Here’s a scene playing out in shops all over right now. A client comes in wanting a sofa redone, you work up an honest number, and a day later they’re back saying ChatGPT told them it should cost half that. A Las Vegas designer, Eryka Glover, told Business of Home about a version of this — a client pushed back on a “high-end, luxury” budget after running it past the chatbot. Her move was perfect. She said, sure, let’s run it again, using your exact words: high-end, luxury. And ChatGPT came back with a higher number than she’d quoted. End of argument.
That story is the whole reason to read this. Not because AI is coming for your craft — it isn’t, and we’ll get to why. But because your clients are already using it, so you may as well use it better than they do. The good news: for the parts of your week that aren’t actually upholstery — the estimates, the “your chair’s ready” texts, the care sheets, the marketing — ChatGPT is a genuinely great little assistant. It just needs to stay at the front desk and nowhere near your cut plan.
Let me show you the workflow, the copy-paste prompts, and the hard line you never cross.
What ChatGPT is (and isn’t) for your shop
Think of it as a writing assistant that works from your numbers and your knowledge. It’s good at turning things you already know into clean, professional words. It is not good — and in some spots it’s flat wrong — at anything that needs measuring, touching, or judging a real piece of furniture.
The one rule that keeps you out of trouble: you bring the price, the yardage, and the fabric call. It writes the words around them. Cross that line — let it price a job or guess your yards — and it’ll burn you. There’s a great line from an upholsterer on Reddit that says it better than I can: “Please don’t ever use ChatGPT for pricing information. Generative AI has literally no idea how to accurately assess a completely non-digital process.” That’s it exactly. Your work isn’t digital. The chatbot only knows text.
The workflow: write estimates and updates in minutes
Every one of these follows the same rule — you own the numbers and the fabric decisions, the bot just makes them read well. That’s what makes it safe.
1. Estimates and line items — you price, it writes. You already know what a job costs. What eats your evening is typing “reupholster 3-seat sofa, new foam, matched skirt” into something a client reads as professional. Hand it your rough notes and your prices, and lock the numbers:
You're helping an upholstery shop write a customer estimate.
Turn my rough notes into a clean estimate with a short scope of work,
clear line items, and an exclusions line. Keep ALL of my prices and
measurements exactly as written — do not estimate, change, or add any
numbers. If I didn't give you a number, leave a blank.
Shop: [name]
Client / piece: [e.g., mid-century 3-seat sofa, client's own COM fabric]
My notes: [e.g., strip & rebuild, new high-density foam, 14 yds client
fabric, matched skirt, ~18 labor hrs at $75/hr, foam & supplies $220,
6–8 wk lead time]
Tone: warm, professional, no fluff.
That line — keep all my prices and measurements exactly, don’t estimate — is the guardrail written right into the prompt. You measured the 14 yards. It just types them up nicely.
2. Fabric-care and aftercare sheets. This is a quiet time-saver. You know a linen blend needs different care than a performance polyester — turning that into a clean client handout is exactly the kind of “make my knowledge readable” task AI is good at. Give it the fabric type and your care points, get a tidy sheet back to hand over with every finished piece.
3. Project-status texts. “Your fabric’s arrived — your chair’s scheduled for next week.” “Just a reminder your balance is due on pickup.” “Current lead time is running 6 to 8 weeks with fabric delays.” These are polite, repetitive, and easy to put off. ChatGPT drafts them in seconds; you tweak so it sounds like you.
4. Deposit and turnaround wording. The awkward-to-write stuff — your deposit policy, your warranty language, the “why it takes this long” explanation. AI writes a clear first draft. (Anything that’s a real contract term, run past a pro before it’s binding — but for plain-English client wording, it’s great.)
5. Marketing and getting found. Your service pages — sofa reupholstery, antique chairs, marine cushions. The before/after post that shows off a rebuild. Your Google Business Profile. Seasonal notes. None of it is your craft, all of it takes time, and a first draft in your voice is thirty seconds out. One shop, McGovern Upholstery, has said publicly that AI’s been a “tremendous help” running the business side of their restaurant-booth reupholstery work. That’s the sweet spot — the office, not the workbench.
When a client says “ChatGPT says you’re too expensive”
It’s going to happen. A client uploads a photo, asks a chatbot what a redo “should” cost, and comes back quoting a number from nowhere. Don’t discount out of reflex and don’t argue — there’s a calm way through, and Eryka Glover’s story is the template. Three moves:
Reframe, then re-run it their way. Acknowledge they did some homework, then explain what the tool can’t see: “Glad you looked into it — but ChatGPT can’t see today’s fabric prices, the actual yardage your piece takes, or the frame work underneath. Its number’s a guess from other jobs online.” And if you want the Glover move: offer to run it again together, using the real scope — high-density foam, matched pattern, frame repair, your fabric grade. Nine times out of ten, once the bot has the real inputs, its number climbs toward yours.
Walk them through the estimate. Transparency wins. Break it down — fabric grade and yardage, foam and supplies, labor hours, frame repair, your lead time. The second they see real costing, “ChatGPT said” stops sounding like a fact.
Compare scope, not just price. Ask what they told the bot. It almost always assumed a simple slipcover-grade job — no frame work, no pattern matching, builder-grade fabric. That’s a different piece of furniture than the one you’re quoting. Line up scope to scope and the gap explains itself.
And yes — you can draft that whole reply with ChatGPT too. You bring the real reason your price holds; it helps you say it without heat.
What this means for you
If you’re a solo upholsterer: this is your evenings back. Batch the paperwork into one short session — estimates, the status texts, the care sheets — and get back to the bench. Save two or three prompts with your shop details baked in. Just keep every number and every fabric call on your side.
If you run a small shop with a helper: make “AI writes the words, the owner sets the price and the yardage” a house rule. Your estimates read consistently no matter who typed them, and nobody’s ever guessing at a number the bot made up.
If you do antiques or high-end custom work: you’re the most protected of anyone here — and you should lean into it. AI can’t assess a century-old frame, can’t judge hand-tying, can’t feel fabric stretch. Say that out loud to clients. It’s not a weakness that a chatbot can’t do your work; it’s the entire reason they’re paying a craftsperson and not printing a slipcover.
If a client’s already pushed back with AI: don’t cave. Reframe, re-run it with the real scope, break down your costing. You’ll keep more of those jobs — and more of your margin — than you’d expect.
What it can’t do — and never should
- It can’t price fabric. Grade, rub counts, UV resistance, fire codes — those come from your mills, distributors, and sample books. Not a chatbot guessing from text.
- It can’t calculate yardage. Pattern repeat, fabric width, direction, cushion count — generic AI is unreliable here. One upholsterer watched ChatGPT call a boat cover 50 yards when the real answer was 12. (Purpose-built shop tools like StitchDesk have a real yardage calculator — but even that is a starting point you check against your own cut plan, especially on antiques.)
- It can’t see a frame. A cracked rail, loose joints, old water damage — it only guesses from words. Structural calls stay yours.
- It gets facts wrong. Dates, lead times, fabric names. Check everything before it reaches a client.
- It can’t do the craft. The rebuild, the hand-tying, the eye for what a piece needs — that’s you. AI just buys back the hours you were losing to the keyboard.
The bottom line
Your clients are already running your quotes through ChatGPT. So use it better than they do. Let it write the estimates, the care sheets, the updates, and the marketing that were quietly stealing your evenings — while you keep every price, every yard, and every fabric call exactly where they belong, in the hands of the person who actually knows the furniture. The chatbot can’t feel a frame or match a repeat. That’s not a bug in your business. It’s the whole reason you have one.
Want to build this into a real system — the estimate prompts, the client updates, and getting found when someone searches for reupholstery near them? AI for Small Business gets you drafting fast without giving up control of your numbers, Get Found by AI: Local Business Visibility helps you be the shop that shows up when people ask, and AI Marketing Strategy turns a steady content habit into real inquiries.
Sources
- Is ChatGPT influencing your client? — Business of Home
- StitchDesk — AI-Powered Upholstery Shop Management
- StitchDesk Launches as the First Operating System Built Specifically for the Upholstery Industry
- Upholstery manufacturers eye AI for efficiencies, cost savings — Designers Today
- Can robots make upholstery? This startup is betting on it — Business of Home
