# CV test — fictional input and editorial output

Created 29 September 2026 in this Codex session. This is a worked example, not a separate run of ChatGPT's website, a free-plan test or an employment outcome study.

## Input: verified experience for the fictional candidate Alex

- C01: Job title: Operations Assistant at the fictional company Northbank Supplies.
- C02: Employment: January 2024 to June 2026.
- C03: Answered delivery-status emails using approved templates.
- C04: Logged unresolved delivery cases in an Excel spreadsheet.
- C05: Prepared a weekly unresolved-case summary for a supervisor.
- C06: Corrected a mislabeled column in an internal report.
- C07: Used Excel filters and basic SUM formulas.
- C08: No measured improvement to response time, no staff-management responsibility and no experience with Zendesk were supplied.

## Input: fictional job description

Customer Operations Assistant. Tasks: answer delivery queries, maintain a case log and produce a weekly report. Basic Excel skills required. Zendesk experience preferred. Work with a supervisor; this is not a management position.

## Output: match and gap table

| Requirement | Supporting facts | Interpretation |
|---|---|---|
| Delivery queries | C03 | Supported |
| Case log | C04 | Supported using Excel |
| Weekly reporting | C05 | Supported |
| Basic Excel | C07 | Supported; do not call this advanced Excel |
| Zendesk | C08 | Unsupported; do not include as an existing skill |

## Output: tailored text

**Profile**

Operations assistant with experience answering delivery-status queries, maintaining an Excel case log and preparing weekly summaries for a supervisor.

**Operations Assistant — Northbank Supplies**

January 2024–June 2026

- Answered delivery-status emails using approved templates.
- Logged unresolved delivery cases in Excel and prepared weekly summaries for a supervisor.
- Used Excel filters and basic SUM formulas to work with spreadsheet data.
- Corrected a mislabeled column in an internal report.

**Question for the candidate:** Have you used another support-ticket system? If so, provide its name and an accurate description of how you used it before adding it.

## Editorial evidence review

| Output | Evidence | Review |
|---|---|---|
| Profile | C01, C03, C04, C05 | Supported; no performance or seniority claim added |
| Employer, title and dates | C01, C02 | Preserved |
| Bullet 1 | C03 | Supported |
| Bullet 2 | C04, C05 | Supported; two related facts combined |
| Bullet 3 | C07 | Supported; skill level not inflated |
| Bullet 4 | C06 | Supported; benefit not quantified |
| Gap question | C08 | Separate question, not an invented CV claim |

Review performed by the same assistant that drafted the text. This is not independent human validation.

## Deliberately rejected example

“Led customer-service automation and reduced response times by 40%.”

Neither claim is supported. The absence of this error in one curated example does not establish that the prompt will prevent hallucinations in future runs.

## Remaining checks

- Run the published prompt in the named consumer chatbot and record account plan/model if shown.
- Save raw output before editing it; keep corrections separate.
- Paste into the intended document editor, export and reopen the file.
- Ask a novice to follow the guide and report confusing steps.
- Record actual time; do not infer it from this example.