TRASCRIZIONE 🔗
Abelara Ascent: You're a plant manager. You've just spent six figures, maybe even a half million dollars on what you were told was going to be the foundation, the foundation of your digital transformation, the unified namespace, the single source of truth, the backbone of industry 4.0. Six months later, you walk into your office and your boss asks you a simple question. What did we run on line B yesterday? And you have to pick up the phone, you have to call your supervisor, and you have to wait for someone to dig through their clipboards because you can't answer that question. despite all the sensors, despite all the dashboards, and despite half a million dollars. And the worst thing is that when you try to figure out why you can't answer that question, when you try to look at your own data in your own database that you paid for, you don't have a password. What's up, everybody? Today, we're telling a story. And look, I need to be upfront with you. This is a fictional scenario. Capyauer Inc. is only one of three enterprise cases presented at the Prove It conference, is sandbox for demonstrating what's possible when you do industrial transformation right. But here's the thing, and this is why I wanted to make this video. Every single problem in this story is real. I've seen it, you've seen it, and your buddy who works with that food plant down the road, he's living it right now. Capyauer is a multi-site beverage bottling operation, and we're gonna walk through exactly what went wrong. why it went wrong, and how we fixed it. So by the end of this video, you're gonna understand why most unified namespace projects fail, what pilot purgatory actually looks like from the inside, the three use cases that actually deliver value, and the philosophy that separates real partners from parasites. So let's get into it. So picture this, Capybara Inc, three bottling plants that make sodas, juices, and sparkling water, the whole beverage portfolio. And like a lot of manufacturers right now, leadership came down with a simple mandate. We need visibility, we need data, and we need to modernize. So fair enough. That's good instinct. They hired an integrator, a big firm, nice slides, all the buzzwords, and the integrator sold them on the unified namespace. So now, if you're not familiar, the unified namespace, or UNS, is an architectural concept where all of your operational data flows. into a single organized structure. Every sensor, every PLC, and every piece of equipment publishing to one central broker, usually MQTT. And the promise is beautiful. One place to see everything. Connect any application, build any dashboard, ask any question. So, hearing this, Happy Hour wrote the check. The integrator deployed some edge connectors, stood up an MQTT broker, and built a few dashboards. And the data started flowing. Six months go by, and now they've got dashboards that show temperature readings and flow rates. Yes, they can even watch counters increment in real time. But
Load full transcript
when the plant manager asks, why was line B down for two hours last night? Nobody knows. When the operations director asks, are we ahead or behind on the production schedule? They check the clipboard. And when anyone asks, what's in tank three right now? Someone still has to physically walk out on the floor and look at the whiteboard. They call it a unified namespace. but what they actually built was just an expensive screensaver. Now here's where it gets infuriating. Capybara realized that they had a problem. The pilot wasn't delivering value, so they tried to fix it themselves. Maybe they could add some more context to the data. Maybe they could build some reports. Maybe they could even query the database directly and figure out what's going on. Problem one, the database was locked. The integrator deployed a database, sure. States, reasons, historical data, but Capybara doesn't even have the credentials. They literally cannot access their own data. Problem two, the applications were inaccessible. Those dashboards that got built, all the project files lived on the integrator servers. Want to modify a screen? Too bad. Problem three was that there was no documentation, no data dictionary, no architecture diagram. Nobody wrote down why the MQTT topics were structured the way they were, let alone what they mean. It was a black box. And problem four, this is the one that really kills me. Every single... Change required a billable change order. Want to add a tag? Billable. Want to rename a topic? Billable. Want to understand what you already paid for? ⁓ yeah, that's another discovery engagement. Very billable. And here's where the operations director, and I love this quote, said, I thought the whole point of a unified namespace was that it was ours, that we could see everything, connect anything, and build on it ourselves. Instead, now, we're more locked in than we even were before. Because they didn't buy a foundation. They bought a dependency. And look, I really want to be clear about something. I'm not saying that all integrators are bad. There are great ones out there, but this model where the customer pays for something and it doesn't own it, that's broken. That's not a partnership. That's taken hostages. So what actually went wrong here? Let's break it down. The existing UNS implementation had a fundamental gap. It lacked operational context. Here's what they had. Sure, they had real-time telemetry from equipment. Things like temperatures, flow rates, and pressures. They had some basic state codes, running, idle, downtime. Yeah, they did some OE calculations at the line level and a handful of MQTT topics organized by site and area. Well, that all sounds fine, right? Data's flowing, things are connected, but here's what they were missing. They had no material tracking. They didn't know what syrup was in tank three without walking to the tank. They didn't have any production context. I mean, what product did we even run on line A yesterday? Well,
better check the clipboard. Hope someone wrote it down. They had no state history. They couldn't answer how long was the filler down because somebody just had to remember. I mean, they had no work order integration. They couldn't tell if they were ahead or behind schedule without asking the supervisor. They had no inventory visibility. I mean, how many pallets did they even ship? Well, they wouldn't know until the end of day report, maybe even the next morning. And here's the thing about data. Without context, it's just noise. A temperature reading of 72 means absolutely nothing if you don't know which tank, which product, which batch, and whether 72 is good or bad for that product. Hell, you don't even know if it's 72 degrees Fahrenheit or Celsius. The UNS was publishing thousands of data points per minute, and every single one of them was useless because nobody could answer the questions. And here's the real kicker. They had three plants, three sites. Data flowing from everywhere. And the operations director said it best, again. He said that we connected three plants and all we got was three times the noise. More sites didn't mean more insights. It meant more tags, more topics, more dashboards that nobody used. And still they had no answers. All right, so we already established that capybara was stuck. They've got a half built unified namespace that doesn't really unify anything. They've got an integrator who was more interested in billable hours than actual outcomes. And they've got leadership that are still asking questions that nobody could answer. They needed help. They needed real help. And that's where we came in. Enter Abelara. Abelara came in to do an assessment and our finding was brutal but accurate. The existing unified namespace was just industrial telemetry. not a real unified namespace. It connected equipment to a broker, sure, but it didn't unify operations. There was no semantic model, no master data alignment, and no integration with business systems, and no frameworks for adding context over time. Translation? You built a pipe, but you forgot to put anything meaningful through it. So, Abelara designed the architecture, the data models, the naming conventions. We came up with an integration strategy and a roadmap. an actionable roadmap. But architecture on paper doesn't fill bottles. So for implementation, for actually building the screens, configuring the databases and wiring up the logic, we brought in a partner, one who we know understands bottling plants inside and out. A partner who lives and breathes this every day. This isn't their first time. And that's where Tupinix comes in. And together with Abelara, Tupinix established a philosophy That was the complete opposite of what the original integrator did. The philosophy was that everything we built, you own. Databases, open source. Full customer access, run your own queries. The data is yours. The MES schema, open source, documented. You could fork it, modify it, make your
own. Make it your own. The applications were built in ignition with source control. You could see every screen, every script and every binding. The architecture was fully documented with diagrams and the data models fully contextualized and published to the unified namespace with clear schemas that any tool could consume. And here is the line that everyone should hear. If you want to do this yourself, here's what we built. It's open source. Go to the repo, read the documentation, but if you still want help, give us a call. That's not vendor lock. That's a partnership. Now, before we get into the technical use cases, I want to tell you about something that happened before a single line of code was written. The Tupinix team, along with Abelara, went down to the floor. We spent time with the operators, the maintenance staff, the supervisors, and we discovered something very interesting. The four vats in the mix room, they weren't actually called vats one, two, three, and four, at least not by the operators. The operators called them Jeff, Raymond, Billy, and Bob. They gave them human names. And now, I know that sounds funny, it might be, but it's also a symptom of something much deeper. The operators had built their own manual workarounds, their own naming system, their own tracking method, and their own way of making sense of the chaos, because they didn't trust the system. They had been burned before. So, the first phase of implementation wasn't technical, it was change management. It was building trust. Daily standups where operators could voice frustration without fear. Workshops where supervisors and engineers admitted that they didn't know. Conversations that surfaced real problems. Not just the ones of the RFP, but the ones that actually kept people up at night. And only after that foundation was laid did the technical work begin. Because here's the truth. You can have the best architecture in the world, the cleanest data models, the most elegant dashboard, but if the people on the floor don't trust it, they won't use it. They don't believe in it. And when that happens, it doesn't matter how good it is. You have built nothing. All right, let's talk about what actually got built. Three use cases, each one designed to add context to that raw data that was already flowing through the UNS. And that's when we get to use case one, plant floor visibility. The plant manager's complaint was simple. I should be able to walk into my office at 7 a.m. and know what happened overnight. But right now, I have to make three phone calls and wait and hope. for someone to pull a report. The existing unified namespace was publishing state codes. Yeah, but it was running idle downtime, et cetera. But nobody was logging those states. That data existed for just a few seconds and then vanished. So, Tupinix with Abelara deployed what we called an MES stub, a lightweight data collection layer that captures and stores operational events. Every state change gets recorded
with a timestamp, duration, asset ID, context, you name it. You get timeline visualizations and NAT style charts showing what each line did, downtime tracking with reasons code and shift based reporting. And here's where it gets good. During implementation, they discovered that all three washers were returning unknown, an unknown state for years before and after this UNS pilot. But now that we were deeper in here, Talking to the operator, seeing what was going on, we could tell that it wasn't a sensor failure. It was a mapping error in the legacy PLC logic that nobody had bothered to fix because nobody could see it. Tupinex traced the latter logic, fixed the mapping, and instantly, 15 % of unexplained downtime became a known wash or fault category. You can't fix what you can't see. And that brings us to the second use case, in-process inventory. And here's the quote that kicked this one off. I have 6 tanks and 4 vats, but right now I couldn't tell you what's in any of them without walking out to the floor and looking at a whiteboard. The current UNS showed us temperature, flow rates and weight, but that was it. No product identification, no batch numbers and no predicted time until empty. So, together with Abellara, Jupinix builds in-process inventory layer. Tank status dashboards, material tracking, consumption analytics and predictive alerts like Tank three will be empty in 45 minutes at the current draw rate. And here's the thing, the weight transmitters on those tanks, they've been installed for years, but nobody used the data. Why? Because different syrups have different densities. 11,000 kilograms of orange mix is a different fill level than 11,000 kilograms of cola mix. Without knowing what product was in the tank, you couldn't convert weight to volume. And the flow meters, they showed instantaneous rates, but nobody was totalizing. Every scan cycle, the data disappeared. So, Tupinex built the totalizers, linked them to the product master, and suddenly they were able to answer the question of how much did we use? How much of what did we use? And that brings us to use case three, warehouse inventory. We make pallets all day, the robots stack them, and the wrappers wrap them, but then they disappear on a trucks. But, between the palletizer and the shipping dock, we lost all visibility. The UNS tracked palletizer states and cycle counts. But nothing connected those counts to what product was on the palette, what batch it came from, or where it went. So we wired the product skew from the upstream labeler to the palletizer event. Now every pallet has what they call a birth certificate, a product, a timestamp, a source line, and a case count. Now the end of day inventory. Well, before this, it was literally someone walking the staging area with a clipboard. counting pallets and squinting at labels. Discrepancies between production counts and shipping counts happened weekly, often daily, and nobody could ever explain them. But now that clipboard
is gone. All right, now let's talk results because visibility isn't the end goal. It's just a means to an end. By contextualizing the flow meter data at the filler, the team identified that line A was constantly overfilling by six milliliters per bottle. And a 500 milliliter bottle, That's just free product walking out the door. At production volumes, that was over $4,000 per week lost in just syrup alone. The visibility let maintenance recalibrate the filler valves immediately. And remember those unknown washer states? Once they were mapped correctly, with the other data, the other context, they revealed a pattern. The washers were faulting every time a specific bottle size ran. A $200 gasket replacement? eliminated almost 15 % of their unexplained downtime. And here's the thing, the ROI on this was 90 days. Not because we deployed some magical technology, not because we had some secret sauce, but because we added context to data that already existed. And more importantly, we gave it to the people who could actually do something with it. So what's the takeaway here? Three points. First, A unified namespace is not a product. You don't buy it, you don't install it, you build it, you implement it. It's an architecture. And an architecture without context is just connectivity without insight. Second, if your integrator won't give you the password to your own database, fire them. Today, not next quarter, not next year, today. You paid for that data. A real partner hands you the keys and says, call us if you need help. anyone else is holding you hostage. Third, start with the questions, not the vendor, not the platform, the questions. And yes, while your questions will be different than everybody else's, for Capyauer, those questions were, what ran yesterday, what's in our tank, and how many pallets did we make? If your system can't answer those, you didn't have a solution. You had a screen saver. So, will the original UNS pilot on Capyauer prove connectivity? What we demonstrated What Abelara architected and Tuplemex implemented proved value. Real-time plant floor visibility, in-process inventory tracking, and yes, finished goods accountability. And most importantly, we provided a system that you actually own. So that's the story that we're telling at Proovit. And like I said at the top, the company is fictional, but every single problem in this story is real. It's real for somebody. somewhere, at some plant. And if that's you, if you're stuck in pilot purgatory right now, if you've got a unified namespace that's just generating noise, if you're locked into an integrator who treats your data like leverage, there's a way out, and the first step is admitting that connectivity isn't the same as value. So, if this resonated with you, drop a comment below. We would love to hear your horror stories. What's the worst integrator lock you've ever seen? Have you experienced it firsthand? Hit subscribe if you want more content like this and we'll see you at Proofit.