49/ 50
＋ Meets every requirement concisely and consistently
－ Does not separately detail the channel B expansion plan
Convert prose into a Korean bulleted report
Convert a prose report into the clipped, hierarchical bullet style Korean business writing expects — grouped by idea, not chopped sentence by sentence.
| Category | Office work › Reports |
|---|---|
| Tags | ReformattingRewritingOffice worker |
Convert this prose into a Korean-style bulleted report (개조식). Rules: 1. **Do not mechanically split sentences.** Group the content into a hierarchy first, then write. 2. Two levels of nesting maximum. If a third is needed, the grouping is wrong — regroup. 3. **Top-level items are conclusions, not labels.** Not "Revenue status" but "Revenue down 12% month on month". 4. **End every line with a nominal form (`~함`, `~임`), consistently.** This is the convention the format requires; a bulleted report in full sentences reads as unfinished. 5. One line per item. If it runs to two, split it or demote it. 6. Carry figures and proper nouns across verbatim. Then: anything in the source that resisted this format, and why.
After pasting, fill in the fields at the bottom (Source text)
What this prompt does
This format is not a bullet list — it requires conclusion-first headings and nominal endings. Splitting sentences mechanically produces something that reads as unfinished.
Model comparison
ChatGPT is strongest in accuracy, format, and length. Claude overexplains, while Gemini adds meaning and violates the required endings.
36/ 50
＋ Clearly groups causes, offsets, and plans hierarchically
－ Overstates format limits and adds excessive analysis
32/ 50
＋ Integrates the shortfall cause into the top conclusion
－ Invents execution expansion and breaks nominal endings
| Criterion | ChatGPT | Claude | Gemini | Leader |
|---|---|---|---|---|
| Instruction following | 10 | 7 | 6 | ChatGPT +43% |
| Accuracy | 10 | 7 | 6 | ChatGPT +43% |
| Specificity | 9 | 9 | 8 | Tie |
| Structure | 10 | 8 | 7 | ChatGPT +25% |
| Right length | 10 | 5 | 5 | ChatGPT +100% |
Scored 1–10 by gpt-5.6-sol with model names hidden (2026-09-24). This is an AI review, not a measurement.
Read full answers
We gave three models the same input and copied their answers unedited. Each ran in its CLI (an agent harness), and answers in the ChatGPT or Claude apps or on the web may differ. Outputs are in Korean.
Source text: 이번 분기 신규 가입자는 목표치에 미치지 못했습니다. 주된 원인은 광고 채널 A의 단가가 오르면서 집행을 줄였기 때문입니다. 다만 채널 B에서는 전환율이 개선되어 일부를 상쇄했습니다. 다음 분기에는 채널 B 비중을 늘리고 채널 A는 단가 추이를 보며 재개 여부를 정하려 합니다. 한편 이탈률은 소폭 증가했는데 원인은 아직 파악 중입니다.
- 이번 분기 신규 가입자는 목표치 미달임
- 주된 원인은 광고 채널 A 단가 상승에 따른 집행 축소임
- 채널 B 전환율 개선으로 일부 상쇄함
- 다음 분기 채널 B 비중 확대 및 채널 A 재개 여부 판단 예정임
- 채널 A 재개 여부는 단가 추이 확인 후 결정함
- 이탈률은 소폭 증가했으며 원인은 아직 파악 중임
형식에 완전히 맞추기 어려운 내용은 다음과 같음.
- “소폭 증가”: 구체적인 증가율이 없어 정량화하지 못함
- “일부 상쇄”: 상쇄 규모가 제시되지 않아 세부 수치화하지 못함
- “원인은 아직 파악 중”: 원인이 확정되지 않아 원인 항목으로 구체화하지 못함
More in this category
| AI Productivity Artifact Generator | |
| AI Workflow Automation Specialist | |
| Comprehensive Image Analysis Report | |
| Corporate Intel Report | |
| Developer Daily Report Generator |