A feedback miner that tells the account team before the customer leaves.

SourcePOST /feedback

Expanso nodeone pipeline, 5 steps, deterministic

1. Receive
2. Shape
3. Ask Jev
4. Gate
5. Route

Jev judgment

• topic
• sentiment
• churn_risk

• alert
• eng
• analytics

1. Expanso, deterministic

   Same fixed rules every time, no model involved.

2. Jev, judgment

   The one step a rule cannot write.

3. Receivelines 11 to 16

   An HTTP server input accepts feedback on POST /feedback.

4. Shapelines 17 to 26

   Parses the POST body if it arrived as a string, then stamps received_at. It applies the same fixed rules every time, with no model involved; the timestamp is the one value that differs.

5. record + questionsanswers + confidence
6. Ask Jevlines 27 to 84

   Expanso packs the whole record and the typed questions into one request. Jev answers. If the call fails, a catch substitutes empty answers marked jev-unavailable, and the pipeline keeps going.

   • topicchoice
   • sentimentscore
   • churn_risknoul
7. Gatelines 85 to 99

   Fixed thresholds over Jev’s answers choose alert, eng, or analytics.

8. Routelines 100 to 123

   A switch output writes to one file per decision.

name: jev-feedback-miner
type: pipeline
description: Feedback miner with Jev — every review and NPS verbatim gets topic, sentiment and churn-risk judged inline. Bugs route to eng, churn risks ping the account team, everything feeds analytics.
namespace: production
labels:
  category: integrations
  pattern: ai-decision
  model: jev

config:
  input:
    http_server:
      address: "0.0.0.0:8080"
      path: /feedback
      allowed_verbs: ["POST"]

  pipeline:
    processors:
      - mapping: |
          # http_server already parses JSON bodies; only parse raw strings
          root = if this.type() == "string" { this.parse_json() } else { this }

      - mapping: |
          root = this
          root.received_at = now()

      # ── Ask Jev ──
      - mutation: |
          meta jev_start = timestamp_unix_milli()

      - branch:
          request_map: |
            root = {
              "state": this.string(),
              "model": "jev-latest",
              "questions": {
                "topic": {
                  "type": "choice",
                  "instructions": "What is this feedback about?",
                  "criteria": {
                    "bug": "Something is broken or not working",
                    "feature_request": "Asking for new functionality",
                    "praise": "Compliments, positive experience",
                    "pricing": "Cost, plans, billing concerns",
                    "support_experience": "About the support interaction itself",
                    "other": "None of the above"
                  }
                },
                "sentiment": {
                  "type": "score",
                  "instructions": "What is the sentiment of this feedback?",
                  "criteria": [
                    "Very negative",
                    "Negative",
                    "Neutral",
                    "Positive",
                    "Very positive"
                  ]
                },
                "churn_risk": {
                  "type": "noul",
                  "instructions": "Does this feedback indicate the customer may churn or cancel?"
                }
              }
            }
          processors:
            - http:
                url: "${JEV_API_URL:https://api.typesafe.ai/v1/systemone}"
                verb: POST
                headers:
                  Content-Type: application/json
                  Authorization: "Bearer ${TYPESAFE_API_KEY}"
                timeout: 2s
                retries: 1
            - catch:
              - mapping: |
                  root = {"answers": {}, "model": "jev-unavailable"}
          result_map: |
            root.jev = {
              "answers": this.answers,
              "model": this.model.or("jev-latest"),
              "ms": timestamp_unix_milli() - metadata("jev_start")
            }

      # ── Confidence-gated cascade ──
      - mapping: |
          root = this
          let topic = this.jev.answers.topic.choice.or("other")
          let topic_conf = this.jev.answers.topic.confidence.or(0)
          let churn = this.jev.answers.churn_risk.noul.or(0)

          root.jev_decision = if $churn >= 0.75 {
            "alert"
          } else if $topic == "bug" && $topic_conf >= 0.7 {
            "eng"
          } else {
            "analytics"
          }

  output:
    broker:
      pattern: fan_out
      outputs:
        - stdout:
            codec: lines
        - switch:
            cases:
              # Production: swap for http_client -> account-team alerting webhook
              - check: this.jev_decision == "alert"
                output:
                  file:
                    path: ./data/jev-feedback-miner/alert-${! now().ts_format("2006-01-02") }.jsonl
                    codec: lines
              - check: this.jev_decision == "eng"
                output:
                  file:
                    path: ./data/jev-feedback-miner/eng-${! now().ts_format("2006-01-02") }.jsonl
                    codec: lines
              - check: "true"
                output:
                  file:
                    path: ./data/jev-feedback-miner/analytics-${! now().ts_format("2006-01-02") }.jsonl
                    codec: lines

123 lines. Copy and Download both give you the file byte for byte.