F-2390Detected Oct 5, 02:33 UTC
How do I transfer my eSIM to my new iPhone?
Found in Nightly · v3.4 · 2026-10-05 on support-golden-v4 · eSIM & activation
Model answer vs gold answer
30% of model-answer words appear in the gold answer
Model answer
support-assistant v3.4Use eSIM Quick Transfer during iPhone setup while both phones are nearby. Unsupported: Your old eSIM keeps working for 7 days so you can switch back if needed.
Gold answer
Human-verified · support-golden-v4Use Quick Transfer on the new phone while both phones are nearby and unlocked. If it is not available, request a new QR code in My Harbor. The old eSIM stops working as soon as the new one activates.
- claimFlagged by the judge as unsupported or contradicted
Retrieved context
bge-m3 · hybrid (BM25 + dense, RRF) · top 3 passed to the generator
2 of 3 retrieved chunks are relevant · gold passage ranked first
- 1
Moving your eSIM to a new phone
Quick transfer · kb-esim-009 · v2026.09
RelevantUse Quick Transfer during setup with both phones nearby.
Similarity0.90 - 2
Moving your eSIM to a new phone
Old device deactivation · kb-esim-009 · v2026.09
RelevantThe old eSIM is deactivated as soon as the new one activates.
Similarity0.86 - 3
Activating an eSIM
Activation times · kb-esim-004 · v2026.09
Not relevantActivation usually completes within 10 minutes.
Similarity0.75
Judge verdict
LLM judge · Claude Opus 5.5 · rubric v3
0.33
rubric score · pass threshold 0.70
The 7-day overlap contradicts chunk 2 (immediate deactivation).
Diagnosis
Unsupported claims · eSIM reactivation
Root cause
Generator prior from other carriers' eSIM policies.
Suggested fix
Add contrastive gold example to the few-shot set for eSIM transfer.
Playbook: unsupported claims
The answer states something that is not present in the retrieved context.
Detection signals
- Claim-level faithfulness below 0.85
- Numbers, fees or durations absent from every retrieved chunk
- Citations that do not contain the cited sentence
Common fixes
- Require a citation after each factual sentence and refuse when none applies
- Reject answers with numeric claims missing from context in a post-generation check
- Lower temperature for billing and policy intents
- Add the query to the regression set with an escalate-or-decline gold answer
Activity
1 event · 0 comments
Relevant ·
Detected by support-assistant v3.4 run and labeled unsupported claims