How to Create Outfit and Styling Posts for a Boutique With AI

Start from your real stock: give an AI assistant your stock list with colours and sizes left, ask for outfit combinations built only from what is on your rails, photograph those outfits on a person or mannequin, then let AI write the shot list, carousel text and captions. AI plans and writes; the clothes in the photos stay real.

Run as a weekly routine, that produces three to five styling posts from about three hours of work, with prompts you can copy below. The caveat worth stating first: styling posts sell because customers trust that the linen shirt looks like that on a person. If AI changes the garment, the post turns into a returns problem. AI model shots are a separate question, covered in our tutorial on try-on images without a photoshoot.

Follow me on Instagram@sagnikteaches

What a styling post does that a product photo can't

A product photo answers "what is it?". A styling post answers the two questions that stop people buying clothes online: "what would I wear it with?" and "how would it look on someone like me?". Four formats do most of the work for boutiques:

Connect on LinkedInSagnik Bhattacharya

• One piece, three ways: a hero item styled for work, weekend and evening. Good for items that are selling slowly because people cannot picture them.
• Capsule: five pieces that make ten outfits. Sells several items at once and suits new-season arrivals.
• Occasion edit: wedding guest, first day back at work, a long weekend away. Ties stock to the reason people are shopping.
• Fit comparison: the same dress on two people of different sizes, with their heights and sizes stated and their agreement to appear. It answers more DMs than any caption.

AI helps most with the first three, because they are combinatorial: working out which of 140 styles go together, in which colours, in sizes that are still available, is the slow part.

Subscribe on YouTube@codingliquids

Step 1: export a stock list AI can style from

From your till or shop system, export a list with one row per style: name, category (top, trousers, knitwear, shoes, accessory), colour, fabric, price, sizes still in stock, and units left. Add one column by hand: a few words on the look ("relaxed, boxy", "fitted, formal"). Ten minutes for a boutique of a hundred-odd styles.

A few illustrative rows, as the AI will see them:

Knitted waistcoat | knitwear | oat | wool blend | $65 |
  10,12,14,16 | 9 left | cropped, boxy
Wide-leg trousers | trousers | chocolate | linen | $72 |
  8,10,12,14,16 | 14 left | relaxed, high waist
Poplin shirt | top | white | cotton | $48 |
  8,10,12,14 | 11 left | oversized, crisp
Satin slip skirt | skirt | sage | satin | $58 |
  10,12 | 3 left | fluid, midi
Loafers | shoes | black | leather | $95 |
  37-41 | 12 left | chunky sole

The slip skirt has only two sizes left, so the filter below removes it.

Filter before the AI sees it. Leave out anything with fewer than three sizes left. Featuring a skirt that only remains in a size 6 produces twenty "do you have it in a 14?" messages and a lot of disappointment. Items down to their last few sizes are better as a separate "last chance" story than inside a styling post.

Step 2: ask for outfits built only from your rails

Paste the filtered list into ChatGPT, Claude or Gemini with this prompt:

You are the stylist for a small clothing boutique.
Below is our current stock (name | category | colour | fabric |
price | sizes in stock | look).

Create 6 outfits for a "[format, e.g. one piece three ways]" post
featuring [hero item]. Rules:
- Use ONLY items in the list. Never invent a piece.
- Each outfit: 3-4 items including the hero item.
- Every item in an outfit must be in stock in at least three of
  these sizes: 10, 12, 14, 16.
- Give the total price of each outfit.
- One sentence per outfit on why it works (colour, proportion,
  occasion), in plain words a customer would use.
- Flag any pairing that relies on a colour I should check in
  person, e.g. two different greens.

Stock list:
[paste]

An illustrative extract of what comes back for "one piece, three ways" around the oat waistcoat:

Outfit 1 - Work: waistcoat over the white poplin shirt, chocolate
wide-leg trousers, black loafers. $280. The crisp shirt collar
and cuffs show at the neck and wrists, and the high-waist trouser
meets the cropped waistcoat so the proportions stay balanced.

Outfit 2 - Weekend: waistcoat worn alone, straight-leg jeans,
white trainers. $...  [jeans and trainers are not in the list]

Outfit 3 - Evening: waistcoat with the sage satin slip skirt...
  [skirt was filtered out; sizes 10 and 12 only]

What to fix: outfit 1 is usable as written. Outfit 2 invents items you do not stock, which the prompt forbids; this happens most often when the stock list has no obvious match for a style ("weekend" with no denim on the rails). Outfit 3 pulls an item the filter removed because the AI remembered it from earlier in the chat. Start a fresh chat for each week's list, and when an outfit breaks the rules, say so and ask for a replacement from the list only.

Expect to reject one or two outfits. AI judges colour from words, so "sage" and "olive" may look fine on paper and clash on the rail. That is why the prompt asks it to flag colour risks, and why you pull the actual pieces together before any photo is taken. Choose the three or four outfits that look best in person.

Step 3: a shot list one person can shoot in 45 minutes

Ask the same chat to turn the chosen outfits into a shot list:

Turn outfits [numbers] into a shot list for one person with a phone.
For each outfit: a full-length mirror or tripod shot, a close-up
of the key detail (fabric, button, print), and one 5-second
movement clip (walking, turning). Order the shots so each outfit
is only put on once. Portrait framing, subject centred.

Read the list before the shoot, not during it. An illustrative first draft for the three waistcoat outfits began: "Outfit 1: full-length tripod shot; close-up of the pearl buttons; walking clip. Outfit 3: full-length; close-up of the shirt collar; turning clip. Outfit 2: …" Two problems. The waistcoat has horn buttons, so the close-up note came from nowhere; the model filled a detail it didn't have. And it put outfit 3 before outfit 2, even though outfits 1 and 2 share the white shirt, which means taking it off and putting it back on. Correct the detail ("close-up of the knit texture at the hem"), reorder so shared pieces stay on, and the shoot drops back to the 45 minutes it should take.

Shoot in portrait at 4:5 (1080 by 1350 pixels), the long-standing default for feed posts and carousels. Instagram's profile grid now shows posts in a taller 3:4 crop, so keep the outfit centred with space above the head and below the shoes, or the grid thumbnail will cut the shoes off. A carousel can hold up to 20 photos and clips, but styling posts rarely need more than eight.

Step 4: write the carousel slide by slide

A carousel that sells has a job for each slide. Give the AI that structure rather than asking for "a caption":

Write on-image text for an Instagram carousel, max 10 words per
slide. British English, no hashtags on slides.
Slide 1: hook naming the hero item and the idea
  (e.g. "One linen shirt, three weekends").
Slides 2-4: one outfit each. Name every piece and its price.
Slide 5: fit note for the hero item (runs small / true to size /
  size up), plus the height and size of the person pictured.
Slide 6: how to buy or reserve: link in bio, DM to hold, in shop.
Use these outfits and facts only: [paste chosen outfits + fit notes]

An illustrative result for the waistcoat, with one line to fix:

• Slide 1: "One oat waistcoat, three days out"
• Slide 2: "Work: waistcoat $65, shirt $48, trousers $72, loafers $95"
• Slide 3: "Weekend: waistcoat alone, trousers, loafers. $222"
• Slide 5: "Fits true to size. Pictured: 5ft 7, size 12"
• Slide 6: "In shop Tue-Sat, link in bio, or DM to hold"

Slide 3 is the one to catch: $65 plus $72 plus $95 is $232, not $222. Models add up prices unreliably when they're writing short text at the same time, so check every total on a calculator or ask for the working in a separate message. Slide 5 is only right if your fit note says "true to size"; if the note says "boxy, size down if between sizes", the slide must say that, because it's the line that saves a return.

Put the text on the slides in Canva or your usual design tool, using the same two fonts and position every week. Consistency is what makes your posts recognisable when someone scrolls past quickly.

Step 5: captions that answer the DM before it's sent

The caption carries the facts the slides could not: fabric, care, full size range, and what to do next. Every fact you put here is a DM you do not have to answer. A template to give the AI along with your brand voice notes:

Caption structure:
Line 1: the idea in one sentence, in our voice.
Line 2-3: the hero item: fabric, fit note, sizes available.
Line 4: a question that invites a real reply
  ("Which one would you wear first?").
Line 5: how to buy: in shop [days/hours], online via link,
  or DM us to hold one until [day].
Hashtags: 3-5, specific to the item and occasion, not generic.

Before and after, for the same waistcoat (both illustrative). A first attempt from a one-line request:

Obsessed with this waistcoat! The ultimate layering piece for every occasion. Shop now!

From the template, with the fit note and sizes in the prompt:

One oat waistcoat, three days out of the house. It's a cropped, boxy fit in a soft wool blend, true to size; sizes 10 to 16 on the rail today. Which would you wear first: the shirt-and-trousers look, the dress, or on its own? In the shop Tuesday to Saturday, online through the link, or message us and we'll hold one until Friday.

The second caption answers fit, fabric, sizes and how to buy, which are the four DMs the first caption would have generated.

Watch the hashtags in particular. Asked for "hashtags", models still tend to return a dozen or more, starting with #fashion, #ootd and #style. Instagram has capped posts at five hashtags since December 2025, so a long list is wasted at best. Broad tags like those also put a small boutique's post among millions of others. For the waistcoat post, five that describe the item and the occasion do more: #knittedwaistcoat #capsulewardrobe #linentrousers #workwearstyle #independentboutique. Put "exactly five hashtags, no generic fashion tags" in the prompt so you don't have to trim them each week.

If your captions still read like a machine wrote them, the fix is usually better voice notes rather than a better model; see writing Instagram captions with AI that sound like you. And line 5 only works if someone answers the DMs promptly, which is the subject of answering boutique DMs within an hour.

Step 6: reuse the same outfits for a Reel and a story

You shot a movement clip for every outfit, so the week's Reel is already filmed. Ask the AI to cut it on paper first:

Write a 15-second Reel plan from these clips: [list clips].
Give each clip a length in seconds, the on-screen text (max 6 words),
and the order. Open with the most striking outfit, end on the
hero item alone with its price. Suggest where the beat should
change, but don't name a specific song.

An illustrative plan that comes back: "0-3 s: outfit 3 walking towards the camera, text 'One waistcoat'. 3-6 s: outfit 1 turning, 'Work'. 6-9 s: outfit 2 walking away, 'Weekend'. 9-12 s: close-up of the knit, 'Soft wool blend'. 12-15 s: waistcoat alone on the rail, '$65, sizes 10-16'. Beat change at 3 s and 9 s." That's five clips at three seconds each, with the price on the final frame, which is the one people pause on. If a plan comes back with clips you didn't shoot ("a flat lay of the full outfit"), swap in one you have rather than reshooting.

Edit it in the Instagram editor or CapCut following that plan. Stories get the leftovers: a poll ("work or weekend?"), a close-up of the fabric with a question sticker, and the fit note as a single slide. None of this needs new photos, which is the point. One 45-minute shoot should feed the whole week.

Staff and customers as models

Fit comparisons work best with real people of different heights and sizes, and staff are the usual choice. Ask each person in writing before their photo is used, agree how long you will keep posts up, and let them veto a picture. If a customer offers to model, the same applies, plus a clear note of what, if anything, they receive in return. Never ask an AI tool to alter a person's body shape or skin in a styling photo; it defeats the point of showing real fit.

Where AI images fit into styling posts, and where they don't

Meta may add an "AI info" label to images it detects as AI-generated, and it expects disclosure for realistic AI video. Our tutorial on labelling AI-generated images and video explains when to add labels yourself.

A weekly styling rhythm for a small boutique

Before scheduling, run each post through a pre-publish checklist for AI-drafted posts. The most common error in styling posts is a price that changed since the stock export.

Four styling posts a week from 140 styles

Imagine a boutique with 140 styles in stock (the numbers throughout are for illustration) that posts one styling carousel a week, taking about six hours: deciding outfits by trial and error on the shop floor, shooting, then writing captions in the evening.

With the routine above, the owner runs the outfit prompt for a "one piece, three ways" post around a knitted waistcoat that had been slow to sell, a capsule of five new-season pieces, an occasion edit for summer weddings, and a fit comparison of a bestselling midi dress on two members of staff. Outfit planning drops from about two hours to 50 minutes including the rail check, captions and slide text from two hours to 45 minutes, and shooting stays about the same because the shot list removes changing back and forth. Four posts take roughly three and a half hours rather than one post taking six.

Two things went wrong in the first fortnight, as they usually do. The AI paired two greens that looked wrong together in person (caught at the rail check), and one caption quoted a price from before a markdown (caught by the till check). Both checks stayed in the routine.

How to tell which styling posts earn their time

• Saves and shares matter more than likes for styling content; people save outfits they intend to copy.
• DMs that mention the post ("the waistcoat from Tuesday"), which your DM sheet can record.
• Second-item sales: how often customers buy the hero piece with another item from the post. Staff can note it at the till for a month.
• Sell-through of the hero item in the fortnight after posting, compared with the fortnight before.

A quick sum for the waistcoat, with illustrative figures: in the fortnight before the post, 2 of 11 sold, a sell-through of about 18%. In the fortnight after, 6 of the remaining 9 sold, about 67%, and staff noted that 4 of those 6 buyers also bought the poplin shirt or the trousers. That's the pattern that justifies a format: the hero moves, and the second-item sales prove the outfit did the selling rather than a markdown.

One edge case to plan for: the hero item selling out in the popular sizes mid-week while the post keeps collecting views. Edit the caption the same day ("now sizes 10 and 16 only; the trousers are still in every size") and post a story saying the same, so the DMs that follow are about what you can still sell.

After a month, keep the two formats that move stock and drop the rest. The DMs and comments these posts produce are also useful buying data; our tutorial on spotting trends before buying stock shows how to use them.

Further reads

• Canva AI for Small Business: Which Features Save Real Time? — Which Canva AI features help with carousel layouts.
• How to Create On-Brand Social Media Graphics With AI — Keep text slides and graphics consistent week to week.
• Best AI Social Media Tools for Small Businesses (2026) — Scheduling and planning tools compared for small teams.
• How to Build a Brand Voice Guide That AI Can Follow — Write the voice notes your captions prompt relies on.
• How Much Time Can AI Save on Social Media Management? — Realistic time savings to expect on social content.
• Should a Small Business Use AI to Write Its Social Media Posts? — Where AI-written social posts help and where they hurt.
• AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.

Sources: Instagram carousel and image-size guidance as published for 2026 (up to 20 items per carousel; 4:5 portrait posts; 3:4 profile grid); Meta guidance on AI-generated content labels.