Write down what you really did
Let the AI interview you, one question at a time, until you have five true stories from your work, with every number labelled exact, estimate or unknown.
You will have: A plain numbered list of five true work stories, each with what you did, the result, and every number labelled exact, estimate or unknown. Every later episode builds on it.
Before you start
- Your Target Card from Episode 1, so you know which job these stories should support
- Your old resume, a list of past jobs, or old emails and reviews to jog your memory. Keep them beside you; you do not need to paste them
- If you do paste anything, first remove your phone, email, home address, ID numbers, date of birth and referees' details. Put [PHONE] or [CITY] in their place
- Never paste your employer's confidential information, such as client names, internal figures or documents. Describe them in general words instead
Make it yours
Change the example (operations job search (fictional)) to your own. Your answers are saved in this browser and used in every episode.
Other series keywords (2)
Your prompt
Paste into: A normal AI chat (ChatGPT, Claude or Gemini). Use the same chat for the whole interview, then copy the final list into your notes.You are a patient interviewer helping me write down what I really did at work. I am aiming for Operations Coordinator jobs in United States. My background: 4 years as a retail assistant manager, moving into office operations.
Your goal: help me find up to 5 true stories from my work (or study, volunteering or home life if relevant).
How to work:
1. Ask ONE short question at a time and wait for my answer. Never ask several questions in one message.
2. Use my story numbers only (Story 1, 2, 3...). Never invent labels such as "3b". If you cannot tell whether something is a new story or part of the current one, ask me, and ask again if I do not answer.
3. For each story, find out: the situation, what I personally did, what changed because of it, any numbers, and who could confirm it (a job title only, no names or contact details).
4. Ask follow-up questions when my answer is vague. Ask me instead of guessing.
5. If I describe team work, keep it as team work. Do not give me sole credit.
6. If I mention a client name, internal figure or anything that sounds confidential, remind me to describe it in general words.
7. When a story is clear, say "Story [number] done". If it is not clear yet, say "Story [number] paused, missing: ..." and carry on.
Numbers:
- Label every number: (exact) only if I clearly say it is exact, (estimate) if I am roughly guessing, or [?] if I do not know. If I give a plain number without saying which, ask me once: "Exact or rough?"
- Never invent, round up or "improve" a number. If I say "I don't know", write [?].
- Never add employers, job titles, dates, skills, tools or qualifications I did not mention.
When I say I have no more stories, or we have 5, give me a plain numbered list in plain text (no tables, no bold), one block per story, listing only the stories we really have:
Story 1: a short title
- Where and when: only what I said, or [?]
- What I did:
- Result:
- Numbers: each with its label
- Who could confirm: only a person I said could confirm that story, otherwise [?]. Praise is not confirmation
- Skills it shows (only ones the story proves):
Leave confidential names and internal figures out of the list. After the list, add a short "Still unclear" section with facts I should check before using them.Review the prompt, then copy it. Never paste real passwords or API keys into a chat.
How to check it worked
- Read each story aloud. Could you explain it in an interview, with the details, if someone asked?
- Every number has a label: exact, estimate or [?]. Change any label that is wrong
- Team results are described as team work, not as yours alone
- No client names, internal figures or other confidential details are in the list
- Where you can, check numbers against something real, such as an old email, rota or report you still have access to
- You copied the final list into your notes. You will paste it again in Episodes 3 and 4
Sample result
Example: operations job search (fictional)Fictional example: Maya Torres's five stories after the interview. Your list will use your own answers.
If something goes wrong
- The AI fills in a number you never gave, such as "cut costs by 20%".
- Reply: "I did not give that number. Replace it with [?] and only use numbers I tell you."
- You cannot think of five stories.
- Ask: "Give me 5 example questions about everyday tasks in my past jobs that might show skills for my target job." Small, regular tasks count.
- The stories sound grand and not like you.
- Reply: "Rewrite each story in plain, simple words, closer to how I described it. Do not make it sound bigger."
Free. Good to know: in a Robert Half survey of more than 2,000 US hiring managers (published March 2026), about two in three said AI-enhanced resumes make skills harder to verify. True stories with honest numbers hold up when someone asks a follow-up question.
More in Prove it
This is part of the series From no replies to interviews: a 10-step AI job search. See all 10 episodes, the tools used and answers to common questions on the series page.