The Blog

The Debate Is Loud. The Testing Is Thin.

Every week brings another hot take about AI gutting the fashion workforce — the designer, the photographer, the buyer, the model. Most of those takes come from one of two places: people selling AI tools, or people afraid of them. Neither group has a strong incentive to show you the part where things go wrong.

So I'm going to stop speculating and start testing.

Who's Running This

I've spent 25 years inside apparel — grew up on a garment factory floor, and helped take the Lee men's business from $5M to $200M in annual sales. I've held most of the roles people say AI is about to take. For the past year, I've also been on the other side of that equation, building AI tools for fashion at Shanley AI.

That combination puts me in an unusual spot. I know what these jobs actually require, not just what they look like from the outside. And I know where current AI tools are genuinely capable versus where they make impressive-looking demos that would fall apart in a factory or a buying meeting.

The Series: 12 Jobs, 12 Verdicts

Twelve jobs, one a week, then a finale scorecard. One question each week: can AI actually do this?

The roles, in order:

1. Trend Forecaster
2. Designer
3. Technical Designer
4. Sourcing / Production Manager
5. Merchandiser / Buyer
6. Fashion Marketer
7. Copywriter / Brand Voice
8. Stylist
9. Photographer / AI Models
10. E-commerce Manager / Virtual Try-On
11. Customer Service / Clienteling
12. Fashion CEO / Strategy

The Finale: full scorecard

Each week ends with one of three verdicts:

• AI wins — the tool does the job well enough that a real brand could rely on it today
• Human wins — the tool falls short in ways that would cost a brand real money
• Hybrid — the tool handles part of the job, and a person still has to own the rest

No hype. No doom. Just the work.

What Each Week Looks Like

Monday — This Week in AI × Fashion. The biggest AI-and-fashion story of the week, and what it actually means for the people doing the work.

Tuesday — The Test. AI goes to work on that week's role, on screen, with every prompt shown.

Wednesday — Hot Take. The insider truth about the job that most AI demos skip.

Thursday — How-To. How a real brand could use AI for this role today, step by step.

Friday — The Verdict. Can AI do the job? Plus one story from 25 years in the industry.

The Rules I'm Holding Myself To

Show the prompts. Every test shows exactly what I typed. You should be able to run the same test yourself.

Show the failures first. A demo that only shows the good output is an ad. When a tool breaks a tech pack or renders fabric wrong, you'll see that too.

Test other people's tools, not just mine. I'll use my own apps where they genuinely fit — trend intelligence, 3D design, tech packs — and name-brand tools everywhere else. An honest test includes tools I didn't build.

Grade it like a boss would. The question isn't "is this impressive?" It's "would I sign off on this, and would it survive a factory, a buyer, or a customer?"

Why Fashion Is a Good Place to Run This Test

Fashion is one of the most demanding places to evaluate AI because every output eventually becomes something physical. A trend call becomes a sample. A tech pack becomes a garment. A buy becomes inventory that either sells through or gets marked down.

You can't hide a bad output for long. The feedback loop is brutal and fast — which makes it an honest testing ground.

This matters well beyond apparel. If you lead any creative or operational team, the question is identical: which parts of the work can AI take on today, which parts fall short, and how do you know the difference before it costs you?

A Worked Example: Why the Photographer Test Will Be Interesting

The photographer job is a good preview of what this series is designed to surface. On one side, brands are already putting AI-generated imagery into their product catalogs, and every shoot that disappears also hits the assistants, stylists, retouchers, and set designers who work on it.

Meanwhile, early AI fashion imagery still struggles with fabric rendering inconsistencies and anatomical errors. Quality has improved, but the problems haven't disappeared. So which is it — does AI win the photographer job, does a human win, or is it a hybrid?

I don't know yet. That's the point. I'm going to run the test and show you what happens.

Follow Along

The series runs every week on Instagram, TikTok, and LinkedIn, and I'll collect the verdicts here on the blog as they land. If there's a fashion role you want me to put to the test, tell me in the comments.

And if you're a brand trying to figure out where AI fits in your own team before making expensive decisions, that's exactly what we help with at Shanley AI.

Let's find out who wins.

The Report Nobody Actually Uses

Every design room has a version of the same ritual. A forecasting PDF arrives — dense, beautifully laid out, and roughly six months behind the market by the time anyone cracks it open. The team flips through it. A few tearsheets get pinned to a wall. Then everyone goes right back to designing from instinct, because the report answered questions nobody was asking on that specific Tuesday.

That's not a failure of effort. It's a failure of format. And it's been happening for decades at a cost that's hard to justify.

What Trend Intelligence Actually Costs — and Who It Leaves Out

Enterprise forecasting platforms like WGSN can run up to $25,000 per year. For large brands with dedicated strategy teams, that's a line item. For a founder building her first collection or a mid-size brand trying to keep margins intact, it's often not an option at all.

The result is a two-tier system. Smaller brands get priced out of professional forecasting entirely and rely on Instagram, gut feeling, and whatever their buyers tell them. Larger brands pay for subscriptions — and then realize their competitors are reading the exact same reports. When everyone drinks from the same well, everyone designs the same line.

The fashion trend forecasting market is growing fast — projected to roughly double from its current size to over $15 billion by 2033 — which suggests demand is real. What the numbers don't capture is how many teams are still not getting what they need from what's available.

The Actual Problem: Speed and Specificity

The issue with most trend intelligence isn't the data. It's the distance between the data and a real decision.

Knowing that utility pockets are trending up 34% is not the same as knowing whether that pocket belongs on a $49 workwear pant, whether your factory in Southeast Asia can execute it at spec, or whether your fit model will kill it in review. Generic trend direction — written for every brand and category at once — can't close that distance. It lands on a design team and sits there, interesting but inert.

A team working against a real line plan, real margin targets, and a real calendar needs something different. They need direction that's already been filtered through the lens of what's producible and what will sell.

What ShaTrend Studio Is Built to Do

ShaTrend Studio is an AI-powered trend intelligence tool built by Shanley Suganda, a Senior Creative Director with 25 years across denim, workwear, knitwear, wovens, outerwear, and fleece. It was built specifically because she's had to act on trend intelligence — and watched too many reports fail that test.

ShaTrend translates trend signals into direction a design team can actually use: silhouettes, fabrications, washes, color stories. Framed the way a design room talks about product, not the way a consulting agency writes about it.

The depth comes from domain knowledge, not just data aggregation. There's a meaningful difference between scraping runway photos into a grid and knowing which silhouette will survive fit approval and which will die on the cost sheet. ShaTrend is built to carry that context — to give teams trend direction that holds up when it meets a real line plan.

A Worked Example

Say you're a denim brand building your next men's bottoms assortment. A standard trend report might tell you that relaxed straight fits are up and that utility details are gaining traction across wholesale. Useful enough, but it leaves your team to figure out how that applies to your specific price architecture, your wash program, and your factory's current capabilities.

ShaTrend takes that same signal and frames it against your brand and category — which silhouette variations are moving at which price points, which fabrication and wash directions are worth pursuing versus decorative noise, and what a credible assortment position actually looks like. The output is direction, not inspiration.

Part of a Connected Development Workflow

ShaTrend is the first step in the Shanley.ai suite. Trend direction flows into Sha3D Studio for 3D garment visualization, then into ShaTechPack Studio for factory-ready tech packs. The intent is to compress a development calendar that typically runs weeks into days — from market signal to a spec your factory can quote against.

Research, visualize, spec. In sequence. Without the usual gaps between each phase where context gets lost and decisions get made twice.

Who Should Be Using It

If you're an independent designer or emerging brand, ShaTrend gives you access to the kind of category-specific direction that was previously only available to teams with large forecasting budgets. You can start with a credit pack and use it when you need it.

If you're at a larger brand, the value is different: a way to interrogate trends against your specific customer, your categories, and your assortment — rather than reading the same generic brief your competitors just downloaded.

And if you'd rather have senior-level direction without adding headcount, Shanley.ai also offers done-for-you trend and development services.

ShaTrend Studio is live at shatrend.studio. The fastest way to understand what it does is to run a query against your next season and see what comes back.

www.shatrend.studio

The Bill Arrives Before the First Sample Does

Most independent designers don't lose money in production. They lose it in the weeks before production even starts — paying for sketches, revisions, technical flats, and the freelancer who turns those flats into a factory-ready document.

Freelance technical designers charge between $300 and $1,000 per style on average. For a first collection of 8–12 pieces, that adds up to $1,500–$5,000 before a single physical sample exists. And that's before you factor in what a bad or incomplete tech pack costs downstream: rejected samples, back-and-forth emails with the factory, and extra sampling rounds at $200–$800 each.

The industry average is 2–4 rounds of sampling before production approval. Each round carries its own shipping fees, revision time, and factory charges — costs that can effectively double the per-sample expense. A tech pack that's unclear or incomplete is usually the reason those extra rounds happen at all.

---

What sha3D.studio Actually Does

sha3D.studio handles the work that eats designer time between the creative idea and the factory conversation. It operates in three modes:

Sketch to 3D takes your flat sketch and produces front, three-quarter, and back views of the garment. The practical value here is proportion checking — catching a silhouette problem on a 3D mockup costs nothing; catching it after a sample is made costs another $200–$800 and two to three weeks.

Photo to Technical Sketch runs the process in reverse. You upload a reference photo — from a runway shot, a vintage piece, a competitor's garment — and the tool produces a clean technical flat from it. No redrawing from scratch, no explaining to a freelancer what you meant.

Design Studio covers the variation work that normally generates the most back-and-forth: colorways, pattern repeats, variations spun off an existing design, or building a new piece from a text description. Changing a colorway at this stage is a few minutes of work. Changing it after a sample ships is a whole new sample round.

---

Locking the Design and Moving to ShaTechPack.studio

Once the design is confirmed, ShaTechPack.studio generates the factory-facing document. The seven-page template covers the full set of information a factory actually needs: colorways, bill of materials, design details, artwork placement, and measurements. You preview it, then export directly to PDF.

This matters because factories — particularly overseas manufacturers — cannot quote accurately, grade sizes, or maintain consistency across a production run without a complete spec document. A well-structured tech pack is what separates a smooth first sample from a months-long revision cycle.

Brands that send detailed, complete tech packs to factories tend to see fewer sampling rounds and lower rework costs per design. The savings on avoided rework alone often exceed what the tech pack process cost in the first place.

---

A Worked Example

Say you're an independent designer launching a 10-piece capsule. The traditional path looks like this:

• Sketches done. Now you need technical flats — find a freelancer, wait a week, pay $400–$700 per style.
• Flats done. Now you need a tech pack assembled — same freelancer, or another one, another $300–$600 per style.
• Total pre-production documentation cost: roughly $2,000–$4,000 for 10 styles.
• First samples arrive. Two styles have proportion problems that a 3D preview would have caught in the sketch phase. Two more sampling rounds, $400–$1,600 in extra costs.

With sha3D.studio and ShaTechPack.studio, the sketch-to-3D preview step catches the proportion problems before the factory touches anything. The technical flats come from the tool, not a freelancer. The tech pack is assembled in a structured template with export to PDF built in. The mechanical work — the part that was never creative in the first place — gets removed from the equation.

---

Who This Is Built For

This workflow is specifically built for independent designers and small brands that don't have a full technical team on staff. If you're a solo designer or running a label with a team of two or three people, you're probably doing one of two things: either spending significant money outsourcing technical work, or spending significant time doing it yourself in tools you didn't build your skills around.

Neither is a good use of resources when the actual design work — the creative decisions about silhouette, fabric, and construction — is what drives the product forward.

The documentation is necessary. It just doesn't need to be expensive or slow.

---

What Each App For

sha3D.studio handles 3D mockups, technical sketches, colorways, and design variations.

ShaTechPack.studio handles the seven-page factory-ready tech pack — preview and PDF export included.

Both tools are part of the Shanley AI suite at shanley.ai.

What's the biggest bottleneck in your design-to-sample workflow right now?