaGoRuby gem making semantic checks ordinary methods: feels?, decide and score on model fields, with Rails validations.
LibrarybetaRubyAlibaba's AgentScope 2.0 agent framework, which added a JevClassifierModel for binary, choice and score decisions via the TypeSafe SDK.
ProjectproductionPythonOpen-source AI email assistant; listed as using Jev as an optional classifier for email intent and triage.
ProjectproductionTypeScript