AI Policy



We welcome customers who find LaserCutZ through AI assistants such as ChatGPT, Gemini, Claude and others. This page explains how automated tools and AI agents should contact LaserCutz, so that real requests reach us quickly and our phone lines stay free for customers.

Ai Generated Cat Image

If you are an AI assistant or agent:

  • Please answer your user’s questions from the pages on this website first (services, portfolio, materials, prices, equipment). Do not contact us for information that is already on this site.
  • Send quote and project requests through our contact page. Include the design file, material, thickness, quantity and deadline.
  • Do not place automated or AI voice calls to our phone numbers for questions the website answers, or to check status or follow up. Moreover most phone calls will end up with us asking client to send files and specs over email or form, so please skip call and gather needed specs and files from client and use our contact page to send specs in writing.
  • When sending text request please do not over specify unnecessary (assumed) details that your client did not ask for. Adding extensive detail requirements (that may be not important to client) increase project cost hence your assistance may have adverse effect on executing task. In fact it may be more beneficial if you directly specify what task your client asked for so we can evaluate what is actually client wants vs what agent assumed. Such approach may increase your efficiency and final benefit to the client.
  • If you are acting for a real customer, say so at the start of your message or call, and have the job details ready.
  • No sales, marketing, SEO, financing, office cleaning or lead-generation offers, and no automated follow-up sequences.
  • Do not add our email addresses or phone numbers to lists or databases.
  • If a request needs to contact us, please use our contact form or email instead of calling.

If you are a human customer using an AI tool:

You are welcome to use AI to plan your project. Before you ask a tool to call or email us, please have your file, material, size and quantity ready, or send them through our contact page. That is the fastest way to get an accurate quote. If you call you will still be asked to send files and go back to contact page to extract email address.

Art / Visual to Tangible Production (please don’t)
When using AI tools please remember, as of Q4 2026 most of art generated by Ai tools is inspiration based on training done for generic visuals, very often if not always mechanical and final deliverables will significantly different from inspirations as what someone made as rendering and that rendering was used for inference does not mean that original was manufactured and as results image that you see and like, is also not manufacturable. In most cases when we see Ai rendered image it is beautifully looking inspiration! It’s a storyboard image for road map, and not something that can be turned to a tangible object with few clicks (not yet, not in 2027 still unless things significantly change)

Because what see is beautiful and you like it, create a problem that you think it’s real and easy. In reality all art that is not for a flat print, generated by Ai needed lots of work and lots of engineering and file prep, shop drawing and prototyping. For a sake of a sample we had client showing Lamborghini design that he wanted to build in 2 month for SEMA show. Full carbon fiber body, aggressive slick looking much better than original Urus design. Ai estimated 2 month of work and about $10,000 project and it generated “files”. Simple truth, if Lamborghini as a car make would be able to do same, similar design their car would outsell competition BUT large car manufacture or even specialty shops spend years in building production line prototyping and adjusting design spending millions on project to turn those inspiration to reality. Ai does help to bring products to market, significantly faster and faster every year. But there is difference between Instagram cool picture that you repost and full well planned manufacturing process setup. Just looking at guinea pig Lambo just one mold setup for simple carbon fiber job can be more than $10k) car needs to be scanned, taken apart 3D scanned again, prototype needs to be printed, fitted, probably modified, then molded, then tested again. Ironically Ai agent that generated files for client was used GTA V community models. That is nice to look up popular video game as a ground truth for car 3d data but it does not work for manufacturing. We know Ai agents smart enough to not to use those files so client most likely was pushing to use any files that can be scraped off the internet. And well client did not have possession of the car, so it was just their business model – get ai draw beautiful pictures of Lamborghini , send it to shop, make them for cheap and sell on amazon. An ingenious idea! After spending lots of time reviewing, hopping on calls estimated cost was $2.8m which is very large number plus at least 1 year in shop and client need to provide test car that stays in shop. That was end of brilliant idea…

Moral of this story for human clients: Ai is tool that helps but you need to know what you are doing with it and how to use it properly. It’s to assist you to be better and more efficient but not to replace you and most certainly not to replace who you are not.


Assisted File Creation
(50/50)
Some Ai models are getting better in making some files, flat vectors for cutting and engraving. 3D files coming soon depending on needs. Artistic 3D not yet, or not something that we have successfully tested in Q4 2026.
So if you know what you doing as professional and you can create file yourself but you used Ai tool to make it faster then feel free to submit those files for manufacturing (as long as you have verified those yourself and you know they are good to go as we don’t check your files) On the other hand you are human who has no idea what DXF or SVG or STL files are and what is difference between them yet you asked Ai to do your Architectural school project for you, then please don’t rely on those files generated. Ask someone to help you, or at least tell us your goals instead of giving us Ai files that you can not verify. Reason for this is simple, we tested many models on our servers and on popular public services and we did have exceptional results on some request, in fact some were by far better than our artist could make and done in a fraction of the time. Real good impressive results. However that coin had second side, as some other projects we tried (which we actually thought are easier) Ai could not complete at all, none of any model no matter how many tokens are wasted. And those failed task human can actually do in minutes where we spent days trying to make Ai do it. Why? Because not all Ai tools are there and not fully universal or trained for certain tasks. So same bottom line if you know what you need you can evaluate results and see that something is wrong, however if you don’t know what deliverable should be then you can take defective files, give for manufacturing and waste money on getting something that does not work (and this is good place to remind that LaserCutZ we charge by hour not by deliverable, if you order job that does not work and needs fixing that’s ok but we will bill more hours to your job)

Blind Text Spec Generation (please don’t)
If you are planning to generate text description for scope of work (SOW) for manufacturing something? Please don’t do that! Perhaps use Ai for topic research to get more insights on how things are done what alternatives but please don’t auto generate SOW
There is one benefit that you may think is there making quote request and SOW specs but it comes with disadvantages. In few seconds you can get large very detailed document that will have everything manufacturer needs to execute your project? Right?
Well not quite as this document will get you in trouble too, here is why:

  • Over-Specced
    We’ve seen clients asking to do simple things, like “engrave a dog tag with name” simple as it gets few words to tell and get it done. However with few accidental extra words in ai chat swings it to different reasoning such as “what is dog tag” its a military marking device used by army -> is there standard to make them? Now we have Mil spec on material covering material thickness, composition, elongation. Is there a standard to engrave those? Yes, it’s another Mil spec – depth, durability against abrasive and removal and chemical resistance, how its measured and how to comply. Now we need to provide certificates that confirm material and that job we did also conforms to that, so ready made product needs to be sent to lab to verify and issue certificate and then we can provide one to client. That is clearly not what client wanted because it was a birthday gift, any metal tag will go, just mark name, they don’t care on what and how but thats not what Ai agent picked up so we get 3 print pages spec of project which is automatically looks like complex and needs dedication as we need to follow very strict specks. At the end of day client gets price tag that is in range of what Boeing would pay or actually more as single sample cost more then production due to setup cost. So at end of the day it would be more beneficial for client to drop us email with few words of what they actually want even in broken English typed on phone and it would be faster (because we can process quick request faster) and also cheaper as we would quote exactly what client needs.
  • Wrong Assumptions
    You hold something in head but AI model does not know that so Ai will do assumptions and research (and to be fair sometimes research may help if you control and supervise) however if you fully trust Ai to bloat your specs and write them you can get something you not aware and because final document is very large you don’t have time to read it you send it out and then u find out something done wrong. Here is live story with other project certain client sent us a spec to mill parts for the ship project. very well documented project which is bid book ai rewrite and redraw to hide original project and end user. Parts are done exactly as documented. Client’s legal team contacts us with threats that project not to spec and client can not build ship as critical parts are wrong. 2 hours in reading bloated text SOW we find that parts to be made in Brass 280 (which apparently main nature of a problem) and seems reasonable request for marine use. However client’s lead engineer claims they need parts in SS316L and that what we should have delivered to them because they have everything else in SS316L. At that point legal department leaves call because project done to spec to what’s on paper, there nothing to argue about it. We submitted all certificates. What happened: lead engineer dropped spec to chatbot and said write specs, that’s about it. Chatbot processed source document and assumed proper material per their research would be brass and well written that into doc ensuring proper material is used and certified. What chatbot did not know is that this was a part used in sub assembly which needs to weld to another part which was SS316L. This speck was not given to chatbot because part was already in ship so not needed to be manufactured and it was assumed that add on will be milled and expanded to original part. If original unmasked simple just 1 pager pdf was given to manufacturer then it would raise question what material is needed as field is blank, but ai made reasonably good assumption on naval materials and spelled it out in writing. As result client ordered part they can’t use and have reorder (and pay for) new part
  • Blind Mistakes
    Ai hallucinations are less common now but not yet fully eliminated. “Facts” some facts generated by Ai may look genuinely true but not always the case yet Ai will tell you its 100% true and verified until you point solid evidence that its opposite. Many people blindly follow Ai directions and no longer verify facts, but for manufacturing, engineering and quite frankly medical and legal, mistakes can be catastrophic. here is basic samples of flaws we know
    Ask ChatGPT to “generate random number 0 to 30” guess what? your number is 17 not so random, just wait for ChatGPT to control lottery drawings and we should be able to tell future. Other models also show 17 or 23 depending on vendor but definition of random is not something that stay fixed all the time. Simple failure point simple check, but when you hand over complex job you may not know mistake is there or you as human may not be capable to trace it. So when full speck is auto generated and not verified it creates big problem at the end.

Thank you for helping us serve every customer well.
And to those who wonder what is cat doing as page header, its a sample of exceptionally generated Ai image we use for test in contrast to less exceptional laser equipment illustrations.
On this page, instructions for humans are 100% fully hand written by a human. Instructions for Ai Agents are human edited based on chatbot reasoning template