jev-intent-classification
Python example classifying customer support messages into 12 intents with Jev, flagging clarification below 0.8 confidence.
Overview
Intent Routing with TypeSafe's JEV Model is a small Python project that classifies a customer support message into one of twelve predefined intents, such as order_status, order_cancel, delivery_tracking, delivery_delayed, delivery_address_change, payment_failed, payment_refund, product_information, product_availability, return_request, return_status and unknown. It uses the typesafe-sdk in a single step.
The result includes the intent, a confidence score, a requires_clarification flag set when confidence is below 0.8, and the probability for every intent so downstream systems can apply fallback logic. Routing decisions remain configurable outside the classifier. (The README expands JEV as 'Judgmental Evaluation of Values', which is the author's own gloss.)
What the decision model decides
| Type | Question | Answers |
|---|---|---|
| Choice | What is the customer's primary intent? | order_status · order_cancel · delivery_tracking · delivery_delayed · delivery_address_change · payment_failed · payment_refund · product_information · product_availability · return_request · return_status · unknown |
How it works
- Take a customer message
- Ask Jev a Choice over twelve intents
- Set requires_clarification if confidence is below 0.8
- Route downstream using the probabilities
Strengths
- Simple, readable example
- Full probability distribution for fallback
- Clarification flag on low confidence
Caveats
- No accuracy evaluation
- Fixed intent set
Getting started
`uv run python main.py`.At a glance
- Type
- Project
- Status
- prototype
- Language
- Python
- Runtime
- Script (typesafe-sdk)
- Has code
- yes
Use cases and integrations
How curated lists describe it
JEV intent classification using Python
logicrw/awesome-jev-projects
Frequently asked questions
What does Jev decide?
The customer's primary intent among twelve options.
What happens if Jev is unsure?
Below 0.8 confidence the result asks for clarification.
How do I run it?
`uv run python main.py` with a TypeSafe key.
Related projects
Customer-service platform rebuilt from OpenAI's ChatKit sample where one Jev call decides intent, emotion and handling agent.
MCP serverbetaPythonE-commerce support sandbox where Jev classifies intent, emotion and escalation and blocks risky agent replies before sending.
AppprototypeTypeScriptCustomer-service chatbot that asks every level of a routing tree in one Jev call and hands off on human requests or low confidence.
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