Workflows & Automation15.1ms Reflex$0 / MIT License
11,400 active downloads
Customer Intent Classifier
15ms support request classification and ticket routing engine
Author: SupportOps AI
GitHub Repository →jevproxy // fast-customer-intent (reflex kernel)
$npx jevproxy run customer-intent
JevProxy Intercept Resolved(Mode: DIRECT_REFLEX)
15.1ms|Cost: $0.0001|0 reasoning tokens burned
STDOUT • Tool Call Output:Status: 200 OK
agent.dispatch(classify_customer_intent)
payload: { "message": { "type": "string", "description": "Customer message text" }, "user_tier": { "...
✔ Decision short-circuited in 15.1ms without roundtrip to frontier LLM.
Upstream token bill saved: $0.0135 on this turn.
Customer Intent Classifier
Compatible with Cursor, Claude Code, Windsurf, OpenCode
TRADITIONAL LLM CALL:UNOPTIMIZED
• Median Latency: 1250 ms
• Cost per Turn: $0.0135
• Mode: Full KV Cache Reload & TTFT Prefill
JEVPROXY REFLEX KERNEL:77x FASTER
• Median Latency: 15.1 ms
• Cost per Turn: $0.0001
• Accuracy: 98.4% deterministic
1-Click CLI Execution
Run this tool accelerated through the JevProxy gateway without manual wiring:
npx jevproxy run customer-intentOpenAI / Anthropic Tool Schema
JSON SpecificationPaste this schema into your agent tools definition or Cursor extensions:
{
"type": "function",
"function": {
"name": "classify_customer_intent",
"description": "Classify incoming support inquiry into deterministic business routes",
"parameters": {
"type": "object",
"properties": {
"message": {
"type": "string",
"description": "Customer message text"
},
"user_tier": {
"type": "string",
"enum": [
"FREE",
"PRO",
"ENTERPRISE"
]
}
},
"required": [
"message"
]
}
}
}Technical Architecture & Usage
Instant Ticket Routing
Route support tickets instantly to Zendesk, Stripe refunds, or engineering alerts without running a 70B parameter model on every incoming email.
Accelerate Customer Intent Classifier with JevProxy
Get 5,000,000 free decision tokens. Eliminate the 3-second tool freeze in Cursor and Claude Code in under 60 seconds.