ref(feedback): update summary prompt and response parsing (#93847)

Author: vishnupsatish

src/sentry/feedback/usecases/feedback_summaries.py

@@ -11,23 +11,22 @@ def make_input_prompt(
     feedbacks,
 ):
     feedbacks_string = "\n".join(f"- {msg}" for msg in feedbacks)
-    return f"""Task:
-Instructions: You are an AI assistant that analyzes customer feedback.
-Create a summary based on the user feedbacks that is at most three sentences, and complete the sentence "Users say...". Be concise, but specific in the summary.
+    return f"""Instructions:
+
+You are an assistant that summarizes customer feedback. Given a list of customer feedback entries, generate a concise summary of 1-2 sentences that reflects the key themes. Begin the summary with "Users...", for example, "Users say...".
+
+Balance specificity and generalization based on the size of the input based *only* on the themes and topics present in the list of customer feedback entries. Prioritize brevity and clarity and trying to capture what users are saying, over trying to mention random specific topics. Please don't write overly long sentences, you can leave certain things out and the decision to mention specific topics or themes should be proportional to the number of times they appear in the user feedback entries.
 
 User Feedbacks:
 
 {feedbacks_string}
 
 Output Format:
 
-Summary: <1-3 sentence summary>
+<1-2 sentence summary>
 """
 
 
-SUMMARY_REGEX = re.compile(r"Summary:\s*(.*)", re.DOTALL)
-
-
 @metrics.wraps("feedback.summaries", sample_rate=1.0)
 def generate_summary(
     feedbacks: list[str],
@@ -50,10 +49,4 @@ def generate_summary(
 def parse_response(
     text,
 ):
-    summary_match = SUMMARY_REGEX.search(text)
-    if summary_match:
-        raw_summary_text = summary_match.group(1)
-        summary_text = re.sub(r"\s+", " ", raw_summary_text).strip()
-        return summary_text
-    else:
-        raise ValueError("Failed to parse AI feedback summary response")
+    return re.sub(r"\s+", " ", text).strip()

tests/sentry/feedback/usecases/test_feedback_summaries.py

@@ -1,25 +1,16 @@
-import pytest
-
 from sentry.feedback.usecases.feedback_summaries import parse_response
 
 
 def test_parse_response_valid():
-    response = "Summary: This is a test summary"
+    response = "This is a test summary"
     summary = parse_response(response)
     assert summary == "This is a test summary"
 
 
 def test_parse_response_valid_extra_whitespace():
-    response = """Summary:   This is a test  summary.
+    response = """   This is a test  summary.
 
     And this is a       continuation of the
     summary."""
     summary = parse_response(response)
     assert summary == "This is a test summary. And this is a continuation of the summary."
-
-
-def test_parse_response_response_invalid():
-    invalid_response = "This is not a valid summary"
-
-    with pytest.raises(ValueError, match="Failed to parse AI feedback summary response"):
-        parse_response(invalid_response)