Remove personal information, for HR and people teams

Find the personal details in a text, swap each one for a placeholder, change nothing else, and get a short list of what a person must still check before the text goes anywhere.

This is the version for HR and people teams. The steps and checks are the same as in the general template; each prompt carries one added line about your setting, and the examples are from your kind of work.

Sample results: runs and scores on this page are illustrations while testing is set up. How we test

Bring: The text to clean

Use This TemplatePrepare a prompt for your own AI chat.

Who it is for: HR business partners, recruiters and employee relations staff who clean up case notes, interview feedback and emails that name staff or candidates before they are filed or passed on.
Time: 10 minutes

Edit this version

Other Formats and Files

The Step 1 chat prompt under Use This Template is the quick start. Choose a full prompt chain, checklist, Claude skill or n8n file when that destination fits your work. Checked Instructions, below, run the whole job with a hard check after every step.

Choose one format

Prompt chain
Copy each step into ChatGPT or Claude, one after another.
Checklist
Follow the steps and check off the checks by hand.
Claude skill
A folder Claude loads, so it follows the steps by itself.
n8n file
Import into n8n and connect your own AI account.

This job also comes as Checked Instructions: the whole job in one set, with a hard check after every step. See the Checked Instructions on the general template.

Using a download:How to use this in n8n How to install a Claude skill

The same job, with and without the template

Sample

This comparison uses the Typical sample. Try an input below for a result guide matched to each sample.

One plain prompt

Remove the personal information from this text. Keep job titles, the department name (Distribution) and the town, Larkfield. (The same text is pasted below.)

Here is the text with the personal information removed:

Subject: Summary of grievance meeting, shift swap request

Hello [Name],

I am sending the summary of Tuesday's meeting. [Name], a packer in Distribution, asked to move to day shifts after her knee surgery in August. Her manager, [Name], said the day team was full. Rosalind (employee ID [ID]) said she had raised it twice before. The only forklift trainer on the night shift, who sat in as her representative, said the same thing had happened to another packer. Tobias agreed to review the roster by October 30. Rosalind can be reached at [Email] or [Phone], and her home address on file is [Address], Larkfield. I have asked Tobias to send his notes to me directly.

Regards, [Name] HR business partner

All personal information has been removed, so this is safe to share.

Problem:It leaves "Rosalind" and "Tobias" on their own, keeps "The only forklift trainer on the night shift", and ends by saying all personal information has been removed.

With this template

Result: 11 details replaced: 6 names, 1 ID number, 1 indirect detail, 1 email address, 1 phone number, 1 street address.

Cleaned text

Subject: Summary of grievance meeting, shift swap request

Hello [Person A],

I am sending the summary of Tuesday's meeting. [Person B], a packer in Distribution, asked to move to day shifts after her knee surgery in August. Her manager, [Person C], said the day team was full. [Person B] (employee ID [ID number A]) said she had raised it twice before. [Role A], who sat in as her representative, said the same thing had happened to another packer. [Person C] agreed to review the roster by October 30. [Person B] can be reached at [Email A] or [Phone A], and her home address on file is [Street address A], Larkfield. I have asked [Person C] to send his notes to me directly.

Regards, [Person D] HR business partner

Look at these yourself

"after her knee surgery in August": a health detail the table did not cover. Health information about an employee is usually kept confidential and shared only with people who need it, so ask HR or legal whether it should stay before this goes anywhere.
"[Person B], a packer in Distribution": a job, a team and a shift together may still point to one person in a small group.
'Larkfield': a town next to a home address can narrow down where a person lives.
"[Role A], who sat in as her representative" and "another packer": a representative and a second packer may be known to the readers.

Read the cleaned text yourself before you share it. An AI tool can miss a detail.

Check:Every detail is quoted word for word from the text: Eleven personal details; Distribution, Larkfield and job titles are left out because they may stay.
Check:Each kind is one of the seven allowed words: Eleven rows, each kind one of the seven allowed words: 6 names, 1 ID number, 1 indirect, 1 email, 1 phone and 1 street address.
Check:No number appears that the original text does not have: The only number left in the cleaned text (30) is in the original.
Check:The reminder to read it yourself is in the final text: The closing line is there: Read the cleaned text yourself before you share it.
Check:No listed detail is left in the cleaned text: None of the eleven details is left; the first names Rosalind and Tobias on their own and the email address are all replaced.
Check:Only the listed details were swapped: Only the eleven listed details were swapped; every other word, number and line break is as written, including the health detail.
Check:The counts add up to the table: The Result line says 11 details: 6 names, 1 ID number, 1 indirect detail, 1 email address, 1 phone number and 1 street address; the table has the same.
For your read:A person reads the cleaned text before it goes anywhere: Does the knee surgery line need an HR or legal read, and does a night-shift role in one named team still point to one person?

Use This Template

Add your material below. Logic Lab prepares Step 1; you copy it into your own AI and check the answer there. The site does not run the workflow.

Example materialSample

Choose a sample

Typical: Employee relations summary sent to a managerEasy: Survey results emailTricky: Reference check notes with look-alike words

One name and one email in a short notice; the may-stay box is left empty, so both are replaced and nothing else changes.

The text to clean (needed)What may stay (optional)

Step 1 Chat Prompt

Review the exact prompt, then paste it into your own AI. For later steps, use the chat chain in Other Formats and Files.

You are a careful privacy reader who spots the details that could identify a person.
The text is an HR or people document, such as case notes, interview feedback or an email about an employee, and it names staff, candidates or managers.

Your task is to list every personal detail in the text below.

The text:
Thank you for completing the new-hire survey, Callum Reyes. We have sent your results summary to callum.reyes@example.org. The next survey opens on Monday, November 2 and closes on Friday, November 13.

What may stay (if this is empty, nothing may stay):
Not provided

Rules:
- A personal detail is a name, an email address, a phone number, a street address, an ID or account number, a date of birth, or any wording that points to one person without naming them, such as "the only night-shift nurse at the clinic".
- A job title on its own may stay, but a title together with words like "the only" or a workplace can point to one person. Quote those words as one detail.
- Quote each detail from the text word for word, as short as you can. Do not correct it and do not reword it.
- List each different form once: a full name and a first name on its own are two rows. List the rows in the order each form first appears.
- Do not list anything the "may stay" line allows. Do not list a detail that is not in the text, and do not skip one because it looks harmless.
- Do not change the text yet.

Return one markdown table and nothing else:
| # | Detail (quote) |

Send to AICopy and open ChatGPTCopies this result and opens a blank chat. Paste and review it there.

Before Step 2, check: Every detail is quoted word for word from the text.

What This Sample Should Produce

Replace Callum Reyes and callum.reyes@example.org consistently. Keep the November 2 and November 13 dates because the input does not identify them as personal information.

A result guide for this invented input, not an AI reply.

Want to see where the materials, steps and checks fit? Watch the general template being built.

The steps, and the check on each

Step 1: List every personal detail
Pick out each detail that names or points to one person, quoted word for word, before anything is changed.
Hands back: A numbered table of the personal details, each quoted from the text.
Check:Every detail is quoted word for word from the text
A plain rule. If it fails, run the step again (the n8n file does this for you).
Step 2: Choose a placeholder for each detail
Give each detail a kind and a placeholder, with the same placeholder for the same person.
Hands back: A table that gives each detail a kind and a placeholder.
Check:Each kind is one of the seven allowed words
A plain rule. If it fails, run the step again (the n8n file does this for you).
Step 3: Write the cleaned text
Replace each detail with its placeholder everywhere it appears and change nothing else.
Hands back: The text with each personal detail replaced by its placeholder.
Check:No number appears that the original text does not have
A plain rule. If it fails, run the step again (the n8n file does this for you).
Step 4: Add the count and the reminder
Put the count, the cleaned text and a short list of what a person must still check into one package.
Hands back: A count line, the cleaned text, a list of what to check yourself, and the reminder line.
Check:The reminder to read it yourself is in the final text
A plain rule. If it fails, run the step again (the n8n file does this for you).

After the last step

Check:No listed detail is left in the cleaned text
A second look by the AI. If it fails, run the step again (the n8n file does this for you).
Check:Only the listed details were swapped
A second look by the AI. If it fails, run the step again (the n8n file does this for you).
Check:The counts add up to the table
A second look by the AI. If it fails, run the step again (the n8n file does this for you).
Check:A person reads the cleaned text before it goes anywhere
You check this one. If it fails, it is flagged for you.

What you need first

The text to clean(needed)
Paste the text as it stands. Paste only what your rules let you put into an AI tool; if they say no, take the names and numbers out by hand instead.
What may stay(optional)
For example job titles, or the name of a town. Leave empty if nothing may stay.

Ways to adapt it

This template does a careful first pass. It cannot promise that every personal detail is gone, and the last step says so. A person must read the cleaned text before it goes anywhere.
Paste only what your rules allow you to put into an AI tool. If they do not allow it, take the names and numbers out by hand and do not use this template on that text.
To let a kind of detail stay, such as dates or job titles, say so in the What may stay box. Details you name there are left out of the list.
To add a kind of detail, such as a license plate number, add it to the allowed kinds in the placeholder step and in its check, both.
For a long text, run one section at a time and keep the same letters for the same person in every section.

Change this version in the Editor. Its added context stays in every prompt and the general template stays separate.

Notes for HR and people teams

Employee IDs, payroll numbers, badge numbers and candidate numbers are ID numbers. Replace them unless the may-stay line allows them.
A job title with a team, a shift or words like "the only" can point to one person. Quote those words as one detail.
Health, leave, family and discipline details do not fit the seven kinds. Leave them as written, list them under the things to look at, and leave them for an HR or legal read.
Use one letter for one person across every form of the name, and a different letter for each other person.

The scoring guide

Sample

Each template check has a criterion marked met or not met; a person must still make the handed checks. (The guides in Quality rate a whole piece of work on three levels.) Under each one is how the example run scored.

Download the scoring guide (Markdown)

Every detail is quoted word for word from the text
A plain rule
Meets:Every detail in the table is found word for word in the text.
Does not meet:A detail has been reworded or corrected, or is not in the text word for word.
Example built from the before and after:Met
Each kind is one of the seven allowed words
A plain rule
Meets:Every row has exactly one of name, email, phone, street address, ID number, birth date or indirect.
Does not meet:A row has another word, such as phone number or address.
Example built from the before and after:Met
No number appears that the original text does not have
A plain rule
Meets:Every number in the cleaned text is also in the original text.
Does not meet:The cleaned text has a number the original does not, such as a placeholder written with a digit.
Example built from the before and after:Met
The reminder to read it yourself is in the final text
A plain rule
Meets:The final text includes the line Read the cleaned text yourself before you share it.
Does not meet:The line is missing, or it has been replaced by a claim that the text is safe to share.
Example built from the before and after:Met
No listed detail is left in the cleaned text
A second look by the AI
Meets:No detail in the table is left in the cleaned text, whole or in part, including a first name on its own.
Does not meet:A listed detail, or part of one, is still in the cleaned text.
Example built from the before and after:Met
Only the listed details were swapped
A second look by the AI
Meets:Apart from the placeholders, every word, number and punctuation mark matches the original.
Does not meet:A word was dropped, reworded or swapped by mistake, such as a verb that only looks like a name.
Example built from the before and after:Met
The counts add up to the table
A second look by the AI
Meets:The Result line matches the table, in total and for each kind.
Does not meet:A count on the Result line is different from the number of rows in the table.
Example built from the before and after:Met
A person reads the cleaned text before it goes anywhere
You check this one
Meets:You read the cleaned text from top to bottom, decided about each item under Look at these yourself, and looked for what the list missed.
Does not meet:The cleaned text is shared without being read through by a person, or the items to check are skipped.
Example built from the before and after:For your read

Example built from the before and after: met 7 of 7, plus 1 for your read

Learn from this template

The AI swaps the details and a rule checks that nothing else changed. What it cannot judge is whether something still points to one person, such as a job title or a rare event. The last step ends with a reminder, and the final check is yours: you read the cleaned text first.

Where to see it: step 4 (Add the count and the reminder).

Read the guide: Where a person must look Take the lesson: Where a person must look (lesson 8)

More to read:Worked example: remove personal info Adapting a template

Practice:Check it yourself Kinds of checks

Which one fits?

Use Review a draft before you send it when the text is a draft to be read for quality and nothing in it needs cleaning.
Use Hand off a task to someone else when you are passing on a task with its materials, not cleaning one text.

Related templates

All for HR and people

Review

Fact-check against sources15 min
Every claim gets a verdict and a quote from your sources

Review

Check every requirement is covered15 min
Each requirement marked covered or not, with the passage quoted

Review

Score work against a rubric15 min
Each mark quotes the work and follows the rubric's own words

Last run: October 17, 2026: Example built from the before and after (sample run)
Tools and versions: Example built from the before and after
What was checked: The files, against each tool’s import format, and 1 sample run, each check’s finding recorded.
Version: 1.0
What changed: First release.

Report a problem with this template How we test

Remove personal information, for HR and people teams

How it runs

The same job, with and without the template

One plain prompt

With this template

Cleaned text

Look at these yourself

Runs and scores

Example built from the before and after

Use This Template

Step 1 Chat Prompt

The steps, and the check on each

Step 1: List every personal detail

Step 2: Choose a placeholder for each detail

Step 3: Write the cleaned text

Step 4: Add the count and the reminder

After the last step

What you need first

Ways to adapt it

Notes for HR and people teams

The scoring guide

Learn from this template

Which one fits?

Related templates

Review

Review

Review