Comparing AI feedback with your original notes
When you read a cover letter an AI wrote for you, check whether it still describes what you actually did before you judge whether it reads well. When a short note gets stretched into full sentences, your role can grow, and numbers or tools that were never in your notes can appear.
Aim for writing that describes your work accurately rather than writing that slips past a detector. If the employer has published its own guidance on AI use, follow that as well.
1. Lay your records next to the draft
Put your experience notes, project records, and the work you produced next to the draft. Read the draft one sentence at a time and find the source for each claim. Missing paperwork doesn’t make an experience false, but it helps to separate what you remember from what you can verify.
| What to check | Question to ask | How to fix it |
|---|---|---|
| Role | Was this the team’s work or mine? | State your own scope |
| Action | Did I actually do this? | Separate taking part from leading |
| Result | What record or observation confirms it? | Keep only verified results |
| Numbers | Is there a unit, period, and baseline? | Delete numbers without a source |
| Skills and job experience | Have I actually used or done this? | Separate familiar terms from real experience |
2. Fix inflated roles and numbers
The example below is hypothetical and only illustrates the review.
- Original note:
As an intern, I sorted customer inquiries and organized recurring questions. - AI draft:
Led a customer-experience improvement project that raised satisfaction by 30%. - Revised:
During my internship, I sorted customer inquiries by type and organized recurring questions, and the team used them to update its help documents.
If you didn’t lead the project or measure satisfaction, remove those claims from the draft. Keep the last sentence only if the help documents were actually updated with your work. If you only organized the questions, write only as far as you can confirm, such as I handed them over as a draft for the next update.
Even without numbers, you can describe the scale of the work, what changed before and after, and the information you based decisions on. Use how to write achievements without numbers to make the result concrete without inventing an improvement rate.
3. Replace abstract phrases with actions
I took on challenges with passion or I used my communication skills doesn’t tell the reader what you did. Put the answers to these questions into your sentences.
- Who did you discuss which problem with?
- When people disagreed, what did you base the decision on?
- What document, experiment, code, or process did you create?
- What did you check after the decision?
For example, I resolved a conflict through communication can become I mapped the lab schedules of two teams whose experiments overlapped and worked out a new timetable with the people in charge. Use this kind of sentence only for things you actually did.
4. Answer the question and connect it to the role
If the question asks about your motivation and you spent the whole answer on a project’s process, the text doesn’t answer the question, however smooth it reads. For every question, mark these two things separately.
- The sentence that directly answers the question
- The experience and evidence that support that answer
Rather than copying phrases from the job posting, connect them to the points where you actually used the same skill. If you wrote problem solving, the text should show how you spotted the problem, compared causes, acted, and checked the outcome. If you repeated the same experience in every answer, check whether each question is looking for a different skill.
5. Check length and content in the submission form
- Do the company name, job title, and dates match your records?
- Did you avoid presenting team results as your own?
- Are there any skills or numbers left that you couldn’t explain yourself?
- Does the answer to the question appear at the start?
- Did you check, in the actual form, whether spaces count toward the character limit?
- Could you explain your actions and decisions in your own words in an interview?
When you need to cut length, remove background and repeated sentences first. Keep the parts that explain what you were responsible for and what came of it.
Frequently asked questions
What if the AI draft contains numbers that aren’t in my notes?
Take those numbers out of the submitted text first, then check the period, unit, and source. If the result wasn’t actually measured, don’t turn a team result or an estimate into a personal achievement. You can still describe the result through before-and-after change and scope without numbers.
The AI gives different feedback every time. Which sentences should I keep?
Choose sentences you can trace back to your notes and explain yourself, not the most fluent ones. However many times the draft is regenerated, the company name, dates, role, and results must stay the same. As in the example comparing documents, the level of detail can change, but check that the facts don’t.
A sample prompt for asking an AI to review
Compare my experience notes below with this cover letter draft. First, mark any roles, numbers, skills, or results that are not in my notes. In your suggested revisions, use only facts that are confirmed, and leave anything you need to know as a question. Also check whether each answer responds directly to the question and stays within the character limit.
Even when the AI flags wording that isn’t in your notes, treat its output as reference material. Compare what it flags with your originals and judge the facts yourself. Rewrite any sentence without enough support, or leave it out of what you submit until you can confirm it.