jev-demo
Customer-service chatbot that asks every level of a routing tree in one Jev call and hands off on human requests or low confidence.
Overview
This jev demo is a customer-service chatbot routed by Jev. Instead of walking a decision tree one call at a time, it asks every level up front in a single call and discards answers it does not need, the speculative fan-out pattern from the docs, so a three-level route is one round trip.
The same call includes a Noul for whether the person wants a human and a Score for frustration. Either can trigger a handoff before the tree walk starts, and low confidence at any level also hands off. The README is short and gives only the run commands and this idea.
What the decision model decides
| Type | Question | Answers |
|---|---|---|
| Choice | Which branch applies at each level of the support tree? | — |
| Yes / no | Does this person want a human? | — |
| Score | How frustrated is the customer? | — |
How it works
- Receive a user turn
- Ask all tree levels plus handoff and frustration questions in one call
- Hand off if a human is wanted, frustration is high, or confidence is low
- Otherwise follow the route using the needed answers
Strengths
- One round trip per turn regardless of tree depth
- Built-in human handoff signals
Caveats
- Minimal documentation
- Demo only
Getting started
`npm install && npm run dev`.At a glance
- Type
- App
- Status
- prototype
- Language
- TypeScript
- Runtime
- Node web app
- Has code
- yes
Use cases and integrations
How curated lists describe it
A customer-service routing demo that batches Jev questions before following the resulting route.
logicrw/awesome-jev-projects
Frequently asked questions
What is speculative fan-out?
Asking every level of a routing tree in one call and using only the answers on the chosen path.
When does it hand off?
When the person wants a human, frustration is high, or confidence is low at any level.
How many API calls per turn?
One.
Related projects
Customer-service platform rebuilt from OpenAI's ChatKit sample where one Jev call decides intent, emotion and handling agent.
MCP serverbetaPythonPython example classifying customer support messages into 12 intents with Jev, flagging clarification below 0.8 confidence.
ProjectprototypePythonE-commerce support sandbox where Jev classifies intent, emotion and escalation and blocks risky agent replies before sending.
AppprototypeTypeScriptOpen-source AI email assistant; listed as using Jev as an optional classifier for email intent and triage.
ProjectproductionTypeScriptDeepSeek Harness plugin that classifies each user message's emotion, urgency and department with Jev and injects it as context.
PluginbetaTypeScriptHermes Agent plugin that classifies each turn's route (skill, research, history, others) with Jev before the model call.
PluginbetaPython