User Feedback

• During the request: the trace is still active, so you enrich it directly.
• Later: you save the trace’s IDs, then attach feedback when the user rates the result.

​

Log feedback during the request

from honeyhive import HoneyHiveTracer

tracer = HoneyHiveTracer.init()  # Reads HH_INGESTION_API_KEY from your environment

# ... your application logic ...

tracer.enrich_session(feedback={
    "rating": True,  # boolean: thumbs up/down
    "comment": "The response was helpful",
})

from honeyhive import HoneyHiveTracer, trace

tracer = HoneyHiveTracer.init()  # Reads HH_INGESTION_API_KEY from your environment

@trace
def generate_response(query: str, user_id: str) -> str:
    response = ...  # your LLM call

    tracer.enrich_span(
        feedback={
            "rating": False,
            "comment": "The model hallucinated the capital",
            "ground_truth": "The capital of New York is Albany",  # the user's correction
        },
        user_properties={"user_id": user_id},
    )

    return response

​

Log feedback later

import uuid
from honeyhive import HoneyHiveTracer, trace

tracer = HoneyHiveTracer.init()  # Reads HH_INGESTION_API_KEY from your environment

@trace
def summarize(document: str) -> tuple[str, str]:
    event_id = str(uuid.uuid4())
    tracer.enrich_span(update_event_id=event_id)  # Must be a UUID

    summary = ...  # your LLM call
    return summary, event_id

def run_summary_job(job_id: str, document: str):
    # One session per job, so each result maps to its own trace
    session_id = tracer.create_session(session_name="summary-job")
    summary, event_id = summarize(document)

    # Store the IDs with the result your user will see
    save_result(job_id, summary, session_id=session_id, event_id=event_id)  # your persistence layer

from honeyhive import HoneyHiveTracer

tracer = HoneyHiveTracer.init()  # Reads HH_INGESTION_API_KEY from your environment

def submit_feedback(job_id: str, thumbs_up: bool, comment: str):
    result = load_result(job_id)

    # Feedback on the whole trace
    tracer.enrich_session(
        session_id=result.session_id,
        feedback={"rating": thumbs_up, "comment": comment},
    )

    # Feedback on the specific span
    tracer.enrich_span(
        event_id=result.event_id,
        feedback={"rating": thumbs_up, "comment": comment},
    )

from honeyhive import HoneyHive
from honeyhive.models import UpdateEventRequest

client = HoneyHive()  # Reads HH_INGESTION_API_KEY from your environment

def submit_feedback(job_id: str, thumbs_up: bool, comment: str):
    result = load_result(job_id)

    client.events.update(UpdateEventRequest(
        event_id=result.session_id,  # or result.event_id for a span
        feedback={"rating": thumbs_up, "comment": comment},
    ))

curl -X PUT "https://api.dp1.us.honeyhive.ai/v1/events/$EVENT_ID" \
  -H "Authorization: Bearer $HH_INGESTION_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"feedback": {"rating": true, "comment": "Great summary"}}'

• The trace must be ingested first. Updates for an unknown ID fail. The tracer logs the error instead of raising, and the API client raises an HTTPException.
• Updates merge. Each update sets the feedback keys you send and keeps the rest, so you can send a rating now and a comment later.

​

Concepts

​

What is User Feedback?

1. Explicit Feedback: User-provided ratings in your app - thumbs up/down, star ratings (1-5), written comments
2. Implicit Feedback: User actions that signal intent - clicking “Copy”, “Regenerate”, editing the response

​

Feedback Schema

{
  "rating": true,                    // boolean: thumbs up/down
  "score": 4,                        // number: star rating
  "comment": "Great response",       // string: user comment
  "ground_truth": "Expected answer", // string: for comparison (reserved field)
  "actions": {
    "copied": true,
    "regenerated": false
  }
}

​

Data Types

​

Nested Data

{
  "rating": 5,
  "step1": { "retry": true },
  "edits": [{ "value": "New York" }]
}

​

Reserved Fields

​

Learn More

​

SDK Reference

• Python SDK Reference - enrich_session(), enrich_span(), create_session()