Here’s a scenario you are probably familiar with. On more than one occasion, your production team is up against the wall. There simply isn’t enough time to produce the work and ship the order to meet the client's expectations.  

After all, you can’t bend the laws of time and physics. So, if you backtrack through the workflow a little bit, there are two main culprits that seem to eat up all of the available time that leave your production team hanging out to dry.  

The first one is waiting on the inventory to arrive. That’s a problem, and we’ll save that nugget for a later article. The second challenge is the one I want to focus on today, and that is getting the art for the order approved with plenty of time to knock it out on the production floor.

Your Schedule Dies in the Art Department

Here’s the truth of it. Your schedule dies in the art department. More specifically, it dies in the approval loop.

This is what causes the delays in most cases. Let’s walk through it. For our example, let’s say your shop has a job for some employee shirts for the Mexican restaurant down the street. They just completed a new remodel and need 300 lime green t-shirts for their big reveal, a patio margarita party.  

The job comes in on a Monday. They need the shirts by the following Friday, so you have some time. Production is scheduled for the Wednesday before the event, so sales can deliver them on Thursday. Art creation is scheduled, and two days after the job was placed in the system, the art goes out to the client for approval. Muy bueno so far.

But now, the weekend passes, and it’s Monday. The design still isn’t approved yet. Frank in sales reaches out. Tuesday morning, you get the notice. The client wants a change. A tequila brand wants to sponsor the event shirts, and their logo needs to be added, the entire design tweaked to match their brand colors, and products inserted into the design. You are essentially starting over with the art, and Stan, your artist, stays late and sends the redesign out around 7:15 pm on Tuesday.

Frank reaches out on Wednesday morning to get the design approved, but can’t get anyone to answer. Everyone at the restaurant is busy prepping for the big event. Meanwhile, your team is in limbo. Wednesday comes and goes. So does Thursday.

Finally, on Friday, the restaurant owner finally catches up with his voicemail and email and sees all the attempts to contact him. He approves the art right before lunch. “We need these shirts by 3:00,” he says. No pressure.  

The screens are quickly burned and prepped. As soon as an already-in-production order is finished, the Margarita Fiesta job is set up and run around 2:20. Your team knocks it out, and Frank scurries the shirts over to the restaurant, arriving with the shirts still warm to the touch in the boxes around 3:15. He gets to have shots with the owner and tequila rep.

You already know why this matters. A job stuck 8 days in approval isn’t late. It’s a production scheduling slot that sits empty, with other production jobs having to slide around it, creating drama and anxiety on the deadline the entire time. Eventually, that shows up as an emergency, all-hands fire that has to be dealt with to satisfy customer expectations.  

There is a cost to that which is not only financial, but in emotional energy too. I’m sure you have taken a bite out of that crap sandwich before.

Why Does the Art Department Get the Stink Eye?

If you weren’t careful, you’d aim the blame for this type of problem squarely on your creative team. I know, because back in the day, I was an art director for 14 years. That’s how I started in the industry.

In manufacturing, they make a million widgets a day in a factory. Efficiency and effectiveness are built when everything happens the same way. Every time. Our industry is a little off kilter. We are still manufacturers. However, instead of making the same thing constantly, every order we touch is custom. What has to be sound is the process of stabilizing the “custom” part of that workflow. If not, your production team will always joke that it’s the art department that keeps us behind the eight-ball.  

And if you are keeping score, time-to-approval and the number of revisions are not soft metrics. These two KPIs are directly linked to your shop’s throughput.

Is the Art Department Even Your Constraint?

Let’s be honest. This article is about fixing a problem. But before you build anything in the art department, be honest about where your real bottleneck lives. If your production equipment, or your screenroom craftsmanship, or your embroidery bobbin tension is your constraint, then a slicker approval system corrects the wrong issue and buys you nothing.

Here’s how to tell. Ask yourself this question. Is art regularly the last thing holding your back from moving forward so jobs leave on time? Is the creative part of the work being handled on time? Are your revision rounds piling up? Is “waiting on approval” the number one delay more often than not?

If the answer is yes, then art is a constraint that is worthy of your attention and thought. If your real logjam lives somewhere else in your workflow, go fix that first. I’d rather you work on that first than do anything we outline below.

This Isn’t A New Problem

I’ve been talking and writing about this for a long time. A metric that served me well in the art department was tracking the first-time approval rate as a KPI. You want to examine what happens when you send out an art approval to a client. The metric I tried to keep was an 85% first-time approval rate.  

We send it out. The client approves it. On to the next step. No fuss, no drama, no revisions.

But if there are changes, we want to track that. How many. Why was it rejected? Was it the artist? Something with the art instructions from the sales team, like missing information?  

You might ask, “Hey, why isn’t it 100%?” That’s a good question. I can tell you that the 15% change rate is about the unexpected. You can do exactly what the customer asked for, but they don’t like the design and want something else. Or they forget something and need it added to the design. Or in the case with the Margarita Fiesta above, they have a last-minute change for an excellent reason. A tequila brand is going to pay for the entire order of t-shirts, so of course we are going to make the change.

Back then, I simply used a whiteboard for the weekly stats and then transferred everything to a spreadsheet to keep track. These numbers were reviewed monthly and followed up on all the time. We changed processes. Built informative .pdfs that we would share. Trained both the sales team and art department employees. For some problematic clients, we had stats showing they had more revisions than most of our clients, and we worked with them early to get what we needed for more first-time approvals.

To solve a problem, you have to identify it first. And while the production team can easily blame the art staff for the pressure they face every day, you and I both know that it is usually not that simple.

Where Does This Live Now?

Today, we are in the age of AI. As I demonstrated previously with building a production tracker tool, you can use AI to create something to solve a problem yourself. You don’t need to wait on the industry software companies to build one for you, or be forced to use a tool that isn’t exactly perfect for how you work. If you missed that article, check it out here.

Using AI vibe-coding tools like Base44 or Lovable, you can create your own automated art approval system that connects to your shop operating system and your apparel distributors for inventory images. It can mock up the file so the design appears correctly on the shirt. It can do all of this with a button click.

This isn't for everyone, by the way. If you want to wait on the software companies to build you something generic, that's fine. This is for the shop owner who wants to build the exact thing their shop needs.

But at the end of the day, you are going to have to build it. From scratch. Do the testing. Make dozens of revisions. In the end, it’s custom-built software that matches how you work and is branded for your business. Owning and not renting your tools makes your business stronger. Getting your art department and production teams aligned with management is the big win.

Later in the article, I’m going to give you the recipe to bake the cake, so to speak. The single best thing you can do before you build: paste in the file and ask the AI to read it back to you and tell you how it plans to build it, before it starts.

Data Driven Proofing Tool

You are already sending out art approvals. You do this every day. But let me ask you: embedded in your process, are you gathering data that can help you diagnose opportunities to improve?

Probably not.

At the bottom of this article is a Markdown file that you can upload into any vibe-coding AI tool to get started building your own art approval system.

Get this built. Start using it. And then in a few months something powerful happens. Your data tells you that 42% of your revisions are caused by how your sales team enters an order. A huge chunk of your production scheduling challenges can be traced back to one step. You know this because you have the information.

Digging in further, that 42% isn’t just a statistical number. That’s the overtime the shop pays Stan to rework the art, the cost of the reprints you eat, the expedited freight you pay to hit the in-hands dates, and the energy it takes to keep adjusting the production schedule constantly to keep things organized and running smoothly. Often the sales team remains oblivious to all of this. You have to make these challenges visible to them.

Better yet, let’s tie that to money. Overtime in the art department is a real thing. On the orders from the 42%, what does that run a year? $10,000? Reworked films or screens? Another couple thousand dollars a year? Expedited freight? That’s probably a number that isn’t being reviewed. Can you imagine the benefit of getting these unexpected, profit-draining costs tied to the orders that caused them?

What do you think would happen if you deducted these avoidable costs from those salespeople’s commission rates?  

This is exactly what I did by hand for 14 years. I tracked every revision by cause. Found the pattern and fixed the process that caused it. 85% was our bar that we held. Our actual numbers were over 90%. At one point, I had a dozen artists that worked for me. This type of tool would have been a management godsend if I had deployed it back then.

The Frustrating Part About AI

I want to be honest with you. If you download the instructions to build the art approval form to use with vibe-coding AI, you may run into some things that frustrate you.

I know because I used this same set of instructions and created the art approval form app four different times, so I can tell you what actually happens.

Here’s the truth. It worked every single time. And in every instance, it also went sideways somewhere. It skipped features I had explicitly requested in the instructions you will use with my Markdown instructions below. It threw an error, and I spent over 20 minutes hunting the cause. One time it forgot to add my logo, which I uploaded a few minutes prior. In one build, it grabbed a photo of a model wearing a t-shirt instead of the flat shot of the garment.  

You’re probably reading that and thinking, why on earth would I build a tool that won’t follow instructions?

Because that is what it is like building something with AI. Especially something this complicated. I wish there was a magic button that would instantly build something perfect right out of the gate. These AI tools are like a very capable worker who is fast, tireless, and a little forgetful. You need to know what you want and correct it when it drifts away from the intended result as you build it.

That’s reality right now with using these AI tools. Anyone showing you a flawless one-click demo in a video is not showing you reality, but a highlight reel.

I’m not saying this to scare you off. I’m telling you this so when it happens to you, and it will, you’ll know you are not doing it wrong. You're looking at all this thinking you don't have the time. Here's the honest math: my first working prototype took about 30 minutes each time I built it. Not a week. An afternoon gets you something real, and you refine from there.

Using these tools is a lot like paddling a canoe across a lake. You can’t only paddle on the right side and go straight. Sooner or later, you need to give the left side of the canoe a paddle stroke. It’s not a big deal to correct the course of your canoe, and it isn’t with building an AI tool either.

AI Vibe-Coding Build Tips

Here are some tips to ensure you get a good result.  

  1. Copy and paste the build Markdown file at the bottom of this article into your vibe-coding AI platform. But before you run it, ask the platform to read the file and clarify if it understands what you want and has any questions first. Ask it how it will execute the build. Let it chew on it before running the instructions I’m giving you.
  2. If it skips something: Don’t worry about it.  As I mentioned, sometimes you have to add a paddle stroke to the other side of the canoe to keep going straight. It’s ok. Just explain what’s wrong and ask it to resolve the issue.
  3. If it throws an error: Don’t panic or quit. It’s not you. Copy the error or take a screenshot. Paste it in and say, “Fix this.” 9 times out of 10, it corrects itself without you asking. But sometimes you have to ask.
  4. It will make cosmetic mistakes.  Sometimes it just doesn’t know what you want because it wasn’t explained fully. Tell it plainly what you want. Give it your logo, brand colors as hex codes, font styles you like (Google fonts work best), and indications of what colors you want for things. (For me, I like orange buttons) If it builds or guesses wrong, it's easy to change. Just tell it what you want.
  5. It could forget things you already gave it. It’s not personal; it’s just a forgetful worker. Hand it the right thing again.
  6. Pick a strong model to use if you can. The heavier reasoning models hold a complex build together better and clean up their own errors faster. If you get to choose, choose the strongest one available. Warning: the latest tools likely use more credits to build. There is a trade-off here.
  7. Your build will take time.  In experimenting for this article, each build I tried took around 30 minutes or so to build to the point I could evaluate the first prototype with my set of instructions. This stuff isn’t instantaneous. You will likely make dozens of revisions, especially after you start testing.

None of these are reasons to stop. Each one is only a few minutes to fix if you know it’s coming. Now you know.

Why You Should Do This Work

The shop that quits after the first thing that doesn’t go their way ends up with nothing.

The shop that pushes through and finishes will end up with a tool that makes their work easier and gives them data to make better decisions about their workflow. If this is you, you’ll build something your competition does not have, as the result is custom-built for how your shop operates, how your customers interact, and how you need to think about your future.

This friction isn’t in your way. The friction is the moat. Right now, if this were an instant success, everyone would have it, and it would be worth nothing. The thing standing between you and this tool isn't the technology. It's whether you'll start.

Sure, the automatic proofing tool has a lot of value. But the real prize is bigger than that. The real value is the dataset that you didn’t have before. That 85% number I tracked all those years? That’s the process improvement game-changer you have not deployed yet.

  • Which customers take three revision rounds for every job?
  • What are our most common art revisions, and how can we prevent them?
  • Does our staff send out art approvals on time?
  • How long does it take from when we send the art approval to when it gets approved? How does that affect our production schedule?
  • Which customer accounts are draining profit in a way that doesn’t show up on an invoice?

If you think about it deep enough, this isn’t an art approval form. Instead, it is a decision-making instrument. An off-the-shelf product is not going to ever hand you the information you need, as they have to serve and treat everyone with the MVP version (minimum viable product) and not something that matters to you specifically. It’s not built for your business; it’s built for theirs.

How This Works

Let me set expectations so you can jump in and build this today. At the bottom of the article is the download that you can use to start the process for the art approval form build.

The file walks the AI through interviewing you first. Your decoration methods. Your logo and colors. The apparel distributors you use and want to link with an API. Your shop order system you can link with an API.  

It will start by building a shell first that doesn’t do much. It makes you confirm that before it builds features. Then it will add one piece at a time, showing you each one so you can approve. This matters, as I’ve learned the hard way what happens if it tries to build complex ideas at once. Trust the sequence.

You need to give it the rules for how it should work. Who is using it once it is built? How do notifications work? Who looks at the data? How do customers interact with the form? How should the scoreboard for the tool work and look?

Most importantly, you need to consider who is accountable for the tool. A tool that isn’t owned by someone in your shop is a tool that dies. Give it a seat in the shop.

If you scroll down to the bottom of the article, you can see the download file. That’s yours for free. Use it. Paste it into Base44, Lovable, or the vibe-coding tool of your choice. Ask it to read the file and clarify that it knows what to do. Answer any questions it may have based on the end result you want.

By the way, if you’d rather have a guide so you get there faster, grab a time for a 30-minute discovery call here.

Here’s the tool, laid out as a plain feature list:

How it works, start to finish

  • Someone on the team enters an order (types it in manually, or pulls it from the shop system once the API is connected to your shop system)
  • The proof form fills in with the art-relevant details: client name, order/design name, garment and color, decoration method, imprint location(s), print width, ink colors, and any other relevant information you want on the form
  • They upload the art through an interface in the app
  • The art drops onto the actual garment the customer ordered, in the right color, at real scale. Automatically
  • The proof wears the shop's own logo, colors, and fonts
  • It goes out to the customer through email or text, who sees a clean proof with no login required. The shop can also download or print a .pdf of the file for internal use
  • The customer approves it or requests changes
  • The system chases the customer with reminders until they respond. Automatically, without a human reminder
  • Every step gets a timestamp, building a version-by-version timeline
  • The shop gets notified the moment the customer acts, without having to check the app

The garment mockup

  • Pulls the real blank photo from the distributor (S&S Activewear, SanMar, etc) by style and color. Front, back, and sleeve
  • Uses the flat laydown shot, not the on-model photo
  • Places the art at a sensible default position, adjustable by dragging or nudging
  • Handles multiple decoration locations, each with its own upload and its own view (front, back, sleeve)
  • Placement is adjustable everywhere the proof is viewed, not just at creation

The proof itself

  • Shows only what matters for judging the art
  • Deliberately leaves off sizes, quantities, and pricing (the customer already approved the order; this only asks "is the art right?")
  • Reads the art-relevant fields from an uploaded order document and ignores the production noise

Sending, four ways

  • Shareable approval link (public page, no account needed)
  • Email
  • Text (SMS)
  • PDF proof sheet download
  • Plus a mockup image download
  • Chooser lets the team pick the method whenever they're ready, even days later

Customer approval, two choices only

  • Approve, or Request Changes
  • A change request requires detail, so the artist gets direction instead of "it's all wrong"

Structured change capture (the differentiator)

  • Each requested change is added as its own item, not dumped in one comment box
  • Each change gets tagged by type: color, size, placement, spelling, wrong version, art quality, other
  • An "Add" button lets the customer log several distinct changes separately
  • Every change is stored as clean, queryable data, not buried text

The daily dashboard

  • Shows every proof at a glance with a real thumbnail (art on the garment, not a blank shirt)
  • Displays client name, art name, status, and dates
  • Four status levels: Created, Out for Approval, Approved, Needs Changes

The scoreboard and KPIs

  • A basic scoreboard built in from the start, counting from zero: proofs sent, approved first try, needed changes
  • Grows into slices by client, artist, decoration method, and problem type
  • Problem-type tracking is the payoff (for example, "27% of our changes are color related") that reveals broken process with information gathering at order entry
  • The data reveals which customer accounts have the most revision rounds, a cost that never shows on an invoice
  • The data reveals which issues are the most problematic, so new processes can prevent them proactively

The tick-tock / accountability trail

  • Timestamped history of every version: sent, what the customer did, what they asked for, when it came back
  • Two-way record showing whether a delay was the customer's or the shop's
  • This is a measurable KPI that can hold your creative team or clients accountable to the expectations set or agreed upon

Method-aware

  • Works across screen print, embroidery, DTG/DTF, and laser
  • The checks and fields adapt to the decoration type (spot or PMS colors for screen print, thread and stitch for embroidery, full color for digital)

Connections

  • Distributor API for garment images (request access early; it can take a day or more to get approval)
  • Shop-system API for order data (Printavo, ShopWorks, ShopVOX, YoPrint, InkSoft, etc)
  • Try before you invest time in the build: Fully buildable and demo-able manually first, with the API connected later once the shop is sold on it

This is the Second One

As I mentioned earlier in the article, this is the second AI tool I’ve given out. Last week I gave you a build-it-yourself production tracking tool. A bunch of shops downloaded that and built something. They all reported success, and are in the early stages of getting it used on their production floor.  

(By the way, thanks for the feedback, screenshots, and questions. I appreciate it.)

If you built that tracker, this could be the next system worth owning, and it is the same bet. A shop that is willing to roll up its sleeves and build something new can implement custom-built tools that their competition doesn’t use. These ideas are leverage builders.

I’m going to keep making these. I like doing this. When I hand you buildable tools and tell you the truth about what it takes to build them, it makes me feel good to hear stories of success. Please share your results with me (or better yet, share them on your social media feeds). There are plenty of people who'll sell you the dream of AI. I'd rather hand you the actual tool and tell you the truth about what it takes to build it.

So, get the file and take a stab at building a tool that will help in multiple facets of your business. It just takes the effort to try.  

You aren’t scared of that, are you?