I’m not sure you’re fully aware of the state of shop tech these days. Right now, you can build your own shop software. The real deal, without hiring a developer or waiting months for the industry platforms to finally ship the feature you actually need.
This isn’t a prediction. This is you building something and launching it in a day or two.
The beauty of AI is that you describe what you want in plain words, give it a few examples, and watch it get built. If I ran an industry software platform, I’d be nervous right now. Especially the ones that have ignored their customers for years. You know who I’m talking about.
So what do you need? An automatic art approval form builder? A tool that reads and organizes your email? Something that preps shipping info, prints box labels, and builds a packing list? A custom quote drafter?
Pick the manual, repetitive task you hate most and build a tool that does it for you. Right now. Today.
I talk with plenty of shops that have already replaced their legacy software, or built custom tools that sync through APIs to the platforms they’re comfortable with. You don’t have to start over. You can build custom tools that make the ones you already use work better.
If you’re already using AI, start with what you’ve got. Claude, ChatGPT, or whatever you use. You can build this in any of them. If you want the fastest path to a working tool with the fewest steps, a vibe-coding platform like Base44, Lovable, or Replit is built for exactly this, and it hands you a live app you can open on a screen.
The fun part of building a custom app is that you don’t have to know how to build it. You just have to ask. AI will recommend where to start, what you need, and what to do next. If you get stuck, say, “I don’t understand. Explain it to me in simple terms. Show me what to do next.”
Let me show you one from start to finish. We’re going to turn a paper production log into an app your crew can use on the floor. Follow along, and you can do this for your shop.
The full set of downloadable instructions is available for free at the bottom of this article.
Connecting the Dots with a Production Log
There’s a reason I want to build this app from this form. I designed it in Excel 25+ years ago to track production KPIs. Why? If you’ve read anything I’ve written, you know accurate data on what’s actually happening in your business is the best way to make important decisions. Data on speed, capacity, and problems lets you improve, reward your people at performance reviews, and master production scheduling to increase throughput and on-time delivery.
Production logs like this one make the data visible. The problem is that they’re filled out by hand. People forget, the handwriting is a mystery, or the sheet gets lost. When you’re mining for your own information, the simpler you make the process, the better the habit sticks.
Which is why using AI to build a better tool is a brilliant idea. We can replace the paper form and put that valuable information one click away.
Why Start With a Log Instead of a Bigger Idea?
Start with a log because you already understand it. The hardest part of bringing AI into your business isn’t the technology. It’s the fog of what to do next.
In my research for my book “2026 Thinking,” shop owners kept describing AI the same way. Promising, powerful, and inevitable. Also confusing, intimidating, and difficult to integrate. One owner put it plainly: “Everyone is talking about AI, but it’s still not clear what it actually looks like inside a shop like ours.”
We could have built anything for this article. A production log is a great place to start, because it isn’t a blank page or a loose idea. You’re starting from a tangible asset you might be using right now. If not, use mine. At the end of the article, I have everything you need to start in a download.
3 Things You Are Already Thinking
Let me name 3 things you’re already thinking. By the way, you’re not wrong to think them.
It will break. Sure, that’s possible. But you built it by talking to the AI in plain language, so you fix it the same way. Tell it what’s wrong, upload a screenshot or two, and it adjusts. You’re not editing code. You’re giving feedback. AI does the rest.
My crew won’t use it. That’s a real risk. Are there other SOPs your team isn’t following either? Let’s lean into that. Finding these problems is exactly why you want a production log. What else are they skipping?
I’ll burn a weekend and end up with nothing. You won’t. We’re going to build small and prove it works before you spend any real time on it.
Why Does Measuring Production KPIs Even Matter?
The truth lives in the data, and it can change how you run your shop, not just how you feel about it. You might not be tracking this stuff right now. Not because you don’t know it’s valuable, but because getting the data by hand has always been a pain. And it’s one more thing for your team to complain about. I get it.
Here’s what you unlock when you capture the data and graph it:
Scheduling. When you know your real setup and run times, you schedule the floor based on what actually happens instead of guessing.
Cost optimization. When you can see downtime by reason, you can see what’s draining money every day.
Employee performance. When you can see output by operator, by shift, by press, you stop guessing about who needs help and who your superstars are.
Finding the leak. When the numbers are easy to sort and connect, patterns jump out. That’s where you find the next thing worth fixing.
Getting the information is step one. Graphing it is step two. Using it to make decisions is the real win.
A production log shows you the real number. You still have to act on it. This app won’t fix your set-up times or your production. That’s your job, once you can finally see what’s going on.
What you’re building is a continuous improvement loop for your shop. The app is just the part you can see. The loop is simple, and it runs forever.
Measure. Find the constraint. Fix it. Measure again.
Low-friction capture makes it easy. The dashboard shows you the biggest leak. Go fix it. Then measure again to see if it worked, and the next-biggest leak floats to the surface.
4 Decisions to Make Before the Build
Before you touch a keyboard, answer these 4 questions. Your answers become the instructions you hand to AI, and they make the build much easier.
What Are You Starting From?
Decide what you want to turn into an app. My production log, your production log, or a blank one from scratch. My advice? Pick the one that matches how your shop already works. The closer it is to what your team does today, the less they have to learn.
What KPIs Matter to Your Shop?
Track the one number that points to your current constraint, not every number you could possibly capture. A common mistake is tracking a dozen things and trying to improve them all. You improve none of them and drown in data you never act on.
Data points are good. For shop improvement, work on one thing at a time. Pick the one that bugs you the most, chase it, and improve it.
Look at my production log for screen printing. It covers set-up times, impressions, misprints, screens used, and downtime. For other decoration methods, similar metrics apply. Whether it is a paper production log or an app, your team will still be adding the information as the job progresses. It works the same way: your crew adds the information as each step is completed, not after the whole job is done.
The whole idea behind this kind of app is that you can capture the data easily, get it flowing, and start connecting the dots on what’s really happening without much friction or labor.
Where Will the Data Live?
Your data needs a home. You’re collecting it, so where does it live? This choice shapes everything downstream. There’s no universally correct answer, just the one that makes sense for you.
For some, a simple spreadsheet in the cloud is enough, with a chart tab for the dashboard. Low friction, nothing new to learn.
You could also connect to Airtable, or use Supabase or the native storage in Base44 if that’s what you build with. You don’t have to know how to set this up. You describe what you want to store and how you’ll use it, and AI builds the structure.
Consider the cost. For some, a spreadsheet is good enough because they already pay for it. You can go more robust if you need to, and it won’t cost more than a modest monthly subscription.
What Should My Dashboard Show?
Before you build, know what answers you want. Knowing how you want to visualize the information helps you filter what’s worth tracking in the first place, so you can make solid decisions.
Some KPIs are just an average in a box, like 445 impressions per hour for print speed, or 5.3 minutes per screen for set-up. Others are better as a pie chart or a line graph, to show a percentage of the whole or results that move over time.
How the Build Actually Works
Here’s where it gets easy. You don’t have to figure out the right way to ask the AI for any of this. I built a file that does the asking for you. It’s a set of plain instructions you download (link at the end of the article), and it turns the AI into a patient guide that walks you through the whole build, one small question at a time.
You upload that one file into your AI tool of choice. That’s the only technical thing you do. From there, it takes over and interviews you like a good consultant would. Here is what that conversation looks like:
It starts with the win, not the tech. The first thing it asks is what you want to be able to see or know that you can’t today. Everything it builds serves that answer, so you never lose the plot in a pile of features.
It reads your form for you. If you have a paper log or a spreadsheet, you upload a photo of it, and the AI reads it back to you in plain words, then asks what’s missing and what you don’t need. If you don’t have one, it just asks simple questions instead. Either way, you can start from my production log in the download.
It asks one thing at a time, and it recommends an answer. How should the timer work? Who sees the dashboard? How do people log in? For every question, it tells you what it suggests for a shop like yours and why, so you can just say “yes, do that” or steer it your way. You are never staring at a blank screen asking yourself what the right answer is.
It says the whole thing back before it builds. When the questions are done, it summarizes everything in plain language and ties it back to the win you named at the start. You confirm, and only then does it build the first version: the entry screen, the roles, the logins, the dashboard with your graphs, all of it.
That’s it. No prompt engineering, no code, no guessing at the magic words. The file carries the expertise so you don’t have to.
What to Expect When You Run It
A few honest things, so nothing catches you off guard.
Give it about 30 to 45 minutes for a first working version. I tested this same file across three different AI tools, and each one got to a real, clickable app in roughly half an hour. Not a mockup. Something you can tap through with fake data and actually try.
It won’t be perfect on the first pass, and it isn’t supposed to be. This is your prototype. Once it’s built, you change it by talking to it: “make the buttons bigger,” “move that to the dashboard,” “use full words instead of codes.” You’ll make dozens of little edits, and that’s the process working, not failing. Maybe even upload your shop logo and change to your branding colors or fonts.
Where your first version lives depends on the tool you pick, and this is the one spot where the platform really matters. Some tools, the vibe-coding platforms like Base44, Lovable, and Replit, save your data and hand you a live web link almost right away, so you can open it on a tablet on the floor. Others will build you a working app to test on your computer, but getting it onto a shared link and saving data over time is an extra step. Neither is wrong. If your goal is a link you can put in front of an operator this week, start with one of the vibe-coding platforms and you’ll get there with the fewest steps.
When it finishes, ask it to save you a plain-language summary of what it built. That becomes your record of how the app is set up, and you keep it right alongside the download file. Two files, both yours, one that builds the app and one that describes it.
Your Team’s Feedback
Eventually the app is ready for real testing. This is where the rubber meets the road. You want this app on a shop-controlled tablet or computer at each workstation. Not on personal phones. You know your shop, and personal phone use on the production floor is always a distraction. Asking staff to put this on their own phone is a non-starter for that reason.
The data gets entered right where the work happens, as the work is being done. By limiting what the user types and giving them selections, you make collection far more frictionless than a paper log sheet.
Start with one team or one employee and get their feedback. Make changes based on what they say and any problems that surface. The goal is clean data that shows exactly what’s happening over time.
Only 3 Things Will Happen With Your Data
Only 3 things will happen with your data, and 2 of them are bad. The data shows you’re getting better (the good), getting worse, or staying the same (both bad).
As you start using the app, build the habit of when your production management team reviews the data. Pay attention to what matters to you. Because you have the data, an emerging trend will sometimes show up in a graph you didn’t expect. That’s why this work is valuable.
Constant review is important. Get in the rhythm of checking. At a minimum, pick a day and time every week to review. Then ask, “What’s the biggest leak right now, and what should we do about it?” The solution to a problem often lives upstream from the actual issue.
Once your whole production line is on the system and you’re confident the data is real, throw it up on a big monitor out in production. I’ve been to shops that rotate the view through different graphs and show performance like a leaderboard. It’s motivating to sit in the number one position. People play harder when they know the score.
Fire Bullets, Then Cannonballs
To recap, build for one work unit first. Prove it works. Refine and tweak. Then expand.
The “Fire Bullets, Then Cannonballs” idea comes from authors Jim Collins and Morten Hansen. A bullet is small, cheap, and low-risk. You fire it to find your aim, not to bet the house. That’s why you start by building and testing your app with one press, one shift, one operator. That’s the bullet. Low cost, low risk, and it tells you whether the idea survives contact with your shop floor.
A cannonball is the big commitment. You only fire it after the bullet has hit and told you where to aim. Rolling this out shop-wide is the cannonball. Once you know your app works and doesn’t need further tweaks, you fire it. The whole floor uses the app on tablets at their workstations, and you capture all of the data.
Build small. Watch closely. Then scale what you’ve proven.
Expanding With Phase Two
Once the app is running and you’re getting good data, you can expand it to make it even easier to use.
At the start, the user types the work order number, selects print locations, keys in quantities, and picks options with buttons. In phase two, you connect your app with an API to your shop system and pull in the order information. Picture it: the operator enters the work order number, and the screen populates with the real information, so they don’t have to key it in or select it.
And because you’re using an API, you could push live data into the Notes from Production section or other parts of the order in your shop system for live recordkeeping. I’ve seen shops connect this to a customer-facing app, where every step of the order is tracked live. The shirts are bought, the art is created, inventory is received, screens are burned, the job is staged, registered on press, printing, then shipping. All of it graphed in real time. It’s like the Domino’s Pizza Tracker, but for your shop.
That takes the idea and makes it 100% about the customer. That’s a cannonball worth firing.
Competent Is Not Enough Anymore
One of the sharpest comments I heard during my “2026 Thinking” interviews was this: “You may have gotten by in the past because you were competent, but being competent today does not cut it, because AI is competent, and if you are merely competent you are vulnerable.”
Let that sit for a second.
Maybe your shop is competent. You turn out decent work. Things seem stable and generally ok.
But out there, others are running scratch-built tools made for exactly how they work, aimed at the goals that matter to them, and they use those tools to constantly up their game. They’re building better insight into their data to uncover opportunities and fix what’s holding them back. They aren’t limping along guessing anymore.
The ball is in your court. You don’t have to wait for the legacy software companies to build the tools or the ideas that make things better. With AI, you build them yourself.
Every day you wait to learn this is another day your competition adds tools like this to their lineup. This is a skill you can learn. You don’t need permission or a big budget. You need the willingness to be awkward, try something new, and build version one. That’s how you find clarity in the things that matter.
What are you waiting for?
______________________________________________________________________
Want the file that builds it with you? It’s a free download, a set of plain instructions you upload into your AI tool. It interviews you, then builds your first version. Start from my production log or your own. Download it below.
And if you want a guide through the process, plus help with the other things in your shop, that’s what I do. Let’s talk.