ChatGPT prompts for nail techs come in two completely different kinds, and the search results mix them up constantly. One kind generates nail art — images, colour combinations, design ideas. The other writes your business: the rebooking text, the consultation reply, the no-show policy, the answer to a review that stung. This post is the second kind. The short answer is that the prompts worth using aren't the ones you copy — they're the ones you load with your own real captions, your own real policy, your own real client. And a few things should never go in the box at all.
The short version:
- A prompt works when it carries your context. Bracketed templates delete exactly that.
- Strip your client's name, phone number, address and health details before you paste.
- Never let AI decide what a product does to a nail, and never let it write a review.
- Everything else is a judgement call, and the judgement is yours.
Why a Gel Company Is Writing This
Zillabeau sells gel. We're not an AI company and we have no special authority on how these tools work — so everything here about how prompting works comes from OpenAI's and Anthropic's own documentation, attributed to them by name. What we can add is the other half: what a consultation reply has to cover, what a no-show policy has to survive, what a caption sounds like when a tech wrote it and not a marketing department.
And here's the honest part. We went back through six months of customer messages — 6,179 tickets with a real customer-written message in them, between March 6 and September 6, 2026 — and counted how many nail techs asked us anything about ChatGPT. Zero. So this isn't a post answering a flood of questions. It's a post about a gap.
The gap is real and it is measurable. The U.S. Census Bureau found U.S. business AI use "hovered between 17% and 20%" between December 2025 and May 2026 — but less than 20% of firms with four or fewer employees reported using AI, against 37% of firms with 250 or more, and in that same window use rose among firms with at least 20 employees while it didn't change significantly among firms with fewer than 20. The gap is widening away from exactly the person reading this: the Bureau of Labor Statistics counts 201,800 manicurists and pedicurists working in the U.S. in 2025, 23% of them self-employed, on a median of $17.19 an hour. That reader isn't buying an enterprise marketing suite. Free tier and better prompts is the realistic intervention.
What Should You Actually Type Into ChatGPT?
Type the things only you know. A prompt is not a magic phrase — it's a briefing. OpenAI's own guidance is to "ensure your prompts are clear, specific, and provide enough context for the model to understand what you are asking," and to "use descriptive adjectives to indicate the tone" you want (OpenAI Help Center). Your real captions, your real policy, your real client — that's the context, and nobody can sell it to you.
Anthropic, writing about its own model Claude, puts the test in a way that lands immediately with anyone who has ever trained a new front-desk hire:
"Show your prompt to a colleague with minimal context on the task and ask them to follow it. If they'd be confused, Claude will be too."
That's the whole discipline. If a brand-new hire couldn't write the text from your instructions, neither can a chatbot.
Anthropic also calls examples the strongest lever available — "examples are one of the most reliable ways to steer Claude's output format, tone, and structure" — and gives the one hard number in this whole literature: "Include 3–5 examples for best results," each one chosen to "mirror your actual use case closely" (Anthropic). That guidance is Anthropic's, written about Claude — treat it as a principle, not a spec sheet for every tool.
And expect to go around twice. OpenAI describes prompting as iterative: "start with an initial prompt, review the response, and refine the prompt based on the output." The first answer is a draft, not a delivery.
Why Do Free Prompt Lists Sound Generic?
Because bracketed placeholders delete the exact thing both OpenAI and Anthropic tell you to supply. A template that says write a caption for our [service] that highlights the [benefit] for [your ideal client] has had every specific stripped out of it before it ever reaches you. That's why the output comes back sounding like a stock photo: nothing went in, so nothing came out.
Look at what's being sold. A prompt pack aimed at nail technicians on Etsy — 100 prompts, $4.99 — has 9 sales and one review. Small numbers, and we're reporting them as small. But they tell you people will pay for something they could write themselves in two sentences.
Here's the reframe: you already own the three ingredients. Your captions that performed are the examples. Your actual policy is the context. Your client, described honestly, is the audience.
Five ChatGPT Prompts for Nail Techs
The brackets in these are for your material — that's the whole difference. Copy, fill, and change the tone words until it sounds like you.
The Rebooking Text
You are helping me write a short text message to a client of my nail salon.
Here are three texts I've sent before that got a reply. Match their tone, not their wording:
1. [paste a real text you sent]
2. [paste another]
3. [paste another]
The situation: she came in for a full set five weeks ago and hasn't rebooked. She usually comes every three weeks. She's easy-going and we joke around.
Write three options. Friendly, not pushy. Under 40 words each. No emoji.
The Consultation Reply
A client has messaged me before her appointment. Here is what she wrote, with her name and contact details removed:
"[paste the message]"
What my service actually involves: [say it in your own words, roughly is fine]
What I will not promise: [anything you can't guarantee]
Write a reply in a warm, professional tone. Plain language, no beauty-marketing adjectives, under 120 words. Do not add any claim about what a product does to the nail — leave that to me.
The No-Show Policy
Help me write a no-show and late-cancellation policy for my nail studio.
What actually happens here: [describe your last three no-shows in plain sentences — what happened, what it cost you, what you wish you'd said]
My booking system lets me: [take a deposit / charge a card on file / neither]
My tone with clients is: [pick two words, e.g. warm and firm]
Write it twice: a short version for my booking page, a longer version for the confirmation email. Then list every situation the policy doesn't cover, so I can decide what to do about those.
The Bad-Review Response
I run a nail salon and I need to reply publicly to this review:
"[paste the review, reviewer's name removed]"
What actually happened from my side: [tell it plainly, including anything that was our fault]
What I'm willing to offer: [refund / redo / an apology and nothing else]
Write a public reply under 80 words. Calm, specific, no defensiveness, no marketing language. Do not invent any detail I haven't given you, and do not promise anything beyond what I listed.
The Instagram Caption
Write five Instagram captions for a set I just finished.
Here are four of my own captions that got the most saves. Match this voice:
1. [paste]
2. [paste]
3. [paste]
4. [paste]
The set: [describe it the way you'd describe it to another tech — colours, technique, shape, length, what was hard about it]
My clients are: [who they actually are]
Call to action: [booking link / DM / none]
Under 30 words each. One of the five should be plain and unclever.
What Never Goes Into the Box
Strip the person before you paste the problem. On a personal ChatGPT account, your conversations are training data by default. OpenAI states it plainly: "ChatGPT, for instance, improves by further training on the conversations people have with it, unless you opt out" (OpenAI). On business tiers the default flips — "by default, we do not train on any inputs or outputs from our products for business users" — but the free account you're most likely using is the one that trains.
You can turn it off: personal accounts have an opt-out in Data Controls. Do that first. Then build the habit anyway:
- Her name. Out.
- Her phone number, email, address, or handle. Out.
- Anything about her health — a nail condition, a medication, an allergy, a pregnancy. Out.
- Photographs of her hands with anything identifying in frame. Out.
Paste the shape of the problem, not the person. "A regular client who hasn't rebooked in five weeks" carries everything the tool needs and none of what she'd mind you sharing. That's a working habit, not legal advice — but it follows directly from what OpenAI says its own default is.
What AI Should Never Write for You
Two things, and they are not close calls.
It must never decide what a product does to a nail. OpenAI says outright that "ChatGPT will occasionally make up facts or 'hallucinate' outputs" and that "outputs may be inaccurate, untruthful, and otherwise misleading at times" (OpenAI Help Center). Ask it for a service description and it will happily tell your clients that a treatment "repairs" or "heals" a damaged nail. The FDA is explicit that intended use can be established by "claims stated on the product labeling, in advertising, on the Internet, or in other promotional materials," and that "certain claims may cause a product to be considered a drug, even if the product is marketed as if it were a cosmetic." A cosmetic beautifies. A drug treats. One AI-drafted sentence moves you across that line. So let it draft how a service sounds; you decide what it does. If you want to see the work behind a single ingredient claim, read what we published on the TPO ban in Europe.
It must never write a review. The FTC announced a final rule banning fake reviews and testimonials on August 14, 2024, and it took effect on October 21, 2024 (FTC). According to the FTC's own announcement, the rule reaches reviews that misrepresent who wrote them, including reviews attributed to people who do not exist. Replying to a review is fine. Generating one is a different act entirely, and it is the one place on this whole list where a shortcut stops being a matter of taste.
Do You Have to Tell Your Clients?
This one is genuinely unsettled, and we're not going to pretend otherwise. What's settled: fake AI-generated reviews and testimonials fall under the FTC rule above. What's not settled: whether a salon needs to disclose that a caption, a DM, or a confirmation email was drafted with AI. We could not find a source saying you must, and we could not find one saying you needn't — so we're not going to tell you either.
Here's what we'd weigh instead. A client who feels handled rather than talked to will notice, and that's a trust cost no rule has to name. On the other hand, nobody discloses a spellchecker or a template. The distinction that seems to hold is intimacy: marketing copy is copy, but a message that reads as personal probably ought to be. You know your clients. That judgement was always yours.
Frequently Asked Questions
Can I paste my client's message into ChatGPT to help me reply? Not as it stands, on a personal account. OpenAI states that ChatGPT "improves by further training on the conversations people have with it, unless you opt out." Turn the opt-out on in Data Controls, then strip her name, phone number, address and any health detail before you paste. The tool needs the shape of the situation, never the person.
Can ChatGPT write my Google reviews? No. The FTC's rule banning fake reviews and testimonials took effect on October 21, 2024, and the agency's announcement describes it as reaching reviews that misrepresent who wrote them. Replying publicly to a review a client left is fine, and a good use of a prompt. Generating reviews for your own salon is not.
Can ChatGPT design nails or generate nail art? That's a different tool and a different search. Image generation for nail design — colour combinations, shape references, art concepts — is its own subject, and most lists ranking for "ChatGPT prompts for nail techs" quietly blend the two. This post is the business side: messages, policies, captions, replies.
Are the paid prompt packs worth it? Look at what you're buying. One guide aimed at nail technicians sells for $4.99 and has nine sales and one review. Inside are bracketed templates — the same shape as the free ones, with your context removed. Your own past captions and your own policy outperform any pack, and cost nothing.
Is AI going to replace nail techs? Not according to the numbers. BLS projects employment of manicurists and pedicurists to grow 9% from 2025 to 2035, adding 18,600 jobs. Nobody has automated a hand. What AI can take off your plate is the unpaid second shift — the captions, the policies, the messages written at 11pm.
Nail techs have always been the people who learn the new thing first and get credited for it last. This one isn't hard, isn't expensive, and doesn't ask you to become a technologist. It asks you to hand over the boring half of your week and keep the half that needs judgement.
So here's our ask. If you use a prompt that works, send it to us — and if there's something you'd never let AI write for you, tell us that too. We'd far rather publish one real prompt from a working tech than another list of templates. Meanwhile, for the business side of the salon, Krystal Oh's story and what happens when a salon gets caught in someone else's news cycle are both worth your time — as is Leafgel educator training, still the fastest way we know to get better at the part no machine is coming for.
