ChatGPT prompts for candidate submittals: 8 prompts that stick to the facts
On this page
Good ChatGPT prompts for candidate submittals turn your own screening notes into the client's format without adding anything the candidate did not say. They give the model the job's must-haves, your notes and your template, and a rule to mark gaps as [NOT COVERED] instead of filling them. Below are eight copy-ready prompts, from first draft to a final fact check. Only paste candidate details the candidate has agreed to share and your organization allows in that tool: as of October 2026, OpenAI may use Free, Plus and Pro conversations for training unless you opt out, but not Business or Enterprise content by default.
The template itself, with a filled example, is in the candidate submittal template, and the timing from screen to send is in the screen-to-submittal workflow. This page is only about using ChatGPT to get from notes to a draft safely. General recruiting prompts are in ChatGPT prompts for recruiters.
What can you paste into ChatGPT for a submittal?
A submittal is built from a person's career, pay and reasons for moving, so the data question comes first. OpenAI's help article on how your data is used to improve model performance says, as of October 2026, that content from Free, Plus and Pro accounts may be used to train its models unless you turn off Improve the model for everyone in Data controls; that rating a response with a thumbs up or down can make that whole conversation available for training even after you opt out; and that ChatGPT Business, Enterprise and Edu content is not used to improve its models by default.
Whatever the plan, these rules hold:
- Use an approved account. If your organization has not approved a ChatGPT account for candidate data, do not paste candidate data into it.
- Have the candidate's agreement. The candidate should have agreed to be represented to this client for this role; see the right to represent template.
- Minimize. Replace the name with initials, and leave out contact details, the current employer if the candidate asked for discretion, and references' details.
- Never paste health, family, religion, immigration detail beyond what the role requires, or anything the candidate told you in confidence. It should not be in the submittal anyway; see what not to write in interview notes.
Which prompts turn notes into a submittal draft?
1. Notes to a full submittal in your template
You are an agency recruiter writing a candidate submittal for a client.
Use ONLY my screening notes below. Do not add skills, employers, numbers,
dates or opinions that are not in the notes. Keep anything in quotation
marks word for word. Where the template asks for something my notes do
not cover, write [NOT COVERED].
Template:
[PASTE YOUR TEMPLATE HEADINGS, e.g. Headline / Why this candidate (3
reasons mapped to must-haves) / What to probe / Logistics / Other
processes]
Client must-haves:
[LIST 3-5]
My screening notes (candidate referred to as "the candidate"):
[PASTE NOTES]- Paste in: your template headings, the must-haves from the job order, and your notes with the name removed.
- Good output: a draft shorter than your notes, with quotes intact and a few honest [NOT COVERED] markers.
- Check facts and data: every claim against your notes before anything else; prompt 8 does this systematically. Confirm the account is approved for candidate data (data rules).
2. Map evidence to each must-have
From my screening notes below, build a table with columns:
must-have | evidence from the notes (quote where possible) | strength
(strong / partial / none) | what the client should ask to confirm it.
Use only the notes. If there is no evidence, write "none" and do not
infer it from job titles.
Must-haves: [LIST]
Notes: [PASTE NOTES, NAME REMOVED]- Paste in: the same must-haves and notes.
- Good output: a table where "partial" and "none" appear when they should. A table of all-strong ratings usually means the model filled gaps.
- Check facts and data: the "evidence" column must be traceable to your notes; delete anything inferred from a job title. Same data rules as prompt 1.
3. Write the three-line blurb
Write a three-sentence candidate blurb for the top of a submittal email.
Sentence 1: who the candidate is in one phrase plus one result that proves
it. Sentence 2: how they meet the client's top must-have: [MUST-HAVE].
Sentence 3: why they want to move and their availability.
Use only the facts below. No adjectives such as "rockstar", "passionate"
or "seasoned".
Facts: [PASTE 5-8 FACTS FROM YOUR NOTES]- Paste in: a short list of facts you chose, not the whole notes.
- Good output: three sentences a hiring manager can read in ten seconds. The formula is explained in how to write a candidate blurb.
- Check facts and data: the result in sentence one is the candidate's, in their words or your notes. Pasting a fact list rather than full notes is also the lowest-data option.
Which prompts handle concerns and tailoring?
4. State concerns honestly
Based on my notes below and the client's must-haves, list up to three
honest concerns a hiring manager might have about this candidate (for
example a gap against a must-have, a short tenure, or a pay difference).
For each: the concern in one sentence, any context from my notes, and a
question the client could ask in interview to test it.
Do not soften or invent context. Do not mention age, health, family or
any protected characteristic.
Must-haves: [LIST]
Notes: [PASTE NOTES, NAME REMOVED]- Paste in: must-haves and notes.
- Good output: the "What to probe" section of the submittal, which builds trust with a client because you raised the issue first.
- Check facts and data: any context offered must come from the candidate, and the candidate should know you will raise it. Remove anything sensitive before pasting.
5. Tailor the draft to this hiring manager
Here is a submittal draft and my notes from the intake call with the
hiring manager. Reorder and tighten the draft so it leads with what this
manager said matters most. Do not add any new claims about the candidate;
only reorder, cut and reword. Under 250 words.
Intake notes (manager's priorities): [PASTE, NO OTHER CANDIDATES]
Draft: [PASTE DRAFT]- Paste in: your intake notes and the checked draft from prompt 1.
- Good output: the same facts, reordered for one reader. How to present candidates to clients covers when to call instead of email.
- Check facts and data: compare old and new versions line by line for claims that changed strength, such as "supported" becoming "led." Keep other candidates out of the prompt.
6. Make an anonymized marketing version
Turn this submittal into an anonymized profile I can send to companies
that have not seen the candidate. Remove the name, current and past
employer names, exact job titles that would identify them, university
names, and any unique project details. Describe employers by type and
size instead (for example "a regional logistics company with about 400
staff"). Keep skills, results and availability. Under 150 words.
Submittal: [PASTE CHECKED SUBMITTAL, NAME ALREADY REMOVED]- Paste in: the checked submittal with the name already removed.
- Good output: a profile nobody at the candidate's current company could recognize. Consent and sending rules are in the MPC marketing email template.
- Check facts and data: read it as the candidate's current boss would; if any detail points to them, remove it. Market a candidate only with their agreement.
7. Brief the client for the interview
From this submittal and my notes, write a one-page interview brief for
the hiring manager: 3 areas to explore with one suggested question each,
1 concern to test, and logistics for the interview. Use only the
information given. Questions must be about the job, not personal life.
Submittal: [PASTE]
Notes: [PASTE, NAME REMOVED]- Paste in: the checked submittal and notes.
- Good output: questions that follow from the evidence, which helps the manager run a better interview.
- Check facts and data: remove any question touching protected characteristics, and send only what the candidate agreed to share.
How do you fact-check a submittal with ChatGPT?
8. Compare the draft with your notes
Compare the submittal draft with my screening notes. List every factual
claim in the draft (skills, employers, numbers, dates, results, quotes,
pay, notice period, reasons for moving) in a table with columns:
claim | supporting line from my notes, copied exactly | status
(supported / not supported / changed in strength).
Do not rewrite the draft. Only check it.
Notes: [PASTE NOTES, NAME REMOVED]
Draft: [PASTE DRAFT]- Paste in: the notes and the draft, in the same approved account.
- Good output: a table with a status for every claim. Delete or fix every "not supported" and "changed in strength" row.
- Check facts and data: the checker is the same kind of model as the writer, so skim the "supported" rows too. The final check is yours.
How do you keep an AI-drafted submittal factual?
- Capture quotes during the call. A submittal with the candidate's own words on two or three key points is harder to distort and more persuasive.
- One candidate per conversation. Start a new chat for each candidate so details cannot leak from one to another.
- Watch the verbs. "Supported," "contributed to" and "was part of" turning into "led," "owned" and "drove" is the most common distortion.
- Read it before it goes. A submittal carries your name and your agency's credibility with the client.
Tools built for this job work the same way. Interview Signal, which we make, drafts the submittal in the agency's template from the call transcript and removes any quote it cannot find in that transcript; even so, the recruiter reads it before it goes to a client. Whatever you use, the test is the same: could you show the client where each claim came from? More ways to lift submittal results are in how to improve your submittal-to-interview ratio.
Questions people ask
Can ChatGPT write a candidate submittal?
It can turn your screening notes into a well-organized draft in your template, which saves time. It cannot know anything you did not give it, so it will fill gaps with plausible guesses unless you tell it not to. Use a prompt that bans new facts and marks gaps, then check every claim against your notes before the submittal goes to a client.
Is it safe to paste interview notes into ChatGPT?
Only in an account your organization has approved for candidate data, and only details the candidate agreed to share for this purpose. As of October 2026, OpenAI says content from Free, Plus and Pro accounts may be used to train its models unless you opt out, while ChatGPT Business and Enterprise content is not used by default. Replace the name with initials and leave out contact details and anything sensitive.
How do I stop ChatGPT from exaggerating a candidate's experience?
Tell it to use only your notes, to keep the candidate's quotes word for word, and to write [NOT COVERED] where a must-have has no evidence. Then run a separate fact-check prompt that lists every claim in the draft next to the line in your notes that supports it, and delete anything it cannot match.
What should a candidate submittal include?
A headline, three reasons the candidate fits mapped to the client's must-haves with evidence, what the client should probe in interview, logistics such as availability, notice and pay expectations, and any competing processes. The candidate must have agreed to be submitted to that client for that role.