• I want to thank all the members that have upgraded your accounts. I truly appreciate your support of the site monetarily. Supporting the site keeps this site up and running as a lot of work daily goes on behind the scenes. Click to Support Signs101 ...

I Built My Own Sign Software Using AI. So Should You.

Bobby H

Arial Sucks.
I've been using CorelDRAW and Adobe Illustrator alongside each other for over 30 years. Both may technically be the same kind of application, but each has a slew of unique features. So I take advantage of the best of what both have to offer.

We have multiple licenses of Flexi at my workplace, but I don't use it on my workstation. Flexi is useful in production, but it's pretty limited in terms of design capabilities. The UI feels like going back to the 1990's.
 
  • Agree
Reactions: 1 user

LizKeenan

New Member
Well thats where you and I differ. I have worked across several softwares, and TO ME the value in the designer is the versatility in being able to use different softwares. For everything Print I would use Illustrator. For sign design, I would use flexi. I understand your point. And it has value, but this whole industry is based around making things work for you. Illustrator is not a hard software to use. If you need to unround a corner you just manipulate anchor points. Illustrator will unround corners that are already told that it HAS been rounded prior from said anchor point. So it depends on the users level of skill IMO. Does that mean it could have better tools? 1000% but I don't see how AI has any bearing on that fact.
THERE ISNT ANYTHING that building an AI based production merchant/softwre does that any other software on the market already does. There are several to choose from. I just don't get the push. Its not helping anything. And until AI perfects the workflow for any particular business, its always going to have some drawbacks in its workflow and how the users interact with said software.
To me it just seems like nerds keep wanting to slowly infiltrate and take over blue/yellow collar industries.. and honestly Im not here for it.
None of what you are saying is wrong, but I still think there are tools to be made. You are right, the examples I have given can be solved with using other tools, Astute has been especially helpful. However there are some that haven't been whipped up yet for super annoying edge cases. For example the below illustrates a script to produce paneling with versaworks/ onyx logic to visualize panels and overlaps while still being able to nest and label. This was created specifically for odd cuts to be nested to waste less material than rip paneling, and since they aren't rectangles you cannot pdf panel them easily, nor would it include the custom cuts and overlap bleed. Plus we use CutMaster so this way we can add the marks we need instead of cutting it on our print and cut Roland with their marks. Is this useful? To us, very. At large? Onyx is still an excellent paneler that I am sure most people are utilizing that just fine.
 

Attachments

  • Screenshot 2026-04-06 200245.png
    Screenshot 2026-04-06 200245.png
    183.7 KB · Views: 23
  • Screenshot 2026-04-06 200159.png
    Screenshot 2026-04-06 200159.png
    248.9 KB · Views: 23
  • Screenshot 2026-04-06 200126.png
    Screenshot 2026-04-06 200126.png
    226 KB · Views: 20
  • Screenshot 2026-04-06 200005.png
    Screenshot 2026-04-06 200005.png
    230.6 KB · Views: 19
  • Screenshot 2026-04-06 195939.png
    Screenshot 2026-04-06 195939.png
    229.6 KB · Views: 24

Pauly

Printrade.com.au
I would love to see the code base of this app.

What stack is it running on?
Why did you choose the stack?
How did you deploy it?

Ive build many apps with AI.
I know what it can do when you have zero clue on how to code, and i know what i can do when you know what you're doing.

Its the same as print.
When you dont know "this looks good"
When you do "these reds are brown, missing nozzles, banding, blacks are green"
 
  • Agree
Reactions: 1 user

Think713

New Member
None of what you are saying is wrong, but I still think there are tools to be made. You are right, the examples I have given can be solved with using other tools, Astute has been especially helpful. However there are some that haven't been whipped up yet for super annoying edge cases. For example the below illustrates a script to produce paneling with versaworks/ onyx logic to visualize panels and overlaps while still being able to nest and label. This was created specifically for odd cuts to be nested to waste less material than rip paneling, and since they aren't rectangles you cannot pdf panel them easily, nor would it include the custom cuts and overlap bleed. Plus we use CutMaster so this way we can add the marks we need instead of cutting it on our print and cut Roland with their marks. Is this useful? To us, very. At large? Onyx is still an excellent paneler that I am sure most people are utilizing that just fine.
Anytime ive had to do overlaps like that I do them manually... I've never felt like I needed a script or ai to do that.
 
  • Agree
Reactions: 1 user

Pauly

Printrade.com.au
To make this more interesting, read in the voice of Morgan Freeman.

I've been in the sign industry for 20 years. For most of those years, I've watched us all complain about our software. Everybody knows the workflow can really SUCK! Everybody does it anyway because there's nothing better. Or they're using software that costs too much and still doesn't do half of what you need.

Screw... that.

I didn't want to waste my life on that kind of work. Why should I? I'm too pretty. I needed a better solution!

So I built one.
Actually, I built several. And they work on real world projects at 4x speed. That's not boasting. That's numbers.

Now here's the part that's going to make some of you skeptical. I built these with AI.

Before you close this thread...
I know. The AI conversation in this industry has been almost entirely about clients showing up with terrible ChatGPT logos and expecting us to print them on a monument sign. I get the frustration. But that's one tiny, annoying corner of what AI actually does.

Here's what nobody in the sign industry is talking about yet. AI can help you BUILD your solutions, custom fit. Not clip art. Not logos. Actual working software that solves actual problems in your shop.

I don't have a programming degree. I'm a sign industry professional. I could not have built these applications on my own. What I do have is domain expertise from years of soul-crushing, monotonous work. I know exactly what I want and exactly where the existing tools fall short. That knowledge is the hard part. That's the part AI can't do right now.

What AI can do is write code. And it's gotten scary good at it.

What it actually looks like
There's a lot of hype. Building software with AI is not "hey ChatGPT, make me an ADA app." That gets you nothing useful.

What it actually looks like is more like this. You sit down and you think hard about the problem. What does the workflow need to be? What are the inputs? What are the outputs? What are the edge cases? You sketch it out. You make decisions about how things should work based on your years of experience doing the actual work.

Then you describe what you need to AI, piece by piece. You test what it gives you. I make web apps so it gives me a webpage to review. It's wrong sometimes. It's wrong a lot, actually. But you catch the mistakes because you know the domain. You know that Braille translation has rules that most programmers have never heard of. You know that sign maps need to reference room numbers a certain way. You know what the production floor needs to see on a shop drawing. You know your problems.

The AI handles the code. You handle everything else. The design, the decisions, the testing, the iteration. It's a tool in your hands, not a magic button.

It takes patience and it takes critical thinking. You have to be willing to look at what AI gives you and say "no, that's wrong, here's why" and push it in the right direction. There are features I rebuilt over and over before they worked right. And there are still some bugs.

But the fact that someone from the sign industry can build production-grade tools like this at all? That's the point.

Five years ago, if I wanted to build a custom application that solves my custom problem, I would have needed to hire a development team (tens to hundreds of thousands of dollars). Developers who don't know the sign industry. I'd spend half the project just explaining what a message schedule is. That barrier is basically gone now. All that cost drops to the price of your AI subscription and your extra time. This is real right now. $100,000 to a $1000? Incredible. The software industry is collapsing... right... into... your lap.

Why this matters for you
Think about that for a second. Every one of you has deep knowledge about some part of this business. Maybe you've been doing vehicle wraps for 30 years and you know exactly how the estimating process should work but no software does it right. Maybe you run a shop and your job tracking workflow is held together with sticky notes and prayers. Maybe you've got an install crew and there's no good way to track field progress.

You know the problem better than any software company ever will. And now you have access to tools that can help you build the solution. Your solution. You don't need a computer science degree. You need to be good at thinking through problems, which... you already are. That's literally the job.

The real AI conversation
Most of the AI conversation has been noise. Bad logos, overpromising, hype. But there's a quieter thing happening underneath all that noise. Regular people in specialized industries are using AI to build tools that solve their own problems. Not billion dollar companies. People like me, building exactly what they need because they finally can.

That's the AI story worth paying attention to. Not the logos.

DO NOT SLEEP ON THIS.

Let me know if you want a tutorial. I'll write a post here to show you how to make something really simple to get your feet wet.

- i am 3toed, struck by lightning, bit by a cobra



I would love to see the code base of this app.

What stack is it running on?
Why did you choose the stack?
How did you deploy it?

Ive build many apps with AI.
I know what it can do when you have zero clue on how to code, and i know what i can do when you know what you're doing.

Its the same as print.
When you dont know "this looks good"
When you do "these reds are brown, missing nozzles, banding, blacks are green"

still waiting for a response.

.
.
.
.
.
.

.
 
  • Agree
Reactions: 1 user

WildWestDesigns

Active Member
Ive build many apps with AI.
I know what it can do when you have zero clue on how to code, and i know what i can do when you know what you're doing.
Are we talking scripts here or are we talking C/C++ apps interfacing with Qt/IMGUI/raygui etc (for a full GUI experience etc) or I would even go TUI here, something like FTXUI?

I have found, outside of Rust (which not a fan of that babysitter of a language), "AI" fails at low low level programming (scripting, because even most devs now are just scripters/web devs only app devs due to the help of Electron and the like(even programs that most on here know use Webview even for their "native" app) and everything that is out there for "AI" to "learn" off of is really based on that aspect of programming, "AI" tends to do better with that). Having that "babysitter" of a compiler (Rust lang) helps with keeping it in check. Especially if one has to go back for many prompts for a particular session (it doesn't do well keeping track of what is what particularly on non trivial applications).

What I'm afraid of, just like with art/animation/coding people/users are going to be even more abstracted away from basic concepts (already happened with the takeoff of Python and Electron/webview, this is just the same thing on steroids).

I would argue that with kids using "AI" younger and younger in school now, they won't have the basics. Just like we have lost the basics with math, reading/writing ("AI" prompting will have to be abstracted away to just voice before one knows it), shoot even critical thinking (already have studies on it and already have places/countries trying to change course on it with going back to more analog methods in teaching). So this "know what you're doing" is going to be a very different beast pretty quick and probably be very lacking in what comes out in future products.
 

Think713

New Member
Are we talking scripts here or are we talking C/C++ apps interfacing with Qt/IMGUI/raygui etc (for a full GUI experience etc) or I would even go TUI here, something like FTXUI?

I have found, outside of Rust (which not a fan of that babysitter of a language), "AI" fails at low low level programming (scripting, because even most devs now are just scripters/web devs only app devs due to the help of Electron and the like(even programs that most on here know use Webview even for their "native" app) and everything that is out there for "AI" to "learn" off of is really based on that aspect of programming, "AI" tends to do better with that). Having that "babysitter" of a compiler (Rust lang) helps with keeping it in check. Especially if one has to go back for many prompts for a particular session (it doesn't do well keeping track of what is what particularly on non trivial applications).

What I'm afraid of, just like with art/animation/coding people/users are going to be even more abstracted away from basic concepts (already happened with the takeoff of Python and Electron/webview, this is just the same thing on steroids).

I would argue that with kids using "AI" younger and younger in school now, they won't have the basics. Just like we have lost the basics with math, reading/writing ("AI" prompting will have to be abstracted away to just voice before one knows it), shoot even critical thinking (already have studies on it and already have places/countries trying to change course on it with going back to more analog methods in teaching). So this "know what you're doing" is going to be a very different beast pretty quick and probably be very lacking in what comes out in future products.
Aside the points mentioned about the coding softwares cause I know nothing about coding, i think your lower points here are where my heads going with all this AI stuff. I'm currently dealing with an operations manager and a boss and a designer all using AI tools, and its actually creating more problems for them. They're having to recreate solutions for things they already have solutions for in the name of "speeding" things up which doesn't really benefit anything.
Its the whole don't use a calculator unless you understand the equation first kinda thing. And its happening at a pace that I think basic intelligence is going to be lost incredibly quickly. Its scary.
 
  • Agree
Reactions: 1 user

WildWestDesigns

Active Member
Aside the points mentioned about the coding softwares cause I know nothing about coding, i think your lower points here are where my heads going with all this AI stuff. I'm currently dealing with an operations manager and a boss and a designer all using AI tools, and its actually creating more problems for them. They're having to recreate solutions for things they already have solutions for in the name of "speeding" things up which doesn't really benefit anything.
Its the whole don't use a calculator unless you understand the equation first kinda thing. And its happening at a pace that I think basic intelligence is going to be lost incredibly quickly. Its scary.
They aren't really totally unrelated. Most people don't really care, because they aren't artists/programmers etc. To them, it's the destination and not the journey (which in most of these endeavours, the journey is just as important if not moreso compared to just the destination).

This is orders of magnitude worse and we are actually seeing it now even with people that should know better, there are issues with this (the "AI Pause" is a real thing (this particular "pause" is about professionals "pausing" whatever they are doing to allow the "AI" to do it's thing, preventing any type of "flow state", to much stop/go). At least with the other concerns with letting tech run amok, it actually took a couple of generations, we aren't even getting through 1 generation with this before starting to see degradation.
 
Top