Real-time Lead Classification,
Routing & Validation

Lead|Chk is a hyper-efficient qualification layer that catches incoming customer communications at the edge. It acts as an elite gatekeeper—instantly isolating high-intent prospects from spam, tech support, and noise—delivering formatted leads to your sales reps or downstream systems in under 2 milliseconds.

Nobody outside your own infrastructure ever sees your lead data. Lead|Chk runs entirely inside your servers — no messages, no customer data, and no classifications are transmitted to any external service or third party. Unlike AI-based alternatives that route your customer conversations through cloud providers such as OpenAI or Google, Lead|Chk processes everything locally using a JSON rules file you control. Your leads stay inside your pipeline, and only your pipeline.

Most operational systems route all text to heavy AI models like OpenAI or Gemini, paying immense 'token costs' just to filter out spam, opt-outs, or simple one-line replies. Lead|Chk catches these inquiries first. It processes basic intents and rejections locally with zero variable cost, shielding your premium enterprise AI layers from expensive, unnecessary usage.

Completely. Unlike external cloud AI options that process your sensitive pipeline interactions on third-party frameworks, Lead|Chk is fully self-hostable. It deploys completely inside your secure private cloud or internal databases, ensuring enterprise security parameters are strictly protected.

Yes. Because of its lightweight, highly compiled framework, it is exceptionally resource-friendly. It easily operates side-by-side with heavy GPU systems or standard CPU web servers without draining processing power, allowing you to run elite automation without expensive hardware upgrades.

Yes — entirely. Lead|Chk is released under the MIT license, meaning it is free to use, modify, and deploy in personal or commercial projects with no restrictions. There are no per-request fees, no subscription plans, and no usage caps. Because you self-host it, your only cost is the infrastructure you already run. The source code, npm package, and Docker image are all publicly available at zero cost.

Hit the unauthenticated /health endpoint — no API key needed. A 200 OK means the API is up.

# Health check — no auth required
curl http://localhost:8080/health

# Expected response:
{
  "status": "ok",
  "timestamp": "2026-05-30T12:00:00.000000+00:00"
}

Pass your key in the X-API-Key request header. The key is set via the PROXY_API_KEY environment variable in your .env file. Missing or incorrect key returns 403 Forbidden.

# Valid request
curl -X POST http://localhost:8080/v1/classify \
  -H "Content-Type: application/json" \
  -H "X-API-Key: your-secret-key" \
  -d '{"message":"I want a demo","channel":"text","lang":"en"}'

# Wrong key → 403
curl -sS -o /dev/null -w "%{http_code}" \
  -H "X-API-Key: wrong-key" \
  http://localhost:8080/v1/classify
403

POST to /v1/classify with a JSON body. The intent field in the response tells you where to route the lead.

curl -sS -X POST http://localhost:8080/v1/classify \
  -H "Content-Type: application/json" \
  -H "X-API-Key: your-secret-key" \
  -d '{
    "message": "I would like to schedule a meeting for next week",
    "channel": "whatsapp",
    "lang": "en"
  }'

# Response:
{
  "status": "success",
  "data": {
    "intent": "schedule",
    "is_rejection": false,
    "scores": { "schedule": 10, "lead": 1, "support": 0 },
    "classification_metrics": {
      "matched_stems_count": 2,
      "processed_chars_length": 48
    }
  },
  "request_context": {
    "channel": "whatsapp",
    "language": "en",
    "latency_ms": 3
  }
}

Each channel applies different pre-processing. Email strips thread history; WhatsApp strips markdown formatting; SMS is capped at 160 characters. Test them explicitly:

# Email — thread history is auto-stripped
curl -sS -X POST http://localhost:8080/v1/classify \
  -H "Content-Type: application/json" \
  -H "X-API-Key: your-secret-key" \
  -d '{"message":"Hi, saw your site. Can we set up a call?","channel":"email","lang":"en"}'

# WhatsApp — markdown and emoji artifacts are scrubbed
curl -sS -X POST http://localhost:8080/v1/classify \
  -H "Content-Type: application/json" \
  -H "X-API-Key: your-secret-key" \
  -d '{"message":"*Hey!* Interested in a _demo_ 👍","channel":"whatsapp","lang":"en"}'

# SMS
curl -sS -X POST http://localhost:8080/v1/classify \
  -H "Content-Type: application/json" \
  -H "X-API-Key: your-secret-key" \
  -d '{"message":"Call me re: pricing","channel":"sms","lang":"en"}'

When is_rejection: true, the intent is forced to general regardless of other scores. Route these directly to your suppression list.

curl -sS -X POST http://localhost:8080/v1/classify \
  -H "Content-Type: application/json" \
  -H "X-API-Key: your-secret-key" \
  -d '{"message":"Please unsubscribe me from all emails","channel":"email","lang":"en"}'

# Response — note is_rejection: true and intent: general
{
  "status": "success",
  "data": {
    "intent": "general",
    "is_rejection": true,
    "scores": { "schedule": 0, "lead": 0, "support": 0 }
  }
}