The ROI of Pre-Roll Automation: Scaling Quality, Reducing Labor, and Hitting 20K+ a Day ft. Shahar Yamay

Listen Now

Bryan Fields: What's up guys, welcome back to another episode of The Dime. I'm Bryan Fields, and with me as always is Kellan Finney. This week we've got a very special guest, Shahar Yamay, CEO of Festi Tech. Shahar, thanks for taking the time — how are you doing today? Shahar Yamay: I'm doing great, thanks for having me. Bryan Fields: And diving in — Kellan, how are you doing? Kellan: Doing really well, really excited to dive into manufacturing — one of our favorite topics. How are you, Bryan? Bryan Fields: Yeah, I'm stoked. I think this nerdy topic is one that Kellan and I are really excited to talk about. Automation is maybe one that some people listen to and kind of know what it means, but really don't. So before we get into it, Shahar, it'd be great for you to give a quick background on yourself and how you found your way to the cannabis space. Shahar Yamay: I actually joined my father about six years ago. He had this cool joint machine that he built for an Israeli manufacturer. He had this idea, this beta, and started trying to push it out. Obviously the right market is North America — Canada and the States. I'd been in high tech for years, and I know how to sell, I know how to market. So we painted one corner of the facility, got a gimbal, and started pushing a little bit of SEO. That's how we started the cannabis journey with Festi Tech. Bryan Fields: What was the first prototype like? Were you surprised? Did you know he was working on it? Were there iterations from what we saw there to now? Shahar Yamay: It's an entirely different machine now. In the beginning we did twisting — now we've built our own approach for doing the folded top, because that's the way to make a joint in our eyes. The twisting top was slower, the insert was different, the software was entirely different. I think we grew with the industry, because the industry changes every minute. We grew with the industry, we grew with the papers, we grew with the material, and today we're in a good place. Bryan Fields: Is that growth element about working with customers who say, "this works for us, but this isn't necessary"? Is it an iterative process of hearing customer feedback, recognizing there are opportunities to improve, and going back to the drawing board to make a difference? Shahar Yamay: A hundred percent. I wouldn't know anything without the clients. I think the number one characteristic you need as an engineer — and I'm not an engineer, I'm kind of the funnel to the engineering, because I'm hearing the clientele and taking that data to the engineers — the best quality for an engineer is to be open-minded and not be in love with your product. Because when you're in love with your product and you don't listen — I mean, take the twisting versus the folding. The market wanted folded, so we went folding. At one point the market decided they wanted infused joints, and we realized vibration from the bottom wasn't going to be good enough, so we added a top-packing rod. All kinds of stuff like that. Now the market wants glass tips — so let's do glass tips. Bryan Fields: What's the iterative process, right? Let's say a customer comes back with an idea — it can't just be "let's get into the lab and make it work." Is there a conversation around what's possible, what can be R&D? I think that internal process is kind of the magic — people outside the space don't really understand how complicated and nuanced these decisions are, because every step along the way there are compromises. If you take one step to the right, maybe you need additional software, maybe you have to redesign. If you take a step to the left, maybe you can't fulfill something — but you should be able to test and consider those things. So how does that work from a roadmap standpoint? Shahar Yamay: From a roadmap standpoint, your biggest bottleneck is production versus R&D, because you're limited. We're a decent-sized company, but we're always limited, because the same engineers pushing the new batch to make sure it's aligned with what we've already committed to are the same engineers who are going to work on R&D for something futuristic — like making sure the air goes through the middle of the joint and not through the side, or building the auto-tube, a fully automated tube-loading machine. So from a roadmap standpoint, we understand our capabilities and we allocate — let's say 20% of our engineering and manufacturing toward R&D, sometimes 40%. We try to hit milestones and try not to push them, but sometimes you need to push a milestone forward a little. The good thing about this market is you always have MJBizCon at the end of the year, so if you want to push something big, you need to be ready — alpha, then beta, then testing — heading toward MJBizCon. The bigger changes, and you can see this across all the automation companies in the industry, come out around MJBizCon, because when you're producing machines for manufacturing companies, you need to show them what you're making. You can't just say "believe me, it works" — you need to show that it works. So that's how we plan our R&D roadmap. We're ready with something entirely new around mid-October, and then we can push it out in December. Kellan: On working with cannabis directly — do you guys hold a cannabis license, or do you have specific partners you work with when you're exposing the equipment to those final stages of testing, where it's interacting with the actual plant material? Shahar Yamay: Here in the States you have hemp, and if you get good hemp, it really behaves like cannabis. Kellan: That's what we've heard — THCa flower too. Bryan Fields: Same, same. But different. Shahar Yamay: THCa flower, yeah. But think about it — in Israel it's not a legal market, so when we're testing, we're using extracted cannabis, which is much lighter than flower and doesn't have all the oils to it. There are adjustments that need to happen between here and there, but I think we've gotten so much better at it because we've done it a few times now. We've been working on this product since 2014, so we've been through a lot of rounds of R&D. Bryan Fields: I imagine there are a ton of failures along the way too, where you think something's going to be awesome, you tell the engineers, they say it'll take six weeks, and then it takes twelve. You're asking where it is, they say they need more time, and then you finally see it and it's not what you expected. Is there anything early in the process that you thought would be an awesome part of the product and it just never worked out? Shahar Yamay: Tons of it. We had this crazy tamping station where we used air to bump the joints up to the roof and let them drop back down — we thought it would be a game changer, and it went to garbage after the first test. You really need to test these things by hand to see if they work. I'm laughing saying it, because numerous ideas that were great initially just went to the garbage. They didn't work. They just didn't work. Bryan Fields: And the ideas are right in principle — physics are still physics, and there's no getting around those challenges no matter how much you'd like to bend the laws. Sometimes they're there for a reason. Shahar Yamay: Great question, by the way — I appreciate that. Kellan: What's it like balancing automation with the fact that these companies probably all behave a bit differently? Some use trim, some use whole flower, they all have their own specs, and your system automates one process. What's it like integrating into so many different types of companies — do you have to adapt constantly? Shahar Yamay: We had one case where the length of a SKU wasn't the same between a facility in New Jersey and one in Illinois — same grind, same SOP, one of the best companies in the States, great operation, I won't name them — but the flower just behaves differently. I'm not an algorithms person, I don't come from that world, but it was just different — a few millimeters longer or shorter, same weight, same grinder, same settings on paper. Kellan: Is there an understanding on both sides that this is just part of dealing with a plant? Bryan Fields: Did you know going in that you'd run into that, or did it just appear? Shahar Yamay: No, we were entirely surprised. Usually with an MSO, the biggest facility, or the one you did your initial testing on, is your first rollout, and then you migrate everything to the next facility expecting an X-millimeter joint, and it comes out longer or shorter, and the team is asking why it isn't the same. But the flower is different — that's just what it is. That's why I like this industry so much, honestly — I consider myself part of the industry now. That's what's fun about it. We're dealing with a living plant. Bryan Fields: The fun part is some days you wake up to a surprise you never knew was even possible, and you've got a new challenge to solve. Shahar Yamay: It's even on a daily basis — some batches are heavier because of moisture content, and you need to make adjustments in the morning. It can happen every morning, or every two weeks. The cool thing about our machine is that it's actually pretty simple — we've invested a lot in it, but it's simple enough that operators understand what they need to do. Surprises come from every angle, every time, and it's a fun game. Bryan Fields: What about pulling lessons from the tobacco industry? Is there overlap where you can look at some of their techniques and think, okay, maybe we can implement this today, or maybe we should pull from some of those secrets down the line? Shahar Yamay: We're always looking at every industry we work in to see what already exists. But because tobacco is a leaf and this is flower — the powdery nature of cannabis — most of tobacco's solutions just don't transfer. Some solutions come from filling systems in different food industries, especially after the flower is ground. On the packaging side, though, since we're originally a packaging company, most of the solutions you see in cannabis today — outside of joint machines, which are their own niche — come from the food business: mylar bags, blister packs. Someone did it before us. It's really about nice implementation, working within regulation, and making sure the quality is there in the equipment. Bryan Fields: What percentage of the industry do you think is actually using automation, generally speaking? Shahar Yamay: Are you talking pre-rolls specifically, or generally? Bryan Fields: Let's just talk pre-rolls. What percentage of the industry do you think is using automated pre-roll equipment versus hand-filling manually? Shahar Yamay: I think most of the industry is still hand-filling. Bryan Fields: Like 75%, you think? Shahar Yamay: I'd say so — and that's mostly the bigger companies that have moved to automation. Bryan Fields: Why the hesitation to switch? Is it fear? Is it disruptive? Is it intimidation around price, or quality? Where do you think the biggest barrier is? Shahar Yamay: I think it's two things. First, the entry point — you need quantities that justify it, because at the end of the day it's an ROI calculation. Every business is basically a big ROI calculator, especially in manufacturing. Second, this industry is so young, and the first wave of automation that came in wasn't great — us included, six years ago, we weren't there yet. So fully automated pre-roll machines carry a bit of an intimidating reputation, or people perceive them as too complicated. I've heard people call them quarter-million or half-million-dollar paperweights more times than I can count. But I don't want to say a hundred percent — our clients like working with us because we adjust, we're there, we always answer the phone, even when we're not perfect. And we're still not perfect — nobody is. Bryan Fields: In a space where you're finding these kinds of surprises, the expectation can't be perfection — it has to be a continuous-growth mindset where if there's an issue, you keep refining it, and there's an expectation that you're a trusted partner who'll be there along the way. I think the ROI angle is interesting, because it really is a numbers game — what are those numbers, and if it makes sense, it should be considered like anything else: if we spend X to get Y, does it make sense? So what size operation is too small for automation, and where's the sweet spot — the gray area where switching from manual to automatic starts to make sense? Shahar Yamay: Above 30,000 units a month, it's a no-brainer — though it really depends on a lot of variables. I'd say 30 to 50K a month is a good starting point. But if you know you're going to push a SKU and you want to get to 70K but you're only doing 10K right now, that's also a good starting point, because you need to be ready. Another entry point is when you're a facility that doesn't have anything yet, and you don't want to invest in a crew of ten people, so you start small knowing you'll grow into it over time. But to your question — the number is 30 to 50K with the right automation. Bryan Fields: That's the element I think a lot of brands sitting right around that line are weighing — it doesn't feel like it's quite enough to justify it. But once they understand the real ROI, it becomes, "maybe we should be considering this," because if you're spending X on personnel to hand-fill and you're scale-limited by headcount, once you remove that constraint and automate, the limitation is no longer personnel — it becomes unit sales. Shahar Yamay: Right, and you also need more than one operator regardless, because that's the problem with people — someone can get sick, something can happen, someone's kid doesn't want to go to school. You need backup. Automation helps a lot with that. You mentioned earlier what people expect from automation — I think people expect "iPhone" — everything perfect, everything just working. But pre-roll machines are a niche. How many fully automated pre-roll machines do you think get sold in a year — 200, 300? Probably less than that if it's fully automated. So there are always going to be hiccups, not just for us, but for the whole category, competitors included. Bryan Fields: I watched one of these machines at an event, and it's fascinating to see all the mechanisms moving — you see things happen naturally through the process that cause issues, and unless you've experienced it before, you don't know where to clean it or where the challenges are. These aren't things you can model out or forecast. Your engineers can run edge-case tests, but it really needs to be in the field, with someone using it, to find those other surprises — which is part of the ebb and flow of the industry, and of automation technology in general. Shahar Yamay: At this point we know about 90% of the cases, so we're not surprised much anymore. But teaching operators, making sure SOPs are there, making sure cleaning SOPs are there — that takes time. It's a beast. A small beast, but a beast. Bryan Fields: Has anyone ever asked you to automate the cleaning process for them? Shahar Yamay: No — and honestly, I wouldn't want to. Interesting question though. Bryan Fields: I wouldn't be shocked if someone eventually asked, "cleaning is really time-consuming and we're not doing a great job of it — could you make the machine clean itself, and just get a robot to load the product?" Shahar Yamay: Honestly, Bryan, it's not that crazy to clean — everything's open, you just need to clean it periodically. It's like teaching your kid to clean his room every day. Bryan Fields: The reason I bring it up is there are so many moving pieces inside a facility, and as teams recognize a manual step, they start asking, "could we automate this too?" That automation mindset makes people ask what else can be changed. Shahar Yamay: Who's going to clean the cleaner, right? You see what I'm saying? Then it becomes an endless loop — good business for us, but I'd try to avoid that one if I could. Bryan Fields: No, for sure — I was just curious whether you'd ever been asked, because I'd imagine people in this space wish it just cleaned itself. Shahar Yamay: We cut the papers, and in the beginning the trimmed paper would just drop and pile up. So we introduced a small vacuum — basically a Dyson built into the machine — that vacuums up whatever the scissors cut. So maybe we'll get there eventually. Maybe that's our next step. Bryan Fields: That's why, when I was looking at these machines, I was personally most fascinated by what happens with all the extra parts beyond the normal filling process — how that stuff moves through the system. Those are the areas where I think people run into challenges and need to make adjustments, and those specific nuances — the ones that probably aren't listed as headline features — are what I think really separate companies in this space. Shahar Yamay: Thank you — happy you noticed that. Bryan Fields: What about people who are fearful of replacing jobs with automation? Are those conversations still happening, or are people past that at this point? Shahar Yamay: That's a good point — I've probably been asked that a hundred times. Every operator I've asked, if they have a good team, moves them to something else. If it's a team they wanted to reduce anyway, they reduce it. But usually they plan to move people to a more manual process. The pre-roll filling process is so manual, so labor-intensive, with so many people doing it, that when you introduce a machine that reduces headcount there, it just makes sense to reassign people to the flower lines, the edible lines, or cleaning. So no, I don't think we're replacing jobs — I think we're just helping businesses do it better. Kellan: Do you see a world where it's fully automated — lights-out manufacturing, essentially? Shahar Yamay: Honestly, there are solutions for visual QC, and we have one too, but I think anyone who believes in quality wants to inspect the product before it goes to end packaging. It's not like you're filling Coca-Cola or Pepsi bottles — it's flower, and it changes between Illinois and New Jersey. Kellan: Three millimeters. Bryan Fields: That's not nothing, though. Shahar Yamay: Yeah, it's not nothing at all in a very small joint. So I think we're not there yet, and as long as that's true — for beverages, sure, lights-out makes sense. For edibles too. But anything flower-related, like joints or moon rocks, you're still going to need to visually inspect and touch it by hand. Bryan Fields: Let's say a team implements one of your automated systems — what does the training protocol look like? Does your team come in and help them get set up? Shahar Yamay: Yes. Before install, when they decide to go with us, we ask them for samples of their paper, and we ask what they're trying to achieve in length and weight, and we make sure the settings are dialed in. Then when we get to the facility — one of our techs flies in from Vegas or New York — they're with the team for five days, which is actually kind of fascinating. Install used to scare me, honestly, but it looks much better now. Day one is making sure they have power, making sure the electricity is there, making sure it's assembled in the right spot, because practically the machine can be up and running 30 minutes out of the box — but they need to be taught the SOPs, so we talk a lot on day one and get everything dialed in. Day two is about teaching all the manual modes, the settings, the recommended SOPs. On day three, they run it themselves, and we're there for days four and five just to make sure they're good. On day four, our lead tech usually tries to make them fail on purpose, to see how they troubleshoot — it's a funny day for us. By day five, they're really running it. So practically, the following week they'll still need us — say they start on a Thursday, and by the following Tuesday something's not working, they'll call our lead tech and be a little frustrated. But by the third week, it's just running. I have this story of a client who, in their second week with the machine, went to three shifts — 20,000 joints a day, a week after getting the machine. Bryan Fields: That's a serious volume. It would take an army of people forever to roll that by hand. Shahar Yamay: To hit that manually, you'd need at least three teams of ten people — so thirty people working — and with automation you need maybe three. Bryan Fields: That's where the math really matters. Like you said, it depends on your expected unit sales, because some parts of the process scale, but hand-rolling is purely a function of manpower. At some point the numbers just don't make sense — you'd need a busload of people rolling forever. That can work, but is it the best approach? And what about the variation in quality? Shahar Yamay: Exactly. Bryan Fields: I get why people feel automation isn't as "nice," and maybe that's fair to say. But there's a consistency element to being in the top tier of quality every single time, which I think is a missing piece — the artisanal approach works short term, but as the industry evolves, consistency is what brands are built on, state by state. Shahar Yamay: A hundred percent. The entire flow of our machine mimics hand-rolling, hand-stuffing — we just do it again and again, mechanically, consistently. You get better joints, especially compared to a vibrating table, because hand-filling is inconsistent. When it's inconsistent, you're eyeballing length before twisting or folding, then you have to add more, and everything gets messy. When you look at other industries — food, beverage — nobody's sitting at a vibrating table hand-filling. You see Coca-Cola just pouring precisely with the red cap on top. Bryan Fields: You don't think Coca-Cola's doing that? Shahar Yamay: That's why they need big facilities. But that ties back to the earlier question about the sweet spot for ROI — obviously a client doing 20K a day, producing 300 to 400 thousand a month, that's a no-brainer, buy two machines. But some clients can even run just five days a month — they push production for pre-rolls for a week or a week and a half, and then do other things. You get one room, the machine's on wheels — push it to the side, bring in flower for edibles, push it back, make gummies. If you're a smaller operator, that's also a smart way to go. You're using automation for flexibility, not just for volume — one or two rooms, the same four or five people doing everything, and you're good. Bryan Fields: I think it comes back to what we talked about — there are small operators for whom automation will never make sense, there are large operators who've probably already automated, and then there's this huge middle group — maybe 80% — who could consider it but are unsure: will it replace jobs, do we have the skill set, do we have the sales volume? That's where the modeling sells itself — if you think this is going to be a core part of your business going forward, run the numbers. If it makes sense, the ROI is clear. If it doesn't, you keep doing what you're doing. I think the best part of technology is that when you hit these kinds of numbers, it's about being positioned to grow with it — sometimes the challenge is that you have to invest today in order to grow into it. Shahar Yamay: There are options, and we offer some ourselves, but there are other companies that do a paper-per-joint model where the machine is never actually yours, so your ROI never really justifies it. If you're paying, say, ten cents per pre-roll and making 300,000 pre-rolls, that's $30,000 a month — over a year, that's $360,000, which is far more than the cost of our machine, and you still don't own it. So either you finance it through a third party, or if you have the capital, put it down, or ask us to help you figure it out. I just think owning the machine, with a company that's willing to give you the best service and always answer the phone, makes more sense. It sounds like I'm pitching right now, but I'm really speaking from experience here. Bryan Fields: That matters, because people already have hesitancy around automation, sometimes from being burned before. It's important to know that if there's a problem — and there likely will be, because things happen — there's someone who'll answer the phone. It's one thing to invest the money into the technology, but it's another thing entirely when you're six months in and something goes wrong. Can you get through it? Will it derail production, or will you get a quick, immediate fix from their team? Because issues do happen, and overcoming that is another real challenge in this industry. Shahar Yamay: That's the biggest challenge — earning the trust that a company will actually be there for you. Why would anyone trust that up front? Right now we're at a point where we can point prospective clients to our current clients and let them ask whether we answer the phone. I always tell people, take my personal cell number, call me directly. If you think you have a problem, we're there. Festi Tech USA, the arm that deals with cannabis, is honestly more of a cannabis company than an automation company at this point — we understand cannabis, we understand the hurdles, we understand what it means to be down when you need product moving. We really stand behind it. Bryan Fields: Are you experimenting with AI at all, or considering it? Anything you can share? Shahar Yamay: More on the visual-analysis side. It's funny you mention it — there are really two approaches to visual inspection with AI today. One is defining what a perfect joint looks like and trying to get as close to that as possible. The other is telling the AI what's bad and having it filter that out. The biggest companies doing visual inspection today are split roughly half and half between those two approaches. For joints specifically, we lean toward eliminating the bad rather than chasing a single "perfect" standard, because what's perfect changes constantly. Bryan Fields: It's super subjective, but I think using AI for QA/QC adds another layer of value, because it's essentially an inspector going through the whole process, flagging good versus bad, and then a person can manually review the flagged pile and say, these are actually fine, they're just a little ugly, or, no, this one's actually bad. Shahar Yamay: Exactly that. Bryan Fields: And think about the customer experience — every time, they're getting a beautiful, consistent product, versus one where maybe the team gets to enjoy the imperfect ones themselves. Shahar Yamay: Exactly. And the coolest part about AI is you can actually talk to it in percentages — even in the alpha and beta we're working on, you can tell it to loosen the overall quality threshold if you need more throughput, or push it as close to 100% as possible. That's where AI gets really cool in production. Bryan Fields: Any other technology or sensors you're looking to integrate alongside the AI, or is it one step at a time? Shahar Yamay: I'm honestly not sure yet how we'll take it. We're watching what's happening in other industries, like high-capacity baking, where inspection is fully visualized with cameras and lasers, so we're considering it. But it really depends on the output level — for something running around 1,500 units an hour, you don't need anything too elaborate. You just need to teach the system what's good and what's bad. That's where we are right now, but it could take us in any direction. Bryan Fields: Any announcements over the next year you can share or tease? Shahar Yamay: We're coming to Vegas with two or three more machines. We're going to present a turnkey pre-roll automation line, A to Z. I don't want to oversell it — practically what we do is build simple machines that work, and if something doesn't work, we make it work. That's just how we operate. Bryan Fields: Last question — what do most people not know about automated pre-rolls? Shahar Yamay: Most people don't know it's already solved. You just need to come grab it. Bryan Fields: Shahar, for our listeners who want to get in touch, learn more, and automate their pre-rolls — where can they find you? Shahar Yamay: They can find us at festitech.com, or just message us on LinkedIn — we're happy to answer. And I'll say, we've got a great group of people. We're a family-owned business, and everyone's always welcome. Bryan Fields: Awesome — thanks for taking the time, this was a lot of fun. Shahar Yamay: Thank you so much, guys. I had a great time. Kellan: Thanks for your time.