ChatGPT for Interior Designers: Ten Prompts for Briefs and Schedules

The prompts that save an interior designer the most time turn discovery notes into a one-page brief, list the questions the brief still can't answer, tidy a product list into a schedule, flag gaps in that schedule and draft the fortnightly client update. ChatGPT writes and sorts; you supply every supplier, product code, price and lead time.

These are paperwork prompts for professionals, not "describe my dream living room" prompts. They assume the design decisions are yours and the documents around them, the brief, the schedule and the emails, are what eat the evenings. Each prompt uses placeholders such as [CLIENT CODE] and [ROOM], so the same text serves every project and no client data goes in.

Follow me on Instagram@sagnikteaches

One invented project runs through all ten: a kitchen-diner and living-room refresh for a family of four with a dog, a budget band of $60,000 to $75,000 including the design fee, and a 16-week target. Every product, price and date below is made up to show the mechanics.

Connect on LinkedInSagnik Bhattacharya

I'd start with prompts 1 and 2, because discovery notes get written twice, once in a notebook and again as a brief. The schedule prompts come next, since a schedule is where one wrong code or price costs real money. The update prompts carry the least risk, because they report facts you already hold.

Subscribe on YouTube@sagnikteaches

Where each prompt sits in a project

The ten prompts follow the order work reaches your desk, which makes them easy to find when a stage starts. The right-hand column is the part that never moves to ChatGPT.

Four rules that apply to all ten prompts

1. Codes, not clients. Use [CLIENT CODE] and room names. Leave out names, addresses, phone numbers, floor plans with title blocks, and photos that show the front of the house or a name on a letter. State the budget as a band, not to the dollar. Some client contracts and supplier terms carry confidentiality clauses, so check them before pasting trade price lists or discount terms.
2. TBC beats a guess. Every prompt carries the same line: if a supplier, product code, price, dimension, finish or lead time isn't in my notes, write TBC. Without it, ChatGPT fills the gaps with plausible-looking specifics, and a plausible lead time is more dangerous than an empty cell.
3. Sums belong in a spreadsheet. ChatGPT formats numbers and doesn't reliably add them. Tell it to leave totals out, and add them where a formula can be audited. How to catch bad figures in quotes is covered in checking trade quotes against your schedule.
4. You send it. Nothing reaches a client until you have read it against the project file. The prompts produce drafts, not finished documents, and none of them is meant to send anything on its own.

Briefs: prompts 1 to 4

1. Discovery notes into a one-page brief

Discovery meetings produce fragments: a client's phrase, an offhand dislike, a budget said aloud once. This prompt shapes them into the page you send back for sign-off, and it drags the gaps into the open instead of papering over them. If you also run a written questionnaire, building and summarising an interior design client questionnaire covers that side; this prompt works on whatever notes you actually have.

Turn these discovery-meeting notes into a one-page design brief for the client file.
Sections: Goals / How they live / Must-haves / Dislikes / Constraints / Budget band and timing / Open questions.
Rules: use only what is in the notes. Where the notes are silent, write "Not stated" and add the gap to Open questions. Do not label a style unless the client used the word. Under 250 words.

Notes for [CLIENT CODE]:
- family of 4 (children 7 and 10), a dog
- kitchen-diner dark and cramped; wants it lighter; "room for everyone at one table"
- living room: TV dominates; "don't want to stare at a black screen"
- dislikes: high gloss, open shelving ("dust"), anything cold
- keep the existing oak floor in the living room
- budget $60-75k including fee; finished before [MONTH]
- one partner travels; decisions through a shared document

Illustrative output.
Goals: a lighter, roomier kitchen-diner with space for the whole family at one table; a living room where the television doesn't dominate.
How they live: family of four (children 7 and 10) and a dog; one partner travels, so decisions run through a shared document.
Must-haves: keep the existing oak floor in the living room.
Dislikes: high gloss, open shelving, cold materials.
Constraints: Not stated.
Budget band and timing: $60,000 to $75,000 including fee; finished before [MONTH].
Style: warm contemporary.
Open questions: Which walls can move? Is the television staying? Which rooms are in scope?

The extra Style line is the tell. It wasn't in the section list, and nobody said "contemporary": the client said nothing cold, which is a dislike, not a style. Delete it and let the gap show in Open questions. Two checks are worth the minute. The budget line should match your notes exactly, and "Constraints: Not stated" is your cue to ask about structural limits, leases and any approvals before concept stage.

2. The questions the brief still cannot answer

A brief with honest gaps is more useful than a tidy one, provided somebody closes the gaps. Feed the draft back in and ask for questions ranked by how much each answer could move the budget or the programme. The ranking is the useful part, because it tells you which question to ask first.

Here is the brief for [CLIENT CODE]. List up to 10 questions I still need answered before concept stage. Group by room. Tag each (B) if the answer could change the budget by more than about 10 per cent, (S) if it could change the schedule, (D) if it only changes the design. Do not ask anything the brief already answers. If a question is generic and would apply to any project, leave it out.

[paste the brief from prompt 1]

Illustrative output.
Kitchen-diner: (B) Is the wall between the kitchen and the hall structural? (B) Are the sink and hob staying where they are? (S) Does the kitchen need to be usable by a fixed date? (D) How many people sit at the table on an ordinary day?
Living room: (D) Where must the television sit for it to stop dominating the room? (D) Are you open to hiding it behind doors or a panel?
Whole project: (D) Would you like sustainable materials where possible?

The last question is generic, so the model broke the "generic" rule once. Every brief gets that question, and unless the client raised it, cut it. Put the (B) questions at the top of your call list: a moved wall or a moved sink changes the budget by more than any finish will, and it's better to hear that on day three than in week six.

3. A "what I heard" email that brings corrections early

Sending the brief back as a short, warm email invites corrections before concept work starts, when changing course costs the client nothing. Give ChatGPT the agreed brief and forbid it from adding promises about dates or fees.

Write an email from [DESIGNER FIRST NAME] to [CLIENT FIRST NAME] confirming what I heard in our first meeting. Use the brief below as the only source. Plain, warm, under 150 words. Ask for corrections by [DATE]. Do not mention fees, dates for later stages, or products. If a point is marked "Not stated", say I will come back to it on [DATE OF NEXT CALL].

[paste the brief]

Illustrative output.
Hi [CLIENT FIRST NAME], thank you for your time on [DATE]. Here is what I heard, so you can correct me before I start on concept ideas. You'd like a lighter kitchen-diner with room for the whole family at one table, and a living room where the television doesn't dominate. You want to keep the oak floor, and you'd rather avoid high gloss, open shelving and anything cold. Your budget is $60,000 to $75,000 including my fee, and you'd like it finished before [MONTH]. Please send any corrections by [DATE]. I'll come back to the structure of the kitchen walls on [DATE OF NEXT CALL]. Warm regards, [DESIGNER FIRST NAME]

Trace every sentence to a line in the brief, and keep the budget as a band, as the client said it. One thing I'd add by hand: the client's own phrase, "room for everyone at one table", quoted back. It shows you listened better than any paraphrase does, and it is exactly the kind of detail a model flattens.

4. Room scope and budget split for the proposal

Proposals need a scope per room and a budget split a client can follow. Decide the split yourself, work out the dollar amounts in a spreadsheet and give ChatGPT the finished figures to format. The prompt buys clear wording, not budgeting advice.

Format this as a proposal section for [CLIENT CODE]: one short scope paragraph per room (max 40 words each) and a list with Area, Share and Amount. Use my figures exactly. Do not add categories, contingency advice or percentages I have not given. If a scope item is missing, write TBC.

Planning figure: $70,000 (middle of the $60,000-$75,000 band, fee included)
Kitchen-diner 55% = $38,500
Living room 30% = $21,000
Design fee 10% = $7,000
Contingency 5% = $3,500
Total = $70,000

Scope notes:
- Kitchen-diner: cabinetry layout [TBC], lighting plan, table and seating, window treatments, paint
- Living room: media wall to hide the screen, sofa and rug, curtains, side tables, lighting

Illustrative output.
Kitchen-diner: a lighter, roomier layout with cabinetry [TBC], a new lighting plan, a table and seating for the whole family, window treatments and paint.
Living room: a media wall that hides the screen, a sofa and rug, curtains, side tables and layered lighting.
Kitchen-diner, 55%, $38,500. Living room, 30%, $21,000. Design fee, 10%, $7,000. Contingency, 5%, $3,500. Total $70,000. A contingency of 10 to 15 per cent is typical for projects of this kind.

The last sentence is the model volunteering a rule of thumb. I couldn't verify "10 to 15 per cent is typical" as a general rule, and the model can't either; your contingency comes from your own projects and your contract. Delete it. Then check the split sums to the planning figure: 38,500 plus 21,000 plus 7,000 plus 3,500 is 70,000.

Schedules: prompts 5 to 7

5. A messy product list into a clean schedule

A schedule is where one wrong code or price costs real money, so this prompt has the strictest rules in the set: ChatGPT sorts and labels what you paste, and adds nothing. To build a schedule from scratch and check it in more depth, see building and checking an FF&E schedule with AI.

Turn this list of selections into a product schedule. Columns: Ref, Item (with finish and size), Supplier, Code, Qty, Unit price, Lead time, Status.
Rules: use only what is in my list. If a supplier, code, finish, size, price or lead time is missing, write TBC. Do not total anything. Number refs by room (K-01 for kitchen, L-01 for living room). Set Status to "Selected" for every line.

Selections for [CLIENT CODE]:
- pendant light, opal glass, x3, [SUPPLIER A], $240 each
- bar stool, oak, 65 cm seat, x4, [SUPPLIER B], $185 each
- three-seat sofa, "oat" linen blend, [SUPPLIER C], $3,150
- rug, wool, 300 x 200 cm, [SUPPLIER D], $1,280
- side table, walnut, x2, $410 each, supplier TBC
- curtains, interlined, made to measure, 2 pairs, $690 per pair

Illustrative output. The schedule comes back looking like this, with one cell that breaks the rules.

The sofa's lead time, 8-10 weeks, appears nowhere in the list. It looks reasonable, which is the danger. Delete it and get the real figure, with the date, from the supplier's written quote. Notice what the model got right: every other unknown stayed TBC, and that is exactly the list of things you now have to chase.

6. A schedule check that finds what you missed

Once the schedule is yours, turn ChatGPT into a proofreader instead of an author. Ask for a list of issues with the Ref of each, and don't let it change a cell. You verify every issue. Some will be false alarms, which is a fair price for catching the real ones.

Check this product schedule for [CLIENT CODE]. Do not edit it. List problems only, each with the Ref: duplicates, blank cells that are not TBC, units that do not match (mm against cm), finishes that conflict inside one room (for example brushed brass beside antique brass), quantities that look odd for the item, and any line where Qty x Unit price is shown but does not match the line total. Then list anything you are unsure about separately.

[paste the schedule]

Illustrative output.
1. K-01: no size or drop height given.
2. K-02: seat height is 65 cm; the island or counter height isn't in the schedule, so the fit can't be checked.
3. L-03: a price is given but the supplier is TBC; confirm where the price came from.
4. L-01: finish "oat" may be a typo for "oak".
Unsure: whether three pendants suit the length of the table.

Triage the five flags. Three were real: the pendant size, the stool height against the island, and the side table's price with no supplier. One was a false alarm, since "oat" is a fabric colour. The last was a design opinion, not a check, so ignore it. The aim is a shorter list of things to verify, not a verdict, and a list that finds three real gaps in six lines is worth running on every schedule.

7. Ordering dependencies as questions, not durations

Lead times and sequencing decide whether the install date holds. ChatGPT can know neither, but it's decent at asking "does X have to happen before Y?" questions that you answer from supplier quotes and trade knowledge. Ask for questions, not durations, and for an owner against each. Drafting an interior design project timeline with lead times takes the planning further.

From this schedule and the target install date of [DATE], list the dependencies I should confirm, as questions. For each: the Refs involved, who answers it (designer, contractor or supplier) and what goes wrong if it is missed. Do not suggest durations or lead times. Do not assume anything is in stock.

[paste the schedule]
Target install: [DATE]. Trades on site: joiner, electrician, decorator.

Illustrative output.
K-01, K-02: Do the pendant positions depend on the final table and stool layout? (designer, electrician) If the position moves after first fix, the ceiling is opened up twice.
L-04: Who measures the curtains, and when? (designer, decorator, supplier) Made-to-measure curtains ordered before the walls are finished risk a wrong size.
L-01, L-02: Is the floor finished before the sofa and rug arrive, and does the building limit delivery times? (contractor, supplier)

Treat these as prompts for phone calls, not facts. The curtain point assumes late measuring, which is common but not universal, so confirm it with your curtain-maker. Don't let the "what goes wrong" lines turn into advice you pass to a client unchecked. The value is the question you hadn't thought to ask.

Updates and site visits: prompts 8 to 10

8. The fortnightly update with a status for each room

Clients worry most when they hear nothing. A fortnightly update with one status per room is cheap to produce when the raw material already sits in your project notes. You supply the statuses; ChatGPT supplies readable email.

Write a fortnightly project update from [DESIGNER FIRST NAME] to [CLIENT FIRST NAME] for [CLIENT CODE]. For each room give a status of On track, Needs your decision or Delayed, then one or two sentences. Use only the notes below. Put decisions needed from the client at the top, each with its deadline. Under 220 words. Warm and direct; apologise only where a status is Delayed.

Notes:
- Kitchen-diner: On track. Cabinetry fitted; worktop template booked for Thursday.
- Kitchen-diner: pendant K-01 delivered, one glass shade cracked in transit; replacement ordered, date TBC.
- Living room: Needs your decision. Sofa fabric choice (oat or stone) needed by [DATE] to hold the delivery slot.
- Living room: curtains measured after painting, ordered.

Illustrative output.
Hi [CLIENT FIRST NAME], here is where things stand.
Decision needed by [DATE]: your choice of sofa fabric, oat or stone, to hold the delivery slot.
Kitchen-diner: On track. The cabinetry is fitted and the worktop template is booked for Thursday. One glass shade on the pendants cracked in transit, and a replacement is on order.
Living room: Needs your decision. The fabric choice above is the only open item; the curtains have been measured and ordered.
I'll write again in two weeks, [DESIGNER FIRST NAME]

The status labels are yours to set. The model buried the cracked shade inside "On track" because the notes said "ordered". I'd keep the kitchen as On track and add a separate Delayed line for the pendant, so the client hears about it now and not at install. Also supply the date for the replacement as soon as you have it: "date TBC" in an email is an invitation to ask.

9. A discontinued or late item, with options attached

Bad news travels better with options. Choose the alternatives yourself, from real products you have checked, and let ChatGPT write the message around them. If you ask it for alternatives, it will name products that sound right and might not exist.

Write a short email from [DESIGNER FIRST NAME] to [CLIENT FIRST NAME]: the curtain fabric on line L-04 has been discontinued by the supplier. Offer the two alternatives below, say what is different about each, and ask for a decision by [DATE] so the install date holds. Do not suggest other products. Calm, practical, under 140 words.

Option 1: [FABRIC NAME 1], [COLOUR], same price, slightly heavier handle
Option 2: [FABRIC NAME 2], [COLOUR], $60 more per pair, closer match to the sofa
If there is no decision by [DATE]: the curtains slip, not the whole room.

Illustrative output.
Hi [CLIENT FIRST NAME], a small change to report: the supplier has discontinued the curtain fabric we chose. There are two good alternatives. Option 1, [FABRIC NAME 1] in [COLOUR], is the same price with a slightly heavier handle. Option 2, [FABRIC NAME 2] in [COLOUR], is $60 more and matches the sofa more closely. Could you let me know by [DATE]? That keeps the curtains on schedule, and the rest of the room isn't affected. Kind regards, [DESIGNER FIRST NAME]

Check the money. The email says "$60 more" without saying per pair. With two pairs on line L-04 the real difference is $120, and a client who expects $60 will query the invoice. State the total effect, and keep the "not the whole room" line only if the supplier's delay really is limited to the curtains.

10. Snags from voice notes, grouped by trade

Walk the site, talk into your phone and let ChatGPT build the list. Transcripts ramble, so the useful steps are grouping by trade and keeping your own words for priority. Paste the transcript text only; photos of a client's home stay out of the chat, and you refer to them by number.

Turn this site-walk transcript into a snag list for [CLIENT CODE]. Group by trade (joiner, electrician, decorator, other). One line per snag: room, what is wrong, photo ref if I gave one. Use my priority words exactly (urgent, before handover, cosmetic); if I gave none, write "priority?". Do not merge snags or invent causes.

Transcript: "Kitchen. Gap at the top of the end panel by the fridge, joiner, photo 14, before handover. Pendant over the table looks off-centre, electrician, photo 15, that's urgent, the table arrives Friday. Living room. Scuff on the wall above the door frame, decorator, cosmetic. Curtain track a few millimetres out of level at the left end, not sure who."

Illustrative output.
Joiner: Kitchen, gap at the top of the end panel by the fridge (photo 14), before handover.
Electrician: Kitchen, pendant over the table off-centre (photo 15), urgent.
Decorator: Living room, scuff on the wall above the door frame, cosmetic; probably caused during furniture delivery.
Other: Living room, curtain track a few millimetres out of level at the left end, priority?

"Probably caused during furniture delivery" wasn't in the transcript. Delete causes: in a list that goes to a trade, a guessed cause reads as blame. The model did keep "priority?" for the curtain track, which is what you want: a visible gap for you to fill, not a priority it decided for you.

The schedule total ChatGPT got wrong

Asked to "add up the line totals" for the six-line schedule above, ChatGPT might return $8,190, a figure I've made up to show the kind of slip that happens. The true total is $8,090, and the line totals below show where it comes from.

A $100 error in six lines is easy to miss, and on a 140-line schedule one wrong line sits unnoticed among the rest. Build the totals in a spreadsheet instead: a line total column that multiplies Qty by Unit price, and a schedule total that sums it. As a cross-check, a SUMPRODUCT of the Qty and Unit price columns should give the same figure in a single cell. Two formulas that disagree tell you a row has gone wrong, which is exactly what a pasted-in total never will. If spreadsheet formulas are a weak spot, the ChatGPT guide to Excel covers asking for them.

Keeping the ten prompts in a Project

A ChatGPT Project holds instructions and files together, so each new chat inside it starts with your voice, your rules and your templates already loaded. For this set, put the four rules and your tone into the Project's instructions, and keep only templates in its files: a blank brief layout, the schedule column list, and two past update emails with every client detail removed. Live client files stay out: anything you upload stays in the Project until you delete it, and in a shared Project every collaborator's chats can draw on it.

The limits suit a studio. A handful of template files sits well inside OpenAI's per-Project allowance, which runs from 5 files on Free to 25 on Go and Plus and 40 on Pro, Business, Enterprise and Edu. Sharing works on every plan, so a freelance assistant or junior designer can join the Project: Plus allows 10 collaborators, and Business can also invite a whole workspace group. A shared Project runs on project-only memory, which stops it drawing on anyone's memories or chats outside the Project. A studio of two or more can use ChatGPT Business at $25 a seat monthly or $20 billed annually (two-seat minimum), and business data isn't used for training by default. A solo designer can start free and move to Plus at $20 a month. Custom GPTs are the wrong place to start now: OpenAI has scheduled them to stop running on 11 December 2026, and its migration route for existing ones turns each into a plugin, so a Project is the simpler home for ten prompts. For the studio admin around these prompts, AI for interior design studio admin covers sourcing, quotes and updates.

Questions designers ask before using ChatGPT on client work

Can ChatGPT choose products for a client?

It can suggest categories to consider, but it can't see current stock, trade prices, lead times or discontinued lines, and it will describe products that don't exist with complete confidence. Choose the products yourself, then use ChatGPT to format, compare and write about the ones you have chosen.

Is it safe to put client details into ChatGPT?

Keep names, addresses, phone numbers and identifiable photos out, and use a client code instead. On personal plans, check the model-training switch under Data controls in settings; business plans don't train on business data by default. Read your client contracts too, because some carry confidentiality terms that cover tools like this.

Which ChatGPT plan suits a one-person studio?

Start on the free plan to test the prompts. If you use them daily, Plus lists at $20 a month. ChatGPT Business needs at least two seats, at $25 each monthly or $20 each billed annually, and suits a studio of two or more that wants shared Projects and no training on business data by default.

How do I stop ChatGPT inventing lead times and prices?

Put the same line in every prompt: write TBC for anything I haven't supplied. Never ask it to look up a price or a delivery estimate. Then scan the output for numbers that aren't in your notes. A lead time belongs to the supplier's written quote, with the date you received it.

Further reads

• Interior Design Proposal Template: Scope, Fees and Phases With ChatGPT — Turn the agreed brief into a proposal with scope, fees and phases.
• Can AI Draft an Interior Design Contract? Terms a Solicitor Checks — What a solicitor should check before you use AI-drafted terms.
• How Interior Designers Use AI for Mood Boards and Presentations — Presentation boards once the brief is signed off.
• How to Price Interior Design Services: Compare Fee Models With AI — Compare fee models before you quote against the brief.
• AI Room Visualisation From Client Photos: A Designer's Workflow — Show a client their own room with the scheme in place.
• Interior Designer Software Compared: Where AI Add-Ons Earn Their Fee — Where design-software AI add-ons earn their fee.
• AI Marketing Tools for an Interior Design Business: Houzz and Pinterest — A three-hour monthly routine for interior designers: one project card and photo set become fresh Houzz, Pinterest and Google profile content.
• ChatGPT for Building Design: Briefs, Options and Code Checks — Brief intake, layout options and a code-check log for a small design practice, with the line ChatGPT must not cross.
• ChatGPT for Designers: Turning Client Feedback Into Revision Plans — Prompts and wording for freelance designers: sort a feedback round into a revision plan, send the client a summary, and flag extra work politely.
• AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.

Sources: OpenAI's ChatGPT pricing page and help-centre notes on Projects, file limits, data controls and business-plan training defaults.