Signal or Noise?

Hey, Vic. Hey, Vinay. Good to see you. Good to see you. Haven't it's a minute. It has been, like, thirty minutes. Yeah. Exactly. Yeah. I feel like wish we had more opportunity to chat like this, but it's good to see you. Yeah. Welcome everyone. If you're just joining us, glad you can make it out to our webinar here. We got Viktor Brains and then you're stuck with me, Vinay, on technology. Going to jump in here in just a second. But, you know, as you're all getting situated, just wanted to let you know, you can pop any questions you have in the chat. We have a moderator. They'll be watching that and trying to get back to you. But if we don't get back to all of you, we will reach out and follow-up after the event. Okay. With that, I'm just gonna jump in. Let's do it. So we'll do quick introductions. My name is Vinay Kumar. I lead the technology organization. I'm chief technology officer here at Veryon. I've been with Veryon for seven years now. So quite some time, I actually started on the original defect analysis product and moved on and now overseeing all of Veryon's products. But defect analysis is definitely near and dear to my heart and I've seen it evolve over the years. You know, aviation industry, I've been in different industries serving and running technology organizations, but, you know, never found an industry like aviation where people are just so passionate. And being able to fuse what I believe is my passion for technology and Vic's passion for aviation together into a great solution. I'm excited to show you guys what we have to offer today and what where that has brought us. Vic, you wanna introduce yourself? Yeah. Same. Thanks so much, Vinay. My name is Victor Lee. I'm an enterprise sales engineer. Similar to Vinay, I've been here about six years with Veryon. I had a couple roles along the way. I started as a knowledge based developer working on our guided troubleshooting solutions that some of you may be familiar with. I transitioned to do a role in customer success. And during my time in customer success, I did some product support, technical support, onboarding and training for all of our commercial operators that use the defect analysis tool. My background, I'm an aircraft, licensed aircraft maintenance engineer by trade, twenty plus years in the industry. I was based out of Toronto Pearson Airport here in Canada. I hold multiple endorsements on wide body, narrow body Boeing and Airbus aircraft. So yeah, really familiar with the troubleshooting process, especially when it comes to repetitive defects. And so yeah, definitely share a passion for aviation and coming from the world, feel like I'm uniquely positioned to talk to you all today join Vinay here on this call, on this webinar, and just to talk through our solutions that are gonna solve these problems for you. So yeah, thanks Vinay. Okay. We're gonna throw up a quick poll here for all of you. Vic, you wanna kick us off on this? Yeah, let's do that. So our question today, and it's gonna pop up to just ask you to click and answer, but do you have a system or solution in place today that effectively tracks reoccurring defects across your fleet? So that could be anything that's native to your defect management system you have already. Maybe it's someone who's doing it manually through Excel sheets. And primarily we see this from operators. It's usually done by ATA record or ATA code. And so, just be curious if you have something in place today at your operation. Once we have some answers, I think we'll show those to you in a moment. Great. And you should see that poll there. I can see a bunch of the results responses coming in right now. So make sure I just go and check the webinar. You should see the poll pop up and just select your answer. We'd love to hear from you. Alright. So, you know, we're here today to discuss the problem. What is the problem? Really, again, it is tracking these repetitive discrepancies that are happening on your fleet, being able to see the trends, what is causing us the most operational impact. And this happens in a number of ways, but it's pretty standard within, you know, the troubleshooting and line maintenance world where we'll have a discrepancy. All aircrafts are going to face these. And the first time, usually the standard operating procedure is to carry out a bite check, carry an operational check. And if we don't have any, you know, resulting faults associated with that discrepancy, we're gonna sign it off as no fault found or bite check carried out, ops check cleared, the aircraft is gonna fly. And then, you know, you may have a repeat of this issue and it comes back. And again, the pilot may have a flight deck effect or, you know, they'll they'll transcribe it into the logbook. And again, on the second occurrence, depending on the issue itself, we may say, you know, do another additional ops check or bite check, no fault found, the aircraft goes into service one. And it's on this third occurrence, a bit of an industry standard that three strikes you're out, right? That's when we consider it to be a repetitive defect, a repetitive issue. And now we need to make a decision and a work action against this problem. Should we defer this problem? Should we change a part? There's going to be an action plan that's created either by the technician, you know, in real time on the line, maintenance control will be involved but the problem is not everybody necessarily has a system that is able to track these effectively and prevent them from growing, right, and getting larger. And I think this is the thing, it's that the deferred becomes the default. So we defer it, we push it to the next station, we push it to the next day, we push it to the next check, right? And so if you want to switch the slides there, Vina is going to go to the next one. It's how do we make a decision based on the pattern we see? You know, Have we changed parts? I definitely hear with a lot of our customers, we want to reduce the no fault found, right? The preference would be we need to address the problem, but not no fault found because if these grow beyond three or even two, you know, and it gets into that fourth occurrence, the fifth occurrence, the sixth occurrence, this is where we really start to see the impact, the operational impact and the costs really grow. You'll see the delays, the cancellations. When we look at the data across our customer base, it's really you can see that that's where those delays and cancellations are really hitting these operators the most is when it grows beyond that. So yeah, again, so what failed fixes cost? You know, we have scheduled disruptions, delays and cancellations, as I mentioned. Those are huge. We all are trying to avoid the AOGs. You know, they are big, big, one of the biggest cost drivers, of course. Unscheduled maintenance, you know, if we aren't tracking these effectively or not seeing them, they're going to cause unscheduled delays. And if you can't and you don't know necessarily that something is repetitive, you don't have the opportunity to make that operational decision not to send an aircraft to a place where you don't have support. And that's going to be the biggest killer. You don't have the parts there. You don't have the people there. And so that's unscheduled, and those are gonna cause a lot of problems. And then last, but certainly not least in the industry, parts. Parts and warranty spend and rogue parts, massive problem. You know, it's there's nothing worse than replacing a part to solve an issue to only have it fail the next day or the next flight and it comes down to the components. So being able to track that is a massive, massive part of this in reducing costs in the industry. So how do we fix it? Just at a very high level, you're not familiar with Defect Analysis today, we're capturing these discrepancies intelligently, we're grouping these defects based on natural language by the description, the corrective actions of those discrepancies, you know, looking for those keywords that really combine these defects together. Surface it automatically so you're not having to manually go in and search through this for yourself. The system is going to automatically bring these and highlight them in a nice, to use user interface that your team will use to validate. And again, it really comes down, let's catch these earlier. It's about being proactive. It's about having a good game plan, having all the information in front of you to make an informed decision on what we need to tackle next, what we need to do first. And, you know, some KPIs I just want to point out. This tool, you know, we have a very large customer base right now delivering real results to these operators. We hear great things every day. You know, Vinay is going to talk about this a moment, he just spent some time with them in person this week, this past week, gathering a lot of feedback and just understanding these pain points and hearing some great stories about success stories about the tool. So, you know, these are some real numbers, increase in aircraft availability. That's really what we want, want to reduce downtime. And also the reduction in unnecessary component replacements and parts expense. I mentioned that in the last slide and I think Vinay is going touch on that a little more because we were talking just prior to this and he mentioned, you know, that was one of the real key things that was resonating with the customers, wouldn't you say, today? Or Yeah. I mean, almost every time I'm out visiting customers and we talk about parts, unnecessary part removals. Yeah. Eyes light up. Everyone is going through that pain and challenge right now. And confident that our focus and shift and some of the new things that we're coming out with in our solution is is gonna be directly addressing that for sure. That's great. Excited. Okay. Speaking of which, I wanna start this by saying if you haven't seen some of our latest blogs or press releases, go check them out on Veryon's website, veryon.com. But so if you're not aware, we just launched our next generation defect analysis. And so this is really big for us. It's a long time coming. I mean, you saw all the value in the previous slides that Vic was talking about that people have been getting from our product for years. And, you know, it's been really great, but we have really been trying to figure out with all of these latest advancements in technology and especially in the AI space, how can we harness that to build features faster and build more important valuable features that customers have been asking for or needed for a long time? And we really see this evolution of next generation defect analysis as taking our product from what it is and the great value that it's brought really to the next level where we can really start to become game changing around some of these stats that you saw on the previous slide and, you know, improve make sure that these aircraft continue to stay in the air and improve how we manage and monitor parts. And so I'm gonna jump into because I've previewed you a little bit about AI, but I wanna talk about AI and how we we see it here at Veryon, especially in our defect analysis product. I mean, you have probably been out there and hearing all the AI buzzwords. I'm gonna talk about one of those buzzwords, hallucinations. And I feel like this is gonna help people understand a little bit. A lot of people talk about hallucinations, and I I feel like the term gets used inappropriately many times. People say, hey, the AI gave me some weird result back. Must be a hallucination. But is it because the AI model wasn't working well, or is it because you didn't provide the right prompt or the right context to the AI to give you an accurate result of what you're really expecting? Could be anything. Right? And so here at Veryon, what we've been figuring out is, you know, generative AI is really great. It is super powerful as a language tool as well, and we know a lot of these log books are written in English, and so as a language tool, it's phenomenal. But the unpredictability of it is actually also a challenge as you start to look at this regulated space that we're in. And so what we've seen is, hey, we've been building these large these models, these machine learning models for fifteen years, right? Over fifteen years. And so how can we take that and use that to provide this context and save space for the AI to make sure that we are still providing results that will adhere to compliance standards, regulatory standards in this highly regulated environment. The other piece of this is we are sitting on a dataset which we've dubbed our Veryon's verified dataset that nobody else has in the industry. We have a dataset that is all of these user actions that have happened in our system, human in the loop for over fifteen years. When someone at all the top airlines, now in business aviation as well, are looking at these defects and saying, no, these are a chronic together or no, these should these should be split. These should not be a chronic. These should be two separate issues. And so all of those user actions were real humans were verifying and contributing to what makes up a chronic and what do people what are people interested in seeing. We have that huge dataset that's verified by experts like yourself that we can use as the foundation to this platform as well. And what we're seeing now, continuing that feedback loop is, you know, better detection, you know, we're we are surfacing higher higher severity chronic that we weren't before. We're attracting the right severity to them. We're get definitely reducing out of service events, aircraft out of service events. And we're also shifting the way your teams work from, you know, trying to figure out what the story is and doing a ton of research. How can we make you do like, help you do less research and focus more on reviewing research or options or ideas that the AI has provided and just reviewing that and saying, yeah, this makes sense. Let's let's poke down this rabbit hole a little bit more. Right? And shifting the way your teams work as well is what we're getting out of this. Yeah. I think I've heard, Vinay, like, that's really resonated with customers that I've spoken to already where, yeah, you nailed it on the head. It's like, where do I need to go? Where do I need to look? Like the validation process, the looking historically beyond thresholds, clustering thresholds, what you would consider to be a repeat issue, you know, drawing attention to, hey, you know, you do have a repetitive issue, but there actually was something similar recently, but it's just outside that threshold that we consider for a repeat. And so it's really just infusing that AI to help them do that job, help them do that review process. It's pretty exciting. So I'm gonna jump in and we're gonna show you some of these new features that are coming out as that are all part of they're out as part of next generation defect analysis. The first one is adaptive chronic detection. So here what we're we're doing a few different things. First of all, this is where we've paired now our generative AI solution with our traditional machine learning solution to be able to provide you more accurate chronics and more accurate results. And obviously raise those high critical issues. But you can see something that's popping up here. We've actually added in a second layer of defense. So the first layer of defense would say, hey, these things should be clustered together. And then when we're not fully sure and generative AI is thinking that, hey, you know, there's actually more I could do here, we pop up this suggestion. And so you can see the box here that's showing up here that's saying, hey, these are traditionally tasks that your team would do. They would go and split or merge Chronix or in our tool. Now the AI is saying, hey, these are things that I believe you would do. Why don't I just go ahead and take care of that for you? And we realized that some of these actions too, how people want to see these chronic scrooped, it could also vary by operator. Some people may be interested in certain things and others different. Now we're actually learning from these actions as you interact with the AI. And as we come and meet with our customers, we're saying, hey, the AI is actually suggesting the base chronic engine should change and update and do some of this additional work for you. And so over time, what's happening is we're learning from these, your responses to these suggestions, and we're reducing the amount of time investment you have to put in the product the longer that you use it. Exactly, exactly. I've seen some good feedback as well where, yeah, it's just by the nuance of the data, you kind of really nailed it. There's sometimes within descriptions or the corrective actions, it's not always clear, but there can be a reference that might be in there that might not have been captured because of the way it was written or something having to do with the nuance of the way the data, you as an operator capture your data within the system that this AI can really again bring that to the surface say hey, you might want to consider this and it's, yeah, it's really sped up that process. So, very interesting. The next one I'm gonna show you, we've a nice little animation here if you can see it come up. We call this Veryon Air Everywhere. If you're not familiar with Air, that is our AI Veryon products. It's sure you're seeing our chatbot interface so you can go in and ask questions and here it's asking for a history of APU bleed issues. And so you can see if I ask that question, it is going and it pulls out all that information. It's going through all of my data that I have access to and pulling that information for me, and I can get a good idea of it. Another thing that we've done in this is the ability to create graphs yourself. So we get a lot of these asks hey, can you create a dashboard for me or a graph that doesn't we're now allowing users to go in and take those AI prompts and say, hey, can you pull a graph for me on this information? And it can do it right there in the chat. I mean, I've seen some great examples of of people piloting this and using this, and they've been able to identify issues with their fleet that have traditionally taken them a very long time because now you have all of the data that's in defect analysis in one place. And hey, you've got all your parts on off, you've got all your defects and chronic, and you start asking questions. You can get some really interesting fleet level intelligence as well as some, you know, some some very good speed to learning the story is what what I've been saying. Because as I said, it's all about a lot of people that are troubleshooting these things or trying to figure out that story, and this helps them get to that that much faster. Absolutely. And we see this air everywhere, this chatbot interface, as a way that we can start to learn and get you experimenting more in AI. And as you start to find value in different prompts and things that you do, we will start to slowly move those into the actual application interface so everybody could take advantage of those things without needing to know all the prompts or what to ask. Right? You're gonna see some of that today. Yeah. That's great. Great. Nick, you played with this for a while. Right? I did. Yeah. I mean, I'm super excited about this one as well. As a troubleshooter, in in my career. Having something like this is such a time saver. It is that story. I think, know, this is a real world example. I know the one that you're talking about where we had a customer we were working with, and there's like, wow. Like, that that would have taken me hours to build that story. That's like what I do. And so to have this at my fingertips to help me build that story quickly, man, the efficiency just goes through the roof. It's, it's pretty cool. I'm pretty excited about this one. If I'm I'm big AI user myself, like, in real in real life as well, right, in in just the normal day to day. We use we're using these. I mean, I I think it just goes to show. I think a lot of us probably who are on this call today, it's we're interjecting these kinds of chatbots into our lives and we can see the power of how it can really help us, right, speed things up, speed these processes up. So this is really cool to have this really layered in with your data, your defect data. Good. Hey. If I make your eyes glow like that, I know we're doing a good job here. So I love it. I love it. Okay. It does. It does. Okay. Here we go. The next one, the next big one that's out now is fault data integration. So we've actually taken fault data information. Traditionally, you'd see these in systems like Boeing solution or Skywise SHM solution. And we take that data and we actually bring it into the chronic. So as you're looking at the chronic, you're seeing all the logbook entries, but you're also seeing fault messages. So you can see that single pane of glass view that we're starting to really build out and focus on. So you can understand where faults are happening and this is eventually leading into our full closed loop solution. Like where we're seeing this going is being able to really understand the faults that happen and the logbook or maintenance events that they force that they create and being able to see those faults end and so you will know that you have actually corrected the chronic and how can we get to that closed loop faster? This is our the start of our process to do that. But first release has gone out now just visualizing and seeing all of these things together in one place. We're already hearing a lot of people love this this capability. Yeah. Yeah. That's great. The the early warning capability of this, you know, is is really great. Not having to wait for another flight deck effect or wait for flight crew or maintenance to write this up again, just knowing, hey, did the action or the component I replaced have an impact? I can know that right away, right as soon as that aircraft takes off, you know, I can see that information. Because I, yeah, I don't think we've really brought it up, you know, this may not be part of the conversation today, but we can ingest this data in real time. Like we have that capability today, so it's pretty cool. It's pretty exciting. Chronic summaries. This one is actually my favorite. Favorite. So I talked a little bit about how we're using that chat interface for some of these expert users or people who are really into going and trying to figure out the story themselves and see what AI can do for them. Again, you're now playing in this controlled environment. And so, you know, we take care of the safety, the compliance, all of that stuff. You're not going and talking to AI and now everything is going back out to the the Internet and, you know, these these AI providers are learning from your data. No. None of that's happening. This is very on secure environment where we're housing all of this stuff along with your your data. But as you're doing that in this safe environment, we're learning from it. And we're saying, that's a great prompt to ask. Why don't we pull some of that into the interface? And so chronic summaries is our first example of that. You're gonna see these AI generated summaries that, again, coming back to this whole, we can tell the story or help you get to the story faster. Traditionally, you'd be going and looking through these all these logbook events and fault data, and you'd be looking at all this stuff and trying to put, okay, what's happened here historically? Now we have AI generating this chronic summary. And so you can see the the the summary right there, the high frequency recurring fault, that section, where we've actually summarized everything that's going on over time, you can see impacts to delays and cancellations, AOG hours at a glance immediately instead of you going and adding a bunch of stuff up. And I'm gonna highlight one piece here that's really something new that I am the most excited about, and it is the beginning of our road part detection capabilities. And so we know road parts have been plaguing the industry for a long time. And so what we're showing you here in parts activity, you can see that little warning message, that's really indicating that we can see a potential rogue part is at play here. And so I'm gonna jump into the next one so you get a good idea of this. But as you're looking into this rogue parts capability, if I were to click into that, I can go and see that this is no longer a chronic across a tail. This is a chronic that's following a part, a component, and not a tail. And so you can see because Veryon's best positioned to do this, we know how to read defects and we understand defect logbook records, and so we understand that these defects being produced by this rogue part are going to also be similar across different tails. And so we're seeing that happen, and now you have a chronic of a part, and we can see, hey, this really looks like a rogue part, and, you know, if you're able to if we get to your SHOP reports, we can also show that NFF, that no fault found, was actually what shows in your in your SHOP finding. And so we we could tell you, hey. We think something's at play here. There there's likely something related to this component, and I'm hoping this is gonna save a lot of downtime. Yeah. Yeah. This is great. This is huge. I hear about this all the time, feedback. If this is someone's close to full time job internally, a lot of companies, you know, tracking down these parts and putting them into quarantine and trying to find out the history. We get asked that a lot, right? Can we see, you know, what are the issues that this particular part or this particular skill number specifically are causing, you know, across our fleet. And one last piece I'm gonna talk about quickly is this what we call our fixed agent. So what this is doing here is taking a look at and, you know, we're gonna continue to roll this out. Veryon has its de identified dataset where because you're as you become part of the Veryon family, you can learn from other operators. Obviously, can't share your each other's data, but the anonymized version of what is happening out there in terms of what are the opportunities to fix certain issues, we do we do give you access to some of that and we start to tell you now, hey, this issue that you have, you know what? First of all, it's it's probably not these two things because you already did a reset or you've already done this. But also, here's what some of the other operators are trying in your case. And you may look at this and say, oh, well, yeah. That's really interesting. I never thought about the outflow valve valve. That's a great idea. Let me let me go and jump into that solution to see if that that could be causing the issue here. But this is really gonna give you guys a jump start, especially for some of your newer technicians, you know, that that may not be aware of all the different issues that could arise and how you could resolve them. They're really learning from field experience with this new agent. Yeah, absolutely. Yeah, would agree. Just having I think access to, you know, what have others done is again, just good information to know to where to head next. Some of these things can get a bit tricky, bit sticky. You've tried a few different things. You just want to reduce that duplication, not having to do things over again, and so point you in the right direction for sure. Okay. Yeah. With that, you know, I everything on the left hand side here now, this is all what we've just launched. So this is all available in our next gen defect analysis, but we're we're going on. We're we're gonna keep building this out. And, you know, we have a lot of plans to to grow this. If I look at the next section, just so you understand where are we taking this next? I wanna talk about these just a a tad. We are gonna start blowing up or double clicking into these fleet based insights. So, you know, not just just on a specific tail level, but how are these issues reoccurring across my three seven fleet or whatever it is. Right? And so looking at that fleet analytics is definitely one area we're gonna double click on and take this further. Another area is component analysis. And I we talked a little bit about rogue parts, but think about understanding the life cycle of your parts, what defects they've caused across your fleets. And so today, have this chronic view where you see all your tails. In the future, now we're expanding that to have a component view. And we can really deep dive into components and their and their life cycles. And then we are gonna the other thing I'm I'm really excited because we started building this is enhanced alerts. So all of these features that you see here are all great, but then we start asking the question, well, why can't I alert on this? Hey, I don't want to wait to have a chronic generated. I want to know if you think there is a rogue part today and it's not maybe not even a chronic on on my tail currently. I want to know about it right away. So on the first time. So first, second time, whatever it is, I wanna be alerted on this. So we're gonna start making it super easy to create alerts. You're in a screen, you're talking to the AI, you can say, hey, can you create me an alert here? Or you're looking at something in a chronic summary and you're saying, hey, you know what? I wanna create an alert based on if you see some something like this again. You shouldn't have to understand complicated logic. You should just be able to make that create that alert in a in a super simple way. And so we're we're gonna be expanding that that out as well. And fault based clustering, I'm not gonna get into a lot of, but it's really you've seen that fault data that we now have. We are going to improve continue to improve the accuracy of our clustering with fault information, and we'll be able to start helping eliminate some of the noise that we know exists in in in these these systems that you get to see fault faults in. And so, really identifying those nuisance messages and how we can really get to the important pieces of the story that you're trying to get to instead of seeing all the noise. Absolutely. Yeah. I feel like that's really been a philosophy of ours the whole time. Right? This is really why defect analysis was created in the first place was to get to that root cause, to reduce that noise, and we just continue that, you know, that legacy going forward with this new release. Okay. We're gonna launch one last poll as we wrap up here. So, you know, if you can go ahead, Vic, I don't know if you can see that. If you can go ahead, I'm gonna stop the sharing. Yeah, go ahead. We'll pop it up on the screen and ask that everyone take a moment to answer that question. And again, if you haven't had a chance already, really encourage you to ask a question, put it in the chat, and again, we do have a moderator standing by that will answer those questions for you, hopefully, and get to those right now. But yeah, we'll give it a moment here and then I think we'll probably sign off for May. Hey, appreciate all of taking the time to come in today and listen to Vic and I. You know, I hope you're as excited about some of this functionality that's coming out as we are and as we just know that you got a couple of really passionate folks who are excited to get this out there and learn from the industry. And, you know, this this is gonna be out there for the next little while. And as I said, we're gonna continue to grow it. And so we'd love you guys to get on board and provide your feedback and help shape this journey. Yeah. Yeah, please. I echo those, sentiments. You know, our best feedback and our best, features and releases really come from you, everyone in the industry that gives us this great feedback and really drives the initiative forward on where we want this product to go. So I'm really excited as well. Yeah, really thank you for your time today to join us. It's been great. And Vinay has been great. Thanks for joining me. And Talk again soon. See you at the next one, Vic.