Skip to main content

Smart Data, Smarter Vehicles: Building the Foundation for the Future of Transportation

Summary

  • Aviation safety through compound AI: Boeing's Jeppesen Smart NOTAMs system uses a combination of MLflow, GenAI, and Delta Sharing to process over 4.5M critical flight notices annually, serving 75% of commercial aviation.
  • Unified data ecosystem at scale: CARIAD developed a singular data platform powering Volkswagen Group's global mobility transformation across all brands and markets worldwide.
  • Data-driven transportation future: By combining smart data architecture, advanced analytics, and AI capabilities, industry pioneers demonstrate how data platforms revolutionize mobility across aviation and automotive sectors.

Smart Data, Smarter Vehicles: Building the Foundation for the Future of Transportation

Watch: Smart Data, Smarter Vehicles: Building the Foundation for the Future of Transportation
Talk by Jon Brown, Architect, Boeing ; Sascha Boerger, Head of Data Management, CARIAD SE
Join industry pioneers Boeing and CARIAD (Volkswagen Group) as they showcase how advanced data platforms are revolutionizing mobility across air and ground transportation. Boeing's Jeppesen Smart NOTAMs system demonstrates the power of compound AI in aviation safety, processing over 4.5M critical flight notices annually and serving 75% of commercial aviation through an innovative combination of MLflow, GenAI, and Delta Sharing technologies. CARIAD follows with insights into their groundbreaking Unified Data Ecosystem (UDE), the singular data platform powering Volkswagen Group's global mobility transformation across all brands and markets. Together, these leaders illustrate how smart data architecture is building the foundation for the future of transportation, from the skies to the streets.
Here's more to explore:
Databricks named a leader in the 2024 Gartner® Magic Quadrant™for Cloud DBMS: https://www.databricks.com/resources/analyst-paper/databricks-named-leader-by-gartner
An open, unified approach to your data, BI and AI workloads: https://www.databricks.com/product/databricks-sql
See all the product announcements from Data + AI Summit: https://www.databricks.com/events/dataaisummit-2025-announcements

FAQs

What is Boeing's Jeppesen Smart NOTAMs system?

Smart NOTAMs is a compound AI system that processes critical flight notices for aviation safety. It uses MLflow, GenAI, and Delta Sharing technologies to handle 4.5M notices annually and serve 75% of commercial aviation.

How does Jeppesen ensure data sharing across aviation?

The system uses Delta Sharing technology to securely share critical flight safety data with aviation partners. This enables consistent safety information across the 75% of commercial aviation it serves.

What is CARIAD's Unified Data Ecosystem?

CARIAD's UDE is a singular data platform powering Volkswagen Group's mobility transformation. It serves as the foundation for all brands and markets within the Volkswagen Group globally.

How do these platforms support transportation innovation?

By combining smart data architecture, MLflow operations, GenAI capabilities, and data sharing, these platforms enable pioneering companies to revolutionize mobility in aviation, automotive, and beyond.

Full transcript

Welcome everyone to the uh manufacturing transportation uh session. I'll be uh kicking us off here with the foundations of industrial data intelligence and then I'll be handing it over to a good friend uh John Brown from the Boeing company who talk through a NODM's application within aviation. So as we begin really want to start talking about what are some of the challenges that industrials are facing and how do we think about uh the ways data intelligence can start to address those challenges.
So as we dive in think about um all the disruptions that are occurring. So certainly in in aerospace and automotive and many businesses there's disruptions happening faster and faster than ever. Um, if you go back uh just a couple decades ago, maybe you could wake up in the morning, take a look at the newspaper, kind of scan uh the periodicals for kind of what happened, and then maybe make a phone call to go figure out what to do for the day.
Um, that world is no longer. Nowadays, there's uh globalization, supply chains are running uh across the world. Um there's disruptions happening in in one part of the world that cascade to another part of the world as uh you all seek uh new supply bases to go after, new markets to enter. Um all of these things affect the way you run your business today and the way you run your supply chains. And in order to uh take counter measures on these disruptions, you need an information picture.
So we'll touch on that in a second. But think about all these disruptions and how you handle them today and the pain that it takes uh to address those. The second biggest trend that I see as I as I talk to clients is around loss of knowledge. Many manufacturers are scared of their workforce retiring. Um during uh COVID certainly a lot of the largest population generation here in America uh kind of retired uh the baby boomers. um that certainly affected aerospace supply chains heavily and production um the markets um kind of uh changed as well.
So that disruption that occurred. So uh being able to capture these these sources of knowledge whether it's the tacet knowledge or the explicit knowledge and being able to represent that in a data system has been very very challenging especially when you think about uh the research of what this ratio of tacet to explicit actually is and some research indicates over 90% of the knowledge within your enterprise is tacet knowledge and being able to uh capture that before people walk out the door um encode it in a data system to make it explicit is is very very challenging.
But I'll tell you some ways of how you can uh impact that with data bricks. But the last one is this never- ending digital transformation that I'm sure many of you have have gone through. So not only are disruptions happening more and more frequently, but so are technology cycles. Technology cycles are being compressed. They're happening uh more frequent. They go quicker. they're measured in in in years, sometimes even months now, especially with all these AI technologies.
And so you're you're stuck kind of experimenting and trying to get these things to production, but you end up um maybe you maybe you end up getting it there to production, but it doesn't actually get a return. So, how do you think about addressing that? I'll show you a couple ways that we're thinking about here at Data Bricks to help you all um overcome this both on quality, cost, and privacy to make sure that you're not just uh experimenting with technology, but you're actually driving into the business processes that run your business.
And so, how are industrials building this new foundation? Well, um they're using the data intelligence platform, of course, and it's really those two components around unified data governance and AI that understands how you do business. And so, the first thing to do to respond to these disruptions, I kind of uh forecasted it a little bit here, but uh you have um a bunch of real-time data streams that you need access to.
So if you're operating 247 across the world, we're going to talk about NOTAMS here in a second, but you need to understand weather disruptions, you need you need weather data, you need to understand um the uh market data that's responding to these disruptions real time. Can I access certain critical materials, you need to you need to know about risk of suppliers? Are they in a part of the world under certain uh regulatory uh concerns? um all these things you need to be able to sense um bring it into your knowledge and have that so that you can make better smarter decision decisions faster.
So data bricks with uh uni catalog and delta sharing and marketplace and clean rooms uh have created foundational technologies to make this instantaneous to be able to access large data instantaneous for all of your suppliers that you need uh to be able to uh understand and operationalize. Second, you need to be able to encode that knowledge, that implicit knowledge into agent systems. So, you heard at the keynote this morning about some of the future of um agent systems here at data bicks.
And one of the ways that if you think about um testing evaluating the ai agents, um that's one way to bring this implicit knowledge into the explicit. It's much easier for an expert to evaluate and review a plan than it is to create the plan. So, being able to reason about that data, put that feedback in the system, you're providing a new pathway uh a completely new information pathway to capture was that a good uh plan that was generated by the agent or not.
And so now you can encode that knowledge, that implicit knowledge into explicit and start to have agents take over uh some of the uh loss of knowledge because they're able to automate uh the things in the business processes uh with guardrails and evaluation monitoring a full tool stack for you to be able to do that. And then lastly, you want to be able to take all these tools within the data bricks platform and put it into action.
And you want to have outcomes. So you don't want to be stuck in this digital transformation endlessly, but you actually want to think about what are the use cases that are going to drive your success. So these are things around product performance. Maybe you want to understand um how to maintain an asset that's generating energy. Um and you want to you want to make sure you bring the right equipment. Uh you have the right uh troubleshooting manual, you're taking the right safety precautions, you have a field service assistant.
Maybe you're trying to understand, I'll show you a live demo here in a second to um how a product's performing in the field, an automotive product or an airplane. Uh be able to connect all that data from engineering to uh social media feedback to uh manufacturing to capture maybe where a product escape may have occurred. in the industrial AI realm, being able to connect to your uh operational data stores like Aviva, uh bring that in, do process monitoring, natural language SPC, um digital twinning, all of these things are are now possible, but actually understand if I save an hour of downtime, how much is that worth?
Being able to really quantify that so that as you roll out that uh machine learning model at scale across uh your your manufacturing network, you know that the impact that drives in uh cost savings or cost avoidance. And then lastly on on digital supply chain things around order reallocation, be able to respond to tariff events or uh evolving trade rules that are happening every every day, every hour. Um or figuring out new new ways to generate new business or or tune your business for automated quoting things where you're engineer to order.
You're trying to understand um various constraints and be able to optimize that for what makes the most profit or is the most uh resource efficient, whatever those may be. So let's take a look into how the platform actually brings these to life. We're going to start with uh the automated quoting uh uh solution here. And you can see this is a AIBI dashboard. Um we can see we can get a quick state of the world of what the AI agent is doing.
Uh we can see how many quotes are accepted, how many are rejected. We're kind of figuring out we can tune this balance to what makes sense for our business at the time. If you think about what this looks like today in your business, you maybe have to talk to your your analysts um who are generating these quotes. They have to figure out, okay, do I have all the information? Um can I actually respond to this? They maybe look up what the true cost of a similar job was in the in the past or a similar load or or you're trying to do airline pricing.
You're doing dynamic uh pricing uh real time for consumers. So you really want to you want to be able to kind of optimize given the current information set at the time of generating that quote make the best business decision and make it quickly so you can win that work. You can win that order, win that business and accelerate your your your revenue stream. So you can you can go in here you can of course uh generate u all sorts of different charts and then you can look into details of the quotes.
We want to see kind of was a certain quote uh uh accepted. We want to evaluate this. Was this the right decision? Maybe we want to reason and think about this at the executive level of what what quote should we be accepting. You can start to make those decisions directly in the meeting. You don't have to do this another analysis review and iteration. You're able to go right in and uh evaluate and have that full information picture available at your fingertips.
If you if we look at a product performance use case um in uh electric vehicles, you want to understand where the battery was assembled. and make sure that wasn't a manufacturing defect and then take a look at how the product is performing in the field. So you can take social media posts, use AI functions to quickly get a sentiment review of this. I don't have to write a line of code. It's just able to do this um automatically because you have data intelligence built into the platform.
If we want to understand the engineering, how it's comparing across model years as we are doing R&D, we're updating the battery. It's a rapidly evolving space. So we want to see is this newest battery with the newest warranty performing the way we we saw it in the lab or has it uh regressed. We can we can also evaluate from thirdparty marketplaces uh on data bricks like acue weather be able to see is it a is it affected or correlated with environmental conditions where this vehicle is operating all these different uh information you are able to bring it together and start to ask questions and if you want to go deeper hey the information I have in the dashboard isn't uh available you can simply ask Genie so if we go back here on the quoting we can ask Genie directly built in the dashboard we don't need to build another dashboard we can ask it right here so we We can ask, hey, how many quotes were were created um over the last thing?
We we're not seeing the right business kind of come in or we think we can do more. We can very quickly get an answer. We don't have to go back to the uh to our analyst to go figure that out. We do that real time here directly during this presentation or in the meeting. So that's a little bit about uh data bricks, how we're bringing data intelligence to life and defining this with outcomes that you can directly insert into your core business process within your uh within your business.
And so really what we've spent the last bit of uh time here is this is all defined in our outcome app. So this is your guide to be able to bring these directly into your business. Each one of these use case areas are something we are building solutions around. We're we're incorporating AI agents. Uh we're incorporating um marketplace data giving you that full information picture of how you actually build out these uh solutions.
So, take a look at that QR code. And this is where uh I'll leave it and I'll turn it over to and welcome to the stage. All right. Well, imagine you're getting into the car and you're going to the grocery store. And if you're like me, I like to get in and out. I like to get there as fast as I can. I like to stick to my grocery list, hopefully not deviate, get the ice cream that I probably shouldn't be getting, and then get back home.
But now imagine that as you're getting to the car, your wife, your significant other, they give you a list or a pile of sticky notes, maybe 50 sticky notes, right? And on there, they they will say different things. It might be, you know, hey, there's birds by the road 5 miles down. Watch out. Or they might say that stoplight is out two two streets down. You don't normally go that way, but just in case, right? Or it might be that there's a road closure and you do need to change your route.
And these aren't organized. They may have acronyms or bad handwriting, whatever it is. That's what it's like to get NOTMS for pilots and dispatchers today. And so what we're going to be talking about is going from that pile of sticky notes to an organized briefing, from fragmentation to flow. But first, who am I? Uh. I'm a senior architect with Boeing Digital Aviation Services working with Jeeppes. I've been with the company for about 16 years working on a variety of projects and programs across flight planning, flight deck and charting.
And for the last year or so, I've been working with the NOA's team and we've been looking at how can we improve our data pipelines and processes. What technology insertion can we leverage uh to automate and leverage some of these cool great things like AI going forward. And so as a part of that effort u working with the team, we wanted to share with you guys some of the insights and experiences that we've had over that over that time period.
Of course, it's not just me. We have several teams uh inside Boeing and Japon supporting NOTMS. This is just a few of those. And if you tally up the number of years that we have on the Gnome's team on how long these people have been doing it, just on this page alone, you have over 100 years of experience focusing on NOTMS. So, we have people have been doing this for a long long time. 29, 25, 20 plus years is not uncommon just in this domain.
So, and also we do have several team members here today. So, if you do have questions or want to learn more, feel free to come by and talk to us afterwards. But so, what is a NOTM? Well, NOTM is a notice to airman know air mission, right? Looks like this. How many of you can read this NOTM up on Okay, I I see like two hands. Okay. I was going to say there's no test after this, but we might need one. I don't know. All right.
So uh a NOTM is essentially a mechanism that countries use to provide data that is temporary or timesensitive in nature. And it's something that's you know very important you know it's flight safety critical information that or it might be information that's going to help the uh airline operations and whatnot. And so examples might be hazards like a runway is closed might be a temporary flight restriction don't fly here.
Okay. It might include temporary outages like lighting is out or there's a nav out of service, right? It could be environmental conditions like no tams, sorry, uh, snow tams or ashtams, right? So, this is stuff that we don't want to miss. Where do they go? They go in the hands of pilots and dispatchers. We send them worldwide to airline operation centers as well as to the flight deck for our pilots. So, you know what's happening here is a dispatcher will get a NOTM at the operation center and they'll say as they're doing their flight plane calculations, they'll I make sure that if a NOTM is going to impact or change their route in some way that they'll they'll update the flight accordingly.
And then, of course, right before the pilot takes off, they're looking at this in their EFB, their electronic flight bag, and the PLC make sure there's no hazards that would uh impact their flight. And here are some examples of some close calls and uh problems we've had in the past when NOTMS were missed or not identified. Some here even in San Francisco. And so this is why NOTMS are are so important. Now the key thing to note here is that they're situational.
It's very contextual based on where you're flying, what aircraft you're flying, right? Some NOTMs may be very applicable to you but not to others. So we got to get the right NOTM to the right person at the right time. and Jeeppes as a worldwide aggregator of navigational data and NOTM information I think is an unique spot where we can really see how the industry as a whole is experiencing you know some of the problems that we have and how can we improve them how can we make sure that the pilot isn't inundated with all this data that they're seeing so what is the problem well essentially it's this this is a famous quote uh that uh we've used for many years and it's seven years old and unfortunately if you ask pilots today what they would feel about gnomes or how they would feel about gnomes it likely would not change it still feel something like this in fact this is some of the feedback that we've gotten from our customer advisory boards and as we reach out and talk to customers and you'll see that uh you know there's there's always room for improvement right uh in fact I know uh a number of pilots I've talked to over the years good personal friends and I made the mistake of telling them what I did for a living.
And of course, they immediately said, "Well, here's all the complaints I have about NOTMS, right? It was it was a little rough, you know. Um, they said, ", don't send me a list of a thousand NOTMS. , don't send me the NOTM that's 5 miles away about birds, right? That's not going to impact me." And so, so we're trying to figure out how can we do things better. And you might be a little upset, too, if you had to read this every time you went to a car.
So here's some random examples uh from some of the customers that we have. Uh unfiltered NOTMs for both domestic and international flights. But as you can see it's you know very commonly very easily could be a 100 for domestic and even up to five six even sometimes a thousand you know NOTMs if you're talking international and the volume just continues to increase. Now today in production we have anywhere from 65 to 75,000 NOTMs that are active probably about one to 1.
2 two million canceled NOTMS over a 93-day period. But what we're looking at here is the ingest of these NOTMS. And so every year we're getting more and more ingest NOTMSs that are coming from all the different ANSPs. And so that might be the FAA, might be NATS or NAV Canada, etc. And so how do we keep up with this uh increasing workload, right? Because the trend is clear. It's not going to go down. How do we find that needle in the haystack?
Well, we have some tools for that today that uh many airlines and customers use. One of them is called NMT, nom management tool. This is a mechanism that airlines can use to configure their their NOTM feed essentially to help filter and sort through those NOTms, make sure that you know they're reducing the cognitive workload that we have uh for dispatchers and pilots to filter out the noise, right? Not see every NOTM that gets delivered or would be delivered to them.
And also we're starting to visually display these NOTms. So this is a screenshot from Havveda release for FD Pro uh supports over 75% of the commercial market worldwide. And we're we're super excited about this. So this is an area where at a glance the customer can go into the app, they can see what water runways are closed, taxi ways are limited, uh those sort of things. And this is something that team's working on for a number of years.
And it's it's really exciting to be able to see it visually just at a glance. What's impacting me? So we've made good progress, right? We have these tools. We're getting there, right? It's it's pretty neat. But I think as an engineer, as an architect, as a as a technologist, I think we can do better. And so, how do we go from hundreds of nodes to single digits? How do we deliver the right nodem to the right place at the right time is the question.
Okay? And that's where we're really starting to look at data bricks. How can we optimize even further? Every nodem counts at this point in the game, right? So um some things that we've been looking at is how do we solve some of these issues that we're seeing with the data coming in on a daily basis. Okay. And you know the problems that we have with semistructured data, missing content, ambiguous data, right? They're not going to go away.
And so can we leverage data bricks to uh help predict, optimize, and plan for some of these issues that we're seeing here? Well, to do that, we're going to need we're going to need a lot of nodes. we're going to need to think about. And so that's where we started to build the hay stack. And we had today we have about 20 million um 20 yeah 20,000 NOTms in our uh sorry 20 million NOTMs in our uh data lakehouse today and more are being added every every day as we get more we're throwing them in there.
And in the past you know these were all in a Hadoop cluster kind of fitting not really being used or leveraged very much. Um, but now that it's in a data platform, it's really opened up our eyes to be able to leverage and able to easily see and reference and use this data for all the team members going forward and not only our team but also teams across all of Boeing. So what we started to do is identify and build out some of the dashboards just like was showing you up there.
In fact, one of those dashboards like the worldwide view um came just a screenshot from data bricks I was showing you earlier. And then from that we're starting to evaluate and identify how can we get a feedback loop and a mechanism to understand what's going on. Right? So today we have a group of analysts that are checking these things real time. Uh of course we have automation that carries and handles a majority of the gnomes that come in.
But for the ones that we quite can't figure out what the ambiguous data or some of those edge cases, right? They go to a a manual group, a manual queue. And so what can we learn from that, right? What can we derive from that to either help make sure it doesn't happen next time, right? uh and actually prevent that from occurring, right? Or predicting or automating future or even when as the analysts are updating the NOTMS and you know identifying what needs to change, what can we learn from their decisions to help improve this going forward.
Now, of course, it's not just helpful for our teams, it's also helpful for the rest of DAS. And that's been a really interesting endeavor um because you know with Unity catalog we've been able to share information with the click of a button and do a lot more data collaboration. Now this isn't new to us. We've been sharing information for many many years. Uh but usually it's been a conversation around APIs and here here's our API here's access to it.
Here's a swagger spec you know you can take a look but they're very opinionated. They're very focused on particular use cases. And I think the conversation has really changed as we've started to say here's the raw data right and so let's talk about your problem. what do we need to change or you know based on the data that we have how can we improve or solve the particular issue that you're having here and so that conversation of is a lot easier when you can just oh yeah here you go take a look tell me what you think and let's let's talk about it some more right in fact we had some cases where you know they were looking at the API but of course it was different or they couldn't identify you know based on this option what they had to do but when we started looking at the data you know we got we got a lot further along identifying okay this is how we can solve that problem for you.
Another example um is a co-orker of mine. We were working with uh CU Boulder and some students there on a research project and they said, "Hey, we really want to use some NOTMS and they reached out to me. Uh we had some information available in our dev workspace and they didn't even know and I didn't even know this was going on um until until they said, "Hey, can we can we leverage that?" And in the past, this doesn't sound like a big deal, but in the past, you know, it would have been probably a week, several weeks or even maybe a month or more trying to figure out who owns that data, where is it?
How can I get it? Can I get an extract from it? Right? All those things where this was just literally an IM. Yeah, let me open it up for you. Not a problem. Right? Handling it in a matter of minutes or hours instead of weeks and months. And then finally, once you have all this information, you start to understand it. that's when you can really do a lot more interesting things and look at AI, right? We're all we're all looking at AI.
And so with this time span of about three days, we did a hackathon, started playing with, you know, some machine learning, deep learning models, and we'll talk a little bit more about that in the future. But, you know, it's it's been pretty exciting to see the transition of where we're at and then where we're going over time. So, what have we done with AI and ML flow? Well, the immediate problem or one of the immediate problems that we started to ask questions about or try to understand better was about Q codes.
And so what what is a Q code? Well, it's a five character string associated to every NOTM. The first letter being a Q and then the four other characters talking about, you know, really the category of classification of that NOTM. So in this case, you have a subject group, a predicate group, and then identifiers for each of those. So if I go back to the previous example of the note that we were showing earlier, the Q code is QMXLT.
So in this way, taxiway limited 2. And this particular notem is talking about taxiways that are closed to aircraft with a wingspan of more than 118 ft. And the importance here is I think fairly obvious, but also the important thing on this NOTM in particular is that Q code is how we can filter, sort, and categorize and make sure it's going to the right place that's going to be important or pertinent to the the flight plans that are going there.
And so over a 3-day period, u we just got the team engaged. They had literally never logged into Daybreaks before, right? They were not AI experts by any means but you know just by playing with a little bit you know you know reading up on some demos and experiences we're able to you know identify what is possible in this space in a short time span and so we looked at several different classification models and techniques to identify hey can we predict a nm code for a nm and the reason why that's important is not just because of filtering but also because it doesn't always come in 100% accurate or correct right so we have what's called cases like quad where instead of getting those four characters, you just get all X's.
Or it might only have two characters instead of four. Or it might be u we get multiple NOTMs when it really, you know, it should it's it's one NOTM but there's multiple in there and they should really be divided, right? Or it might be a case where the the Q code that we have it's ambiguous or you can have multiple Q codes. So maybe we should have actually be applying more than one in this particular case, right? So how do we solve those problems?
And that's where we started to look at AI and classification models. And I think we learned some interesting concepts, interesting ideas, um, with a high degree of accuracy on leveraging just, you know, some examples like this. And the the question we've really started to ask ourselves is, okay, how can we do this even better? How can we build a compound AI data flow that leverages these things, improves the accuracy and efficiency of our data pipelines, and do so in a way that's, you know, cost- effective, right? because the last thing you want to do is send everything to a uh an LLM, right?
Generative LLM and eat up a ton of money, a ton of tokens and include a lot of that latency into your data flow. And so this is where we're looking at maybe a start we can use a basic machine learning classification models that are quick, easy, you know, cost effective. And then for the ones that we feel you know maybe have or need to have multiple classifications that's where we can look at deep learning models apply that sort of thing so that we can identify uh two or more you know may not be quite as accurate but or may take a little bit longer but at least it's still deterministic in nature and then finally you know if we figure out that those problems maybe there's other issues with the semi-structured contact uh you know or text textual information of the data right so so people are typing this information in and other times s we not be able to may not be able to parse it really well.
One example is time sheets. We actually can parse time sheets pretty pretty well. Um but there are some edge cases that you know we we have or struggle with a little bit. So in that particular case if we have a nodem and you can't parse the time sheet. All right you know you have a nodem and it's it's runway closure right as an example. But then the issue here is is if I can't parse the time sheet I just know it's effective from this date to that date.
And if if you can't read that from a machine perspective, you can't really identify this flight plan can can or cannot fly on that runway automatically, right? Someone has to manually read it. Um where in cases of a time sheet, it might only be closed for certain hours of the day. And so we don't want to limit, you know, flight plans or aircraft flying into a particular air airport or runway just because we couldn't parse the darn time sheets, right?
So once you identify these different issues right in in the data, we can start to identify the solution um based on the the possible AI models that we have available to us and create what we call a compounded AI data flow and eventually get to what I call the perfect nodem right the perfect nm that is completely machine readable right it doesn't have any ambiguity in it and it's something that we can share downstream so that any any entity, any pilot, right, can make the right decision at the right time.
And so what were the results out of the the period and some of the work that we've been doing here? Well, it kind of looked like this. Uh using data bricks, we were able to parse and clean all the information that we had, run it through these models, and with a high degree of accuracy to pre uh sorry, predict the uh the Q codes. So instead of manually sending this to the analysts every time we don't know what the answer is, you know, we can automatically say this is what we think it is and give them really a second pair of eyes, right?
Or or multiple eyes on the problem and help speed up that process and also give some sort of clarity or confidence level of what we think that Q code is. Now, of course, it's not all AI um you know that we found in value of data bricks. You know, sharing bulk data is another one pretty common, I think, across uh all of our our companies. So, how do we do that? Well, unique catalog, of course, right? And delta sharing.
But it's been a interesting conversation as we talk more and more about how to share not only the typical NOTM feeds that uh that we do share across all of our business units, but start to share more and more of that data. you start to think about other possibilities and things you can do with it, right? So talking more about NOTMS and flight planning, talk about how we can speed up and improve the processes that we're saying nodes to update our 424 data extracts.
So the data, the navigational data that's going to the cockpit and the FMC's, you know, what can we do with simulation or analytics or even how we can change the briefing package even further with the more and more data that we have available to us. So what's next? Well, this is still in development. We're still kind of figuring things out, seeing where where things are are going, but you know, we're super excited about it.
And obviously, as we've learned more and more about even, you know, the technologies of this conference, right, uh, Lake Base and, you know, all the opportunities or technologies are coming out on a regular basis, we're we're hoping to take advantage of that. And, you know, right now, I think we're at the stage where it's a pretty exciting journey for us and it's only going to get better. And so, with that, we want to share some of the lessons learned uh, from that journey. and not only what we've done but also some of the values here and I think these are pretty common right I'm sure you've seen them before I used them but it's a good reminder right um simply because we felt that this was going to be a multi-year effort uh and endeavor to go and understand this technology integrate it and you know we're we're still doing that but at the same time we've already gotten immediate value even just from our dev instance right we started to push this to UAT started to have more conversations with customers uh internal stakeholders etc so you know I think that The really lesson learned here for me was don't wait, you know, start immediately and as soon as you play with it, you're going to learn a ton right away.
So that being said, you know, what I would say is just taking it back to the beginning. So thinking about that route back to the grocery store, right? If your data becomes less fragmented, then the route will become clear and your data will flow. So thank you

Learn more about the Databricks Data and AI platform.

The information provided herein is for general informational purposes only and may not reflect the most current product capabilities or configurations.