prompts.chatprompts.chat
Login
All classifiers
ChoiceFeedback & research

Feedback topic

Tag a piece of product feedback with its main topic: usability, performance, pricing, missing feature, bug or praise.

F@f

Problem

Feedback arrives from surveys, app reviews, sales calls and support tickets. Without consistent tags, nobody can say which themes are growing.

Goal

Give every piece of feedback one main topic, so themes can be counted and bugs reach engineering.

Input / state

The feedback text and where it came from.

feedback
The feedback text.
source
Survey, app review, sales note, support ticket and so on.

Decision question

ChoiceQuestion key feedback_topic

What is the main topic of this feedback?

Options

  • usabilityHard to find, understand or use something.
  • performanceSlow, laggy, crashing under load, or using too many resources.
  • pricingPrice, plans, limits or value for money.
  • missing_featureWants something the product doesn't do.
  • bug_reportDescribes something not working as it should.
  • praisePositive feedback without a request or complaint.

Allowed outcomes

Decided in your code from the answer and its confidence. Thresholds are yours to set and test.

  • UsabilityChoice is usability and confidence meets your threshold.
  • PerformanceChoice is performance and confidence meets your threshold.
  • PricingChoice is pricing and confidence meets your threshold.
  • Missing featureChoice is missing_feature and confidence meets your threshold.
  • Bug reportChoice is bug_report and confidence meets your threshold.
  • PraiseChoice is praise and confidence meets your threshold.
  • Researcher tags itConfidence is below your threshold: probability is spread across options.

Example scenarios

Predefined demonstrations. No model is called; the outcome shown is an illustrative judgment.

Example state

{
  "source": "survey",
  "feedback": "Took me ten minutes to find where export lives."
}

Outcome

Illustrative
Usability

Choice is usability and confidence meets your threshold.

Why

Hard to find a feature.

Using it well

  • Long feedback often covers several topics. Split it into sentences or points in code first, then tag each one.
  • Keep the topic list short and stable over time; changing it breaks trend lines.

Request definition

POST https://api.typesafe.ai/v1/systemone
{
  "model": "jev-latest",
  "state": {
    "feedback": "<feedback>",
    "source": "<source>"
  },
  "questions": {
    "feedback_topic": {
      "type": "choice",
      "instructions": "What is the main topic of this feedback?",
      "criteria": {
        "usability": "Hard to find, understand or use something.",
        "performance": "Slow, laggy, crashing under load, or using too many resources.",
        "pricing": "Price, plans, limits or value for money.",
        "missing_feature": "Wants something the product doesn't do.",
        "bug_report": "Describes something not working as it should.",
        "praise": "Positive feedback without a request or complaint."
      }
    }
  }
}

Request shape from the System One API reference; a Kev server accepts the same body. Replace the placeholder state with your data. Requests need your own API key or server.

Version History

v1Current Versionabout 8 hours ago@f

Initial version

Comments (0)