By Jennifer Dubose

From CAM to Quoting: Breaking Through Your Shop’s Bottlenecks | Ep. 546

The RFQs are coming in. Your machines have capacity. But the quotes are waiting on the same handful of people who are already busy keeping the shop running. Quoting takes experience, eats up valuable time, and comes with an especially frustrating reality: much of that effort goes into jobs you’ll never win. And when a purchase order comes back immediately, your first thought might be, “What did we miss?”

Recorded live at IMTS, we welcome CloudNC co-founder Theo Saville back to MakingChips, alongside Jason Bowes, to talk about this hidden bottleneck—and how AI could help relieve it. Building on CAM Assist, their AI tool for automating repetitive CAM programming tasks, CloudNC is expanding into quoting and estimating. Their approach starts with preparing the quote, surfacing risks, and generating an initial routing so an experienced person can review the work instead of building everything from scratch.

With more than 1,000 shops now using CloudNC’s technology, Theo and Jason share what they’ve learned since Theo’s previous appearance, including lessons from running their own aerospace and defense shop. We discuss why quote turnaround matters, where chasing precision becomes counterproductive, and how estimates, programming decisions, and actual cycle times should inform one another.

We also revisit Theo’s original ambition: making CNC manufacturing as simple as submitting a part and pressing go. How close are we? What still needs to happen? And how do you earn the trust of experienced machinists who will be responsible for the results? With growing demand putting pressure on shops to deliver faster, this conversation explores where AI can help today—and what it will take to go further.

What’s Covered in this Episode

  • (1:11) Buyers show up at IMTS with real shopping lists
  • (1:54) Why metal parts take 10 weeks and what CloudNC set out to fix
  • (3:38) CAM Assist, an AI that gets programming 70 to 80% of the way
  • (4:53) Jason on sleeping on the shop floor and why a real shop matters
  • (5:59) Why quoting is the natural next step after automated programming
  • (6:50) Ground Control automates FAI generation and PO review
  • (7:30) Why CloudNC built its own shop to shrink the problem
  • (9:43) From beta to over a thousand customers, starting in the US
  • (10:44) What the workflow looks like from RFQ to CAM to inspection
  • (13:08) The owner as bottleneck and the E-Myth trap
  • (16:38) Find Kennametal at IMTS and the Next Level Shop
  • (17:28) Running a shop on the software: ones and twos, and first over the line
  • (19:55) Don’t over-engineer the quote, just win the work
  • (21:57) What a day looks like with a quote agent, CAM Assist, and ERPs
  • (24:51) Increase spindle uptime with Hennig Workflow
  • (25:41) Estimated targets vs. actual part targets
  • (28:55) Lightweight quoting vs. full toolpath calculation
  • (30:05) Milling today, lathes on the roadmap
  • (31:34) The goal: lift the ceiling from 20 machines to over 100
  • (32:51) Can’t I just vibe code this?
  • (37:15) Let your website answer buyer questions with Navu
  • (38:28) Using AI to flag red flags on a print instead of quoting it
  • (44:28) Where vibe coding does make sense

Resources Mentioned

Read the Full Episode Transcript

[0:00] your vision that you described that, you know, you started out to solve. It reminds me of the, the little meme videos you see of, you know, when, when people say, this is what my family thinks I do, and they go put a piece of raw material in the machine and they close the door and they open it back up and it’s just there, it’s beautiful, right? Like working towards that. Yeah. Like that’s what you were trying to build, right? It’s just that like, set it in there and it’s done. And it’s not that easy as you’ve discovered, but you guys have certainly gone a long way. It is literally the hardest problem I have ever seen in my entire life. Mr. Mike. What’s going on, Paul? We are recording yet again. How many times today? I don’t know. I’ve lost count. So many. We’re, I mean, so this will be coming out later, but we’re at IMTS. We’ve been, we’re on day three. We’ve, I don’t know. I feel like I’ve recorded 97 episodes so far, but it’s not quite that many. This show is palpable energy here. It is, there’s so much demand, so much enthusiasm, so much confidence in, you know, in the demand and the economy. It’s, it’s really exciting. You know, as a, I mean, you’re an exhibitor, I’m a, I’m a shop owner, like I’m here to shop, right? And, you know, I think about years past, people in my boat are, I mean, we’re interested in things, but we maybe didn’t have a hard shopping list. Sure. Like the people I’m talking to this year, it’s like, oh no, I need to spend a million dollars, right? Or I need to spend $2 million.

[1:32] I’m, I’m looking for these three machines. I’m looking for workflow solutions. I’m looking for inspection solutions. And you’re gonna get CMMs, software, all of it. I mean, you’re an exhibitor and I, we were talking earlier and I think you’re seeing the same thing, right? Like they’re coming in ready to buy versus ready to look. Yeah. And they’re trying to solve this real problem. Yeah. This good problem to have that they have overwhelming demand. So today we have a couple of great guests that are, have been working on solving a problem for a little while. And Theo, we had you a couple of years ago with, so we’ve got CloudNC today. Yeah. Welcome guys. Good to be back. Awesome to be here. Yeah. Yeah. So why don’t you refresh the listeners on who you are, what you do, what problem you’re solving in the market. Can do. Uh, so I’m Theo, co-founder and CEO at CloudNC, came into the industry over a decade now. And the idea at the time was that I’m standing in front of a CNC machine. And if I, if I was standing in front of a 3d printer, I could just throw in a part and press go and get my parts next day. But whenever I wanted metal parts, I would have to wait 10 weeks and that disconnect of I can have it tomorrow and I don’t need to talk to anyone versus I have to wait 10 weeks and that’s killing all of my projects. That didn’t make a lot of sense to me. And I just started pulling the thread and I started digging and digging and digging. First I assumed, well, you know, maybe if we build machines with better sensors, then the software will be able to take the data from the machines. And then you’ll be able to machine automatically.

[2:55] And then I saw the software and I was like, Oh my God, like we’ve got a lot of work to do before we get to auto instrumentation and I just ended up at like the very bottom of the problem, scratching my head on why is this so hard? Why can’t I just fire a step file into a CNC machine? Can we build something that would be able to do that? Now the problem is so difficult that the answer is no, you can’t go there straight away. And so what we ended up building was the first AI assistant for CAM, something that would do the job like 70, 80% of the way, take away all of the boring repetitive grunt work, the operations that you do every single time, the simple stuff, and then, you know, much like you would do with, I know, like a GPT created blog article, you would fine tune. You’d like take something that was output, but wasn’t finished. And then with that starting point, you’re going five times faster. And back when I started, there was, we already had a skills problem. Now I’ve got people coming up to the stand saying, I can’t even hire a second CAM programmer. Like that’s how tight we are now. So I need to get more out of the people that I have. And that’s what we’re trying to do. And we’re expanding now beyond just CAM. We want to make anything that is happening in the factory, particularly quoting right now. If it is like repetitive and boring, we want to make that automatic and seamless and push the skill level into more interesting, higher value places.

[4:11] Leo, your vision that you described that, you know, you started out to solve, it reminds me of the little meme videos you see of, you know, when, when people say, this is what my family thinks I do, and they go put a piece of raw material in the machine and they close the door and they open it back up and it’s just there, it’s beautiful, right? Like working towards that. Yeah. Like that’s what you were trying to build, right? It’s just that, like set it in there and it’s done. And it’s not that easy as you’ve discovered, but you guys have certainly gone a long way. It is literally the hardest problem I have ever seen in my entire life. Well, it’s the hardest business that there is in the world, right, Paul? Where’d you hear that? I read it in this really good new book. It’s very kind. Yeah. Credit to Jason. Like part of that journey was building our own shop. So Jason, tell us about your role in all of this. Awesome. I’ve been with Theo for seven years now. I like to say it’s my job to make his dreams come to reality. Execution doesn’t matter, ideas are great. I have a few of my employees here. I want to make sure to say that again. Someone’s got to think of the ideas. Someone’s going to make them into reality. And yeah, I’ve spent the last seven years across various functions. I’ve slept on our shop floor for a whole year in the run-up to my wedding. I understand the problems deeply. You’re an understanding fiance. Uh, yeah.

[5:26] She’s very, very understanding. Um, yeah, most people are sleeping on the shop floor after their wedding. But it was all worth it in the end. We’ve got a great shop down there. We wouldn’t be able to do anything software wise without having that shop. I don’t think you can truly understand the problems of machining without experiencing it firsthand, that allows us to build better software for our customers. And that quoting is a natural progression out of automated programming because what do you need to know to know what the price is? You need to know how much cycle time, how much machine time you’re going to take. 100%. And we see that’s the biggest bottleneck right now. Demand is absolutely booming. Everyone’s inundated with quotes. The worst bit about quoting is really high skill, often done by the most senior person, getting it wrong is really bad. Yeah. You under cost it, you’re guaranteed to win it. And you’re going to lose money on it. And 90% of what you do is not going to be won anyway. Right. So it’s a real tough, tough one for shops, but necessary. Yeah. When you win a job, you’re then you’re just like, oh man, did I only win this? Cause I totally screwed the, screwed it up. And I was, I missed something that was critical. Yeah. We’ve definitely got that, that quick PO back and we’re like, Oh no. What did we do? How much money did I leave on the table? What did I miss? Call them back right away.

[6:43] Sorry. It was supposed to be two X. I missed a decimal point. See what they say. America is re-industrializing with unprecedented growth in aerospace and defense manufacturing. If you’re a manufacturer trying to scale growth, you need ground control. Ground control automates specific workflows like FAI generation and PO review. They help over 200 manufacturers break OTD records and collect revenue faster with easy to use software tools. The best part, they are FedRAMP compliant and can handle ITAR and CUI data. And if you’re a ProShop customer, reach out for an exclusive integration to seamlessly create inspection plans without any manual data entry. Check them out at rungroundcontrol.com. So I’m curious, did you guys start the shop so you could have a place to innovate your software or did you have the shop and the software was trying to solve problems you already were facing in the shop? You know, the truth is it’s actually a little more complicated and uglier than that even, but when we started out the first two years, basically spent realizing month by month, how hard the problem that we just tried to tackle was, and when we hit the two-year point, we realized, oh my God, we are not going to be able to ship anything for five or eight years that we wouldn’t be able, that any respecting shop would actually purchase. We will not be able to build a useful piece of software. We have to do something else.

[8:08] And so our idea was if we build our own shop, then we compress the problem. One type of machine, one narrow set of customer. Maybe we’ll only do like aluminum that fits into a certain dimension. We’ll make the problem simple enough that the software we can build today can actually deliver some impact. And eventually we will get to the point where we’ve built enough software that we can give it out to everybody. And that middle journey took about six or seven years, like six or seven years of trying to like scale up and make parts and investors loved it at the time. So it kept us alive for long enough to deliver that technology. And then this point hit maybe two and a half years ago where we just realized, huh, someone else might buy this by now. Yeah. Yeah. Someone else might genuinely buy this. And we, and we tried selling it and people loved it. Very good. Did you ever have the thought that you could sell parts to fund the software developers? So you could actually. Cause I started my shop. See how excited the software developers are? Yeah, I know, I know. I mean, I joke about that, but when I graduated from college, I wanted to start a car company and we didn’t have the money for it. Didn’t have the expertise for it. So we said, why don’t we start a machine shop? And then very naively we thought, you know, after just a year or two, we’d have enough free cashflow and profits and free time that we can build a prototype and go find investors and start our car company, little did we know we were getting into the hardest business in the world and there was no free cashflow, no extra profits, no spare time. We were sleeping on the shop floor. Oh, you don’t need a nice car when all of your time is spent sleeping on the shop floor.

[9:39] That’s true. You’ve got no time for your other things. No time to drive it. Indeed. Yeah. So when did you come out with the first version? Yeah. How long have you been out on the market? What does that look like? First version was bought two and a half, three years ago now? Three years ago with the original beta version that we put out just in Fusion. Really, I would say we’ve been out for about two years now in the market. Scaled today to over a thousand customers already. Using this globally, focus here in the US. Yeah. Yeah. It’s been a great journey. It’s been really fun. And I think you were telling me yesterday, the day before, although you’re located in the UK, you just sell in the US. I mean, it’s the best market. Let’s be honest, it’s got the most people. I don’t know if you’ve noticed, but Americans love buying things. It is honestly, it is so much harder to sell something in like the UK or Europe than it is here, right? Currents are risk takers, you know, and much, much faster to adopt a new piece of technology. Yeah. I just really like building a go-to-market here and selling product. It’s really fun. Awesome. That is awesome. Yeah, absolutely. We are out in other countries now, but when we first went to market, probably for the first 18 months, two years, it was exclusively the US market. Okay. Just to back up just a hair, cause I, you know, we’ve talked about quoting, we’ve talked about CAM, we’ve talked about, let’s, let’s, let’s just do a high level overview of if I’m a CloudNC user, what separate workflows there look like for me? I mean, I mean, so I know quoting and, and CAM automation and, and you’re developing out both of those just, just briefly there, like how deep can I go there and, and also, you know, maybe we start kind of talking about where are we going from where we’re at today. If I’m sitting in my shop, it starts with email coming in from the customer with the RFQ, I want to convert that into a quote, I want to convert it into a price. I want to convert it into a manufacturing plan.

[11:23] I want to do that very highly autonomously. I only want to be providing little nudges, like saying, I know, put that on an Integrex, Oh, you have an Integrex. I’ll store that in a database. Oh, you know, this customer is a pain in the ass. Please add a plus 10% pain in the ass fee for every time they need a quote. All right. I never have to do that again. Keep building up that database quality. That’s the other thing that I want to do a very high level of automation on because the amount of time in our workshop that people are like sat them high hand, filling in fares, filling in forms, it’s an obvious thing that, you know, just should be streamlined and automated now, and then of course you get to the CAM itself. I want to be very highly automating the grunt work of running those machines. Then eventually I’m going to want to get into creating the quality inspection routines for the CMM machines as well. And maybe the quality plans, like I just want to go through the factory and keep on looking for what is the next incredibly boring monotonous task? What about the next one? What about the next one? And our customers are just turning up and saying, can you add this? Can you add that? Can you add that? And these days we can just say, yes. Right. Five years ago, we couldn’t say yes. Yeah. We could have said, we’ll see you in two years time. Maybe we’re sorry. Yeah. You know, Paul and I were just, we just finished another conversation about AI and I love the way Paul talks about like what you’re doing with ProShop and chip is, you know, just closing that loop. And it sounds like, you know, you’re kind of chasing that same thing, right? It’s like, let’s close this loop of all these, these tasks that have to be done. And as Paul put it, Paul often puts it, but put it a few minutes ago was, you know, let’s, let’s make the machinists as successful as we possibly can, right? I mean, let’s give them everything they need when they need it to be successful at, at, at cutting a chip.

[13:08] Yeah. Right. And something that blews, blew me away when I was first learning about the industry was why is it that the average shop doesn’t get so much bigger? You know, if you go to, go to most other industries, you see like the size of a supermarket or an Amazon warehouse, these places are massive and that’s just like one tiny piece of them. But if you look for the mega scale workshops that don’t really exist, like what have you got, Foxconn, 70,000 machines, they’re just doing iPhone bodies all day long, that simple. You can just have the same machine over and over again. But every workshop that I ever knew, including our own that tried to achieve mega scale, got punched and knocked right back down to be a much smaller operation because as soon as the workshop can’t fit in the owner’s head anymore, as soon as it can’t fit in one person’s head, then everything crunches and fails and grinds to a halt. And so there’s just like this, this ceiling of, depending on your work, somewhere between 20 to call it 80 machines, like 80 to a hundred for serial stuff, 20 for job shopping, no one gets bigger than that. And if we, yeah, if we can relieve these problems, then that means that the machinist, the shop owner can achieve a scale, what, 10 times bigger than what is possible today. I’d love to see that future. I never remember the exact numbers, but like to your point, the, you know, there’s 19,000 roughly like job shop, machine shops in the United States. It’s roughly like 10 machines, 12 people is the average, right? So, I mean, like when you’re talking about mega shops, there’s not, there’s not a lot of mega shops if your average is 10, right? And, and there’s a lot of one and two machine shops, right? And you’re right. I mean, Paul and I have talked about it many, many times on this podcast.

[14:45] I mean, it’s the more the owner sits in the middle is the bottleneck and I’m an owner, you know, I’m speaking as an owner, the more that I’m the bottleneck, the more problem we have growing ourselves. Yeah. Well, I think the, this gets into sort of the premise of my book is that, which was stolen totally from the E-myth, is that a machinist, a tool and die maker, someone that’s excellent at the craft thinks they can go into business and just do that thing, but they don’t realize that there’s 10 times more activities that have to happen well, that are not nothing to do with machining. They have to quote well, they have to, you know, hire people and retain them well. They have to plan things. They have to schedule things. They have to do quality. They have to do all these things. You gotta invoice. You gotta collect it. You gotta bank it. I mean, like everything. They don’t, they’re not experts at that. And they don’t think about how to systematize that or, or automate that very well. And so they hit their head up against the ceiling, sometimes in a catastrophic way, if they grow too fast and they realize that they don’t have the foundation underneath that new size and sometimes they come crashing back down to a smaller size again, yeah, and so software can be a massively powerful tool to automate some of those things, to help provide best practices, to structure systems that just make that, that don’t have to occupy all their mental capacity just to keep that side of the business running.

[16:10] Yeah. Yeah. When you’re consuming your whole mental capacity, just doing repeated operations. Not problem solving, not like building the next process. It’s why I like big batch and serial production. You build a process and it runs and then you build the next thing. But when all of your energy is just going into make it work, make it work, make it work, make it work the same thing over again, you don’t have time for anything else. Exactly. I don’t know about you, but I get very bored. Let alone your kids and some family life. No, just make your kids work in the shop. There you go. Hey everybody. Guess what Canna Metal is doing at IMTS 2026. It’s called the Next Level Shop, an exclusive booth experience packed with live machining demos, game changing product announcements, and industry experts ready to talk tools and shop floor challenges. And here’s the headline. Motorsports legend, Cary Hart is teaming up with a custom fabrication crew to machine motorcycle components for an Indian sport chief live on the show floor. And guess what? One lucky visitor gets to take that bike home. You’ll also find automotive manufacturers and Canna Metal customers showcasing the tooling that keeps their operations running. Plus digital machining demos with partners like Autodesk, Fusion, Toolpath, and Module Works. It all kicks off on Monday, September 14th with a major announcement from Canna Metal’s president of metal cutting.

[17:31] So don’t be late. Are you ready to take your shop to the next level? Find Canna Metal at booth 431-800. It’s in the West building on level three. We’ll see you there. Jason, you’re running a shop. You’re, I mean, you’re our listener. You’re, you’re trying to get out there and you’re making parts. You’re trying to cut chips. You also have the benefit of having this software company running alongside parallel with you. Right. And, and trying to, trying to change the industry and change how you operate. Tell us your experience, right? I mean, you’re, you’re running a shop. What is this doing for you? Yeah, absolutely. So I honestly don’t think we could have run the shop without the software. We came from a very different place of not having a working shop to then lay a software into, and I had to build the software alongside, completely, with that infrastructure there, you know, not a single machine in that shop. We tried to scale it too quickly, even with the software, we found there was a lot of challenges and bottlenecks to doing that. Tried to serve too many customers at once. And that’s, as Theo said before, we, we niched down, it’s now an aerospace and defense focused shop doing work for tier one defense suppliers in the UK. And that constrains the problem a little bit. We layer the software in there because we do a lot of ones and twos offs. And that’s where CAM Assist, our, our AI for CAM really shines because as everyone knows, once you win the work, if you’re doing only one and two, the machining time is kind of irrelevant.

[18:53] It’s all about that programming time and getting it on the machine to try and get that machine utilization up. So that was the biggest way we could get an unlock in the shop. And the second one was obviously winning the work in the first place. So being able to quote fast, particularly with the bigger suppliers, you know, you got to remember someone on the other side of that purchase order you want is sat behind a desk, they have thousands of parts of the big suppliers to go through. Often first over the line wins. So we focus really on building software that allows you to get first over the line. That’s such a great point. Cause I, I think people sometimes forget, you know, when you look at like your buyer persona, right? I mean, there’s just a buyer on the other side of the line. They have 12 things they need to get done today. And one of those is order this part that they need in eight weeks. And they may not. Or two weeks. Or two weeks or whatever. Sometimes a month. Like it just, like they got an email from engineering or production or whatever that said, Hey, we need an order of these or their, or their system said, we’re going to be out of these in eight weeks. Make sure to order some. Right. And so they just got notified. They need to buy some. They’ve got their kid’s birthday party this afternoon. Like they’re just trying to get their checklist done so they can get the hell out of the office. And I think sometimes, I know we’re guilty of this. I mean, I think sometimes we put so much effort into getting that quote back that it takes us a long time. Right? Like one of the things I talk about with my team is like when we’re quoting, we don’t have to job cost, right? It’s called an estimate for a reason, right? It doesn’t have to be this perfect job costing, you know, scenario of this is exactly what it’s going to cost.

[20:25] No, it’s not. It’s going to be wrong. It is going to be wrong. Let’s just make sure we win on the wrong. Right. I mean, like it’s, it can’t be so wrong that we lose. But you know, I think we overemphasize sometimes how much they really care about what the quote is. They just need the number and they need it quick so they can be done with that, that piece of their to-do list today, right? Absolutely. We find a lot of shops over-engineer the costs at quoting and it could take weeks to be able to do that or days. And you’re going to lose a lot of work doing it that way. A hundred percent. Also the inputs are hard to guess anyway. So you’re likely to be wrong. As you say, it’s more important to be there to win the work and to understand the contribution of getting contribution into the shop. Your spindles are not turning. You’re not making money. There’s no point having the perfect quote if you don’t have enough work. It’s better said if you’re not making chips. You’re not making money, but I get your point. You see, he set you up perfectly for that. And I saw you, you were like, we got to get on that. Same page there. I, I, I totally get that piece. I run a shop, you run a shop and you know, I have manufacturing engineers, programmers, I’ve got an estimating team.

[21:40] I assume you still do too. Yes. But you, but you have this software supporting them. What does that day look like for them versus like, I’m not using CloudNC, right? I don’t have that. So, so I get that email, I get that model, I get that print. It goes to my estimating team. They’re breaking that down. We get the PO. It’s, it’s now transitioning from them to the, to the manufacturing engineering team to say, okay, this is how we’re going to make it. That’s what we need to buy for material and tooling and you know, whatever. So I know what that looks like. And I think that’s what the majority of shops, like they know what that looks like, right, because that’s how most of us operate. What does that look like for you? The honest truth is very similar, except the software provides the leverage to do that with way less people and way more efficiently. You still have to get the job in. On our quote agent, the new products, the agent will actually produce the initial draft of the quote for you. But the human is still going to come in and review it. But instead of sitting down and trying to have to assess everything, pulling out all the risks with the job, it’s going to do that and put them right in front and center for you. It’s going to generate the routing automatically and you’re just reviewing over the top of it. The idea is that you can get it out in a couple of minutes instead of 15 minutes per part. And when you compound that effect over however many parts you’re getting in each month, that’s massive. And the same is true on the CAM side of things as well.

[22:55] We’re going to get you 80% of the way there with the tool paths. The programmer still needs to review, simulate, verify, and put it on the machine. But one programmer can now do the work of three, four, five programmers, depending on the work. When you’re done with the quote and you have developed this sort of estimated plan, how does that get into the ERP system of the shop that then needs to make that part, assuming they win it? Ah, would you like to put in a feature request? Okay, I guess I set myself up for that. Yeah, give us a week. It’s interesting. I’ve heard that mentioned at the show five or six times already, or the seventh. Uh, but so we will have to figure it out. Like it’s going to have to get into the ERP system. I don’t know how many people we’re going to be selling to right now who actually have an ERP system versus not right now, but I guess that’s going to be the next most common ask. I’m sure. Yeah. Yeah. Oh, I mean, I like, I mean, for me, I mean, cause I do, I have Paul’s ERP. You know, if a solution doesn’t integrate with that, like it’s I, I I’m immediately going, okay, well, how am I going to though? Right. Cause it’s, it is so core to delivering the part at the end of the day. Yeah. Right. It cannot be by hand. It can’t be by hand. I would, I would hate myself if it was, if it was by hand.

[24:11] We’d be failing our entire mission if it were by hand. So yeah, we’ll figure out a way to do it. And I know ProShop does, and I’m sure a lot of others do Paul. I’m not trying to. You got an API? What’s that? You got an API? Absolutely. Yeah. I mean, so as long as you could tap into there, I’m like, we’re good. Right. Yeah. You know, I still don’t understand how the APIs work. But in your own shop, I’m curious, you know, you win a work, you win a job. How do you then? Well, they know if we don’t actually use an ERP in our shop. Oh, that’s right. We actually have this conversation. I think this is getting ready to turn into a sales call for Paul. Paul is ready to sell us any day now. And look, as you scale, you really do start to need one. And that’s where it comes in. And the reality is you need that system of record, auditability. We’re AS9100D. It comes with onerous audit requirements. Yeah. Our job is, you know, we’ll have to integrate with ERPs such as ProShop to make sure they can go into the system of record and continue on through the shop journey. Our job is to automate the tedious tasks, not to the system of record. If you’re looking to increase your spindle uptime without adding any head count, check out the Hennig Workflow Automation System. I’ve seen it. I love it. It’s a compact automated pallet delivery solution that lets you work ahead of your machine so your shop never stops. You can load up to 40 pallets with material, fixtures, programs, and just let the workflow cook. The Hennig Workflow is designed for high mix machining, works with a wide range of machines, and even with 125 pound payload on top of a 300 millimeter pallet, yet still takes up less space than most competitors pallet automation system.

[25:43] That means more production, fewer manual changeovers, really just a practical path to high profit lights out machining. Learn more about Hennig Workflow by visiting hennigworldwide.com slash workflow. I want to sort of dive into a little topic. You know, so back in my shop days, I was the estimator for the first probably 12 years before we hired an estimator to replace me. So I quoted thousands of births. Wait, Paul, was he better or worse? 30 years my senior. So he had a lot of experience. He was good. Yeah. He was quite good. I’d say he missed less than I did. I just, I wanted to point out like as owners, we got to hire good people and get the hell out of their way. Oh yeah. Yeah. Oh, absolutely. Yeah. Yeah. No, this guy was a deeply experienced aerospace estimator and yeah, he was good. The point is like when you, you’re always going to be wrong, right? You’re never going to hit it on the nail. The way that the sort of the process that I, that sort of evolved in my head over time, you know, you, you use whatever tools you have at your disposal to get it as accurate as it needs to be to get the quote back on time and win it. At a margin you think pretty likely you’re going to make, make money on. You then flow that into the manufacturing plan, right? It becomes the foundation of your plan, you know, two mill ops and a lathe op and, you know, set 37 total minutes of cycle time, whatever it might be. When the programmer then gets their hands on that, they look at it with a closer level of detail than the estimator did because they now have to actually program the whole thing to make it. And as they start, you know, doing their simulation, they should be updating the target time that came from the estimate, but not the estimate, like that still should be retained, like that knowledge of what you guys thought it was going to take.

[27:29] So we differentiate between like an estimated target and an actual part target because the part target will then also define your schedule and the larger a quantity you’re doing, the more important it is that that get as close down to the minute as possible. So if you’re off by 10 minutes and you’re making a thousand, that’s 10,000 minutes, you’re going to be off by weeks on how long the machine time is going to be. And thousands and thousands of dollars. Yeah, absolutely. And then you’d misquote everything else because your lead times are off as well. So the cycle continues around and then the job comes back and then you do it all again. Yeah, the programmer should, should make that more accurate. And since you guys do both, maybe your software can even tell itself that, you know, you should be more accurate here because that’s the dream. Yeah. Yeah. Closing that loop around. Yeah. Yeah. And then once you actually start making it and you have real cycle times, then that should be fed back into what the actuals should be. And that should be helping teach the algorithms about quoting, right? Exactly. Not just the quoting, but the machining, like the version of CanAssist that’s in deep development at the moment. Is much more interactive in that way. And designed to truly learn the preferences of the workshop and the machinist, just like QuoteAgent. Every time you talk to it, it learns something about you, your preferences, how you run your shop. Like the way that I use a computer now, I mostly just use cursor. And every time I correct it at something, it is learning so that it doesn’t make that mistake again, in theory.

[28:50] Usually I have to tell it three times, but, and I want that same experience of having to tell it three times. Yeah. It is just like teaching a child. But we want the same experience in CAM. I think it’s what people most ask us for. Like when they’re, when they’re, if you look at all of the feature requests and really dig down, what are you truly asking me for? It’s, can you please learn how I like it done and keep on doing that? Right. So I guess this, this creates a question that I don’t know the answer to. When you are quoting a new part number, is it actually doing all the toolpath calculations for the quote, or do you do something that’s lighter weight and then you go deeper once you actually want to create the G code? It’s exactly that. So we’re doing a very lightweight version. It uses the same core engine as the CAM engine, but of course, no one wants to wait 20 minutes for a computation to calculate all the toolpaths. So we’re not generating toolpaths at quoting, but we’re looking at how the part will be held, the machine ability of it. We’re looking at tool accessibility, what tooling you’ll need, but in a very lightweight, quick calculation. Okay. And when it comes, because you don’t need to be that accurate at quoting, it’s about getting close enough and I’m a firm believer in just when enough work to keep the, keep the spindles turning and you’ll be fine there. Keep making chips. Yeah. Making chips. Um, we’re going to get you, we’re going to get you there.

[30:05] Do you have a baseball cap I can wear? A spindle turning without touching metal is not helping anyway. It’s got to be actually making chips. That’s how we get our utilization up. That’s right. Just cut air all day long. But of course, when you get to the CAM program, it’s got to be accurate. It’s got to run on the machine and that’s when we do the real heavy computation of actual work of toolpaths. Now, are you guys doing, I know you’re doing milling, are you doing turning too? Or is it just milling so far? Just milling. Milling is a hard challenge. It is. Yeah. It’s a big challenge. We do a three and three plus two axis milling, all vertical and horizontal lathes, three plus one as well, but turning in the future. Can I selfishly ask like where lathe is on the roadmap? 80% of the work I run is lathe. Oh, it’s lathes. It has to go through a lathe first. So. Oh, actually I have a question then. Do you program on the controller or do you program in CAM? CAM. CAM. Okay. Awesome. I think that makes you the exception. It really does. A lot of turning. Programmed on the machine these days. Oh yeah. No, I mean, yeah, no. It depends on the complexity of the parts, right? You’re doing simple shafts and tubes and stuff. We do simple shafts on CAM. I don’t think a program hits my floor without. Without being CAM. Oh yeah. I do think that’s unusual from our experience. On lathes. Maybe I’m doing this all wrong. No, you’re doing it right. You are doing it right. Believe me. Are you guys, did you guys just tell me I’m running my company terribly? No, we think you might be further ahead than everyone else. That’s why you’re a top shot, man. Exactly.

[31:23] Yeah, fair enough. So, I mean, is it on the roadmap? It is on the roadmap. If I, if my co-founder has this thing where he says, please stop committing to roadmap dates on, you know, public forums. Yeah, sure, sure, sure. But what I will say is that for a really long time, it’s been so far in the future that I wouldn’t even talk about it, you know, and I would disappoint you, but it’d be like, Oh, I don’t know. I don’t know. Not as soon as I would like, but not nearly as far away as it used to be. Um, I think that, you know, hopefully we’ll be having a conversation about it by next IMTS. Okay. We’ve kind of talked about where we’re at, what, what product is available today from cloud and see what that workflow looks like, anything, anything new and exciting that, that the audience should know about that’s coming in the near future, I’ll talk about a vision, maybe by next IMTS, right now, the biggest workshop that we can basically build and run ourselves is about 20 machines doing drop shopping. Okay. What I would like is to come back to IMTS in two years time and be able to say that number is now over a hundred. Okay. Because we have solved enough of the problems that create that ceiling, that the ceiling is lifted. Okay. And the ceiling does need to lift. That’s what I want to achieve. And so we’re just going to keep on working through problem after problem after problem until that becomes true. So, you know, my, uh, my shout out to the audience was B if whatever is that thing that is holding you back from running 10, 20, 50 more machines, we want to know about it. We want to know about it in detail. We want to know about it in depth.

[32:49] Tell us and we will go and we will build the solutions to that. Okay. And we don’t know what they all are yet. You rarely know what the next three bottlenecks are behind the one that’s right in front of you. That’s the interesting thing about bottlenecks, right? Every time you clear one, you’re like, Oh, it’s just three more. Yeah. Like whack-a-mole. Yeah. Yeah, it is. I, I always love asking you about this feeling in AI. There’s a huge, like undercurrent, certainly in manufacturing, much larger of, well, I can just vibe code this, right? Like I can, I, it’s no big deal. I can, I can just hop on cloud. I can spend all night, have a couple of Monster Energy drinks and I’ll have an ERP system in the morning, or I’ll have a quoting system in the morning, or I’ll have a cam system in the morning, like for example, I mean, I have just out of my own curiosity, I have, I have given chat and cloud and you know, whatever, like some prints and some models. And I’ve said, write me G code for this. And it writes something. And, and actually the first time I did that, I brought in my lead programmer and I said, read this G code and tell me if it makes part number 27019, which that’s just one we all know off the top of our head. And he, he reads the G code and he goes, yeah, actually it, it makes it perfect. Like, it’s like, okay.

[34:08] He’s like, how’d you do that? You don’t know G code, you know? And I was like, well, I dropped it into, you know, cloud, I think at the time. And I know there’s an absolute massive world of people that are doing that. So why can’t I just do that? Why do I, why do I need what you’re doing? So I’d say it’s easy to vibe code something that sort of works on a, on a whim and a prayer. What’s really hard to know is what to actually build. First of all, if you know exactly what to build, then yeah, you can probably build something that will solve your problem now. Very few people actually know exactly what to build and how to build it. And with the AI, even if you know what to build, you’re not going to be able to do it as well as a real software engineer who knows how things should be built as well. So it won’t be as performant and it won’t be as reliable. I’ve vibe coded some wonderful applications that it turned out didn’t have a database in them, and then I pushed another commit and then all of my lovely data that I had spent absolutely ages putting together was just like, was literally lost because it had been storing it as like cookies in my web browser, rather than on like Amazon S2 and, you know, it gave itself big pat on the back for what great software it had built and I thought I was a genius and now I know I’m not.

[35:23] If you’re like, I can let go of eight software engineers. No, you can’t. You definitely can’t. Anyone can build a prototype now. That is definitely true. Some people can build things that genuinely solve really difficult problems. Like some people will end up running their shops on Claude. But in most cases that will not compete with guys like us who know exactly what to build for a wide variety of customers. And we can build extremely wide, expansive pieces of software that are all carefully linked together and cross all of the T’s and dot all of the I’s and have the robustness, the reliability, the certifications, you know, we will go deep, deeper than you will. It will not be worth your time to try and vibe code everything. Cause I’m sure when you started vibe coding, you got like very far in that first night on those two monsters. And then you spent the next week or so trying to make it all work, trying to make it work, and then eventually you just like give up. Right. That’s the thing. I like the analogy of a, of an iceberg for software engineering. It’s very easy to make progress on that bit that everyone sees in the UI. Yeah. But at the end of the day, you’re going to upload your customer’s parts to an unsecured database. You don’t know where it’s going. You’re going to have your auditors come in for your AS9100 audit and audit a system you can’t trust. And these models are probabilistic to the G-code point, like, yeah, they might get one right.

[36:39] But the next one, they might crash your spindle and the margins for error in manufacturing are too wide for you to possibly trust that. Yeah. You know that like, so I’ll give an example of the opposite of how like we’re using that. So I, I mean, I’ve got these like hardcore experienced machinists that it will take them six months to trust your output from the quoting system, right? Like, cause they’ve been doing this for 30 years. They can look at a print and then five minutes, they can tell you, oh, we’re gonna have two lathe ops, three mill ops, each one’s this time, right? And, and they are not going to trust your output or until you prove it to them 99.9% of the time, right? And I, and I think that’s something a lot of our listeners can relate to. Yeah. Hey folks, listen up. You know, that manufacturing buyers don’t ask simple questions. They’re asking about product compatibility, materials, specs, lead times, application fit, and they want answers before they even talk to sales. That’s where Navoo comes in. Navoo adds AI chat to your website, trained on your own content. So visitors can get reliable, accurate answers in real time.

[38:04] It helps buyers find the right product, understand the right application and take the next step without forcing your team to answer the same questions over and over. And setup is simple. Navoo works with your existing website. It connects with tools like HubSpot, Salesforce, and Slack, and gives you visibility into what visitors are actually asking. See what your website has to say at navoo.co slash making chips. That’s N-A-V-U dot C-O slash making chips. So I didn’t try to solve that problem in, in our use of AI. What I said is, Hey, let’s, let’s create a tool that they can upload that print, upload that model. And just do a review of it. Like tell them what they should be paying attention to because you know what they have, you know, where they have failed is they missed that surface finish. They missed that outside process. They missed that we need to part mark the part, right? They didn’t see that on the print. They didn’t see that on page two in note seven, right? They missed that there was a plus or minus two tenths tolerance. That’s where they failed, right? That once the job came in and we did it, you know, we joked earlier, like sent a quote, get a PO back immediately. You’re like, Oh shit.

[39:37] What did we miss? Well, we missed that there’s a two tenths parallelism, parallelism over a 14 inch, you know, bore, right? Like they missed that. That’s why we won that job in four seconds. Right. And you’re going to lose 200% on making that part. So all I’ve done is I’ve, I’ve said, Hey, use this like little agent we’ve recreated to give you the red flags, right? I mean, I think that’s okay, but I would not trust ourselves to build a tool that just says, okay, I’m going to take that. I’m going to quote it and I’m going to program it. Yeah. We would, we’d be crashing every machine. Exactly. I’ve got an analogy I like for this. I bet all of us could build an airplane. Would you get in the airplane that I built? Hell no. I five coated a Cessna. Do you want to have a go? Yeah. That’s a great point. Yeah. We build mission critical software and precision machining is a mission critical industry. You miss that toll. Like you make any mistake and it compounds and it snowballs and it costs real money and it destroys real components or crashes a real spindle. That is mission critical software. It’s not, you know, making a video out of SeaDance or writing a blog on the internet.

[40:57] You messed this up. There are very serious, very real business killing consequences. And not only business killing, but I mean, we do critical work. It could be people killing, right? I mean, it’s, it’s, it is real. And I don’t want to vibe code that at two in the morning on a monster energy drink. And I bet you tried. We tried. Well, another, and this is actually an actual cautionary tale of a friend of mine that has a business. He was not having good luck with his website that was like a Wix or something like that. And so he decided to with Claude or some other tool, you know, vibe code a website. And he was very specific about making it really good with SEO and, you know, discoverable and all those AI friendly and all that. And he had incredible results. Like his, his website went from having like a hundred visitors a day to like a thousand visitors a day, just over like a course of a few weeks. But then when you wanted to make any change, it was not a CMS. It’s not like you could go in and change a photo or make this text bold. You had to literally just prompt. Rebuild the website. Only he could do it. Cause if you didn’t quite prompt it properly, it would just implode the whole thing. So it was like this deck of cards that was so fragile that you didn’t dare, you know, mess with it. And if you try to build an application that’s running your business, it might look great on the front end, but like you were saying, you know, it’s, it’s just ready to like break apart and destroy your data and you lose it all because it’s stored as cookies.

[42:29] It’s like a rollercoaster with a missing segment, you know, the cars are gonna like, they’re going to go over the segment, you know, they’re gonna arc over wonderfully and re-engage, you know, 49 times out of 50, but I don’t want to be riding that rollercoaster time number 50. You know, the, the analogy I like to make and Paul’s known this, but you guys are both much younger than Paul and I. I used to come in, I mean, back up 30 years and, and like I created an amazing quoting tool over the weekend in Excel. I vibe coded, right? Like this amazing quoting tool in 1998 in Excel. And when I bought my current shop, they were quoting in Excel off of a template. And I was like, you know, the math doesn’t work. And I start kind of going through the template. Well, you know, at some point someone did a save as, you know, instead, they saved instead of did a save as, right. And they broke a formula, right? So we were under quoting everything. That’s why we were winning so damn much. Why were we so busy? Like we were working geniuses. Yeah. I don’t, I don’t remember what the break was, but like we were excluding material or something like that.

[43:44] And it’s very compelling. Yeah. And that’s where the danger is to me in like trying to just vibe code this, right? Like, I mean, you have core business functions. You need like trusted partners that are, that are doing like real development, real testing. I can’t quote on and not include material. That would be a major problem. Of course. Your clients would love you for it. They would. Well, we were winning a lot of jobs, Paul. Yeah. Can you trust your software? Right. That’s it. You have to be able to trust your software. Your software isn’t saving you time if you don’t trust it. It will bite you at some point. So if you trust your software, then you can go and work on other problems. Software is supposed to completely automatically solve some kind of problem for you. Yeah. I mean, if, you know, to that point, like if I vibe code something, I, I need to trust, but verify it. Yeah. I want core data. I don’t want to verify it. I just want to trust it. Yeah. Right. But we shouldn’t discourage people from using them on non-core processes. Like the tools are incredible. By the way, sure. Building vibe coded ways, just to give hints and tips, find manufacturability issues, automate data flows, all wonderful use cases.

[44:59] Dashboarding. Dashboards, customer ports. Whatever. It’s great. Like vibe code, like have fun. Yeah. Buy a case of Monster Energy drinks and go to town. Core data. Like let the, let the professionals handle it. Well, even in the case of like a shop that serves, uh, you know, 1345 medical devices, they have to have a software validation process before they can implement a new revision of software. So they literally have to put in inputs and verify that the outputs are correct before they can, if they don’t, they will fail their audit. Yeah. So there you go. Yeah. That’s a lot. Yeah. Yeah. So even developing software for them, you have to develop a validation process or else they have to come up with one on their own, which is a big lift for a shop. That’s just trying to keep their spindles turning. Yeah. Keep making chips. Sorry. Excuse me. Keep making chips. Keep making chips. Dude, I should know better. Come on. We make software so you can make chips. That’s right. There we go. That’s a good way to wrap up. Yeah. So if people want to learn more about CloudNC, where would you direct them? Straight to cloudnc.com. Yeah. Or our Instagram, check out our videos. We’ve got some really cool stuff. You guys, we’ve got a, we’ve got a 250,000 frame a second camera that is hooked up in one of our machines right now that we’ve been making some very cool stuff with.

[46:19] So you can see like ceramic, you can see cutters that are exploding in real time. Like you can see the crack forming through the ceramic or through the carbide before it blows apart. That is super cool. You can see the high pressure coolant like starting to very slowly come out and then like it’s patterns change. Uh, I know what page I’m getting ready to go. I know that’s, that is so cool. Yeah. All right. Well guys, thank you so much for being on the show. That’s great to learn more about what you guys are doing. It’s very cool. Thank you so much. Yeah. You guys are having a, like a real impact on the future of the industry. And I just appreciate that as someone that’s a fan of the industry. Yeah. Nothing else. I mean, we need more tools, more great brains trying to solve the problems of manufacturing because we need to, you know, raise the bar across the board because there’s a lot of parts to make, a lot of challenges to solve, and we need all hands on deck. If you’re not automating the mundane, then you’re not making as many chips. If you’re not making chips, you’re not making money. Bam.

This transcript was auto-generated and may contain errors.

Connect with Theo and Jason of CloudNC

Connect with MakingChips

Subscribe

For the latest episode straight to your inbox, business-building tips and other exclusive content, Subscribe to #MetalworkingNation today.