Skip to main content

Manufacturing and Transportation Industry Forum | Sponsored by: Deloitte and AWS

Summary

  • Smart factory transformation: Applied Materials and other manufacturers use Databricks to power predictive maintenance, quality control and production optimization—cutting downtime and waste.
  • Autonomous system development: Automotive and transportation leaders deploy generative AI to revolutionize product design and accelerate time-to-market for autonomous vehicles and smart logistics.
  • Airport and airline partnership: Heathrow and Virgin Atlantic demonstrate real-time operational data integration, improving passenger experience, flight efficiency and sustainability across air travel.

Manufacturing and Transportation Industry Forum | Sponsored by: Deloitte and AWS

Watch: Manufacturing and Transportation Industry Forum | Sponsored by: Deloitte and AWS
Talk by Andy Isenman, Head of Technology, Cloud and Data, Heathrow ; Caitlin Gordon, Manufacturing & Energy Marketing Lead, Databricks ; Dr. Andrej Levin, Partner & Managing Director, Boston Consulting Group ; Richard Masters, VP Data & AI, Virgin Atlantic Airways ; Shiv Trisal, Global Industry Leader - Manufacturing, Databricks ; Victor Dsouza, VP, Ai and Data, Applied Materials (Sponsor Speaker)
Join us for an inspiring forum showcasing how manufacturers and transportation leaders are turning today's challenges into tomorrow's opportunities. From automotive giants revolutionizing product development with generative AI to logistics providers optimizing routes for both cost and sustainability, discover how industry pioneers are reshaping the future of industrial operations. Highlighting this video is an exciting collaboration between Heathrow Airport and Virgin Atlantic, demonstrating how partnership and innovation are transforming the air travel experience. Learn how these leaders and other companies are using Databricks to tackle their most pressing challenges — from smart factory transformations to autonomous systems development — proving that the path to profitability and sustainability runs through intelligent operations.
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

How do manufacturers like Applied Materials use predictive maintenance?

By collecting sensor data from production equipment on Databricks and applying machine learning, manufacturers predict failures before they occur, minimizing unplanned downtime and maintenance costs.

How are Heathrow and Virgin Atlantic collaborating on airport operations?

This unique partnership shows how unified data systems optimize flight schedules, ground operations, and passenger experiences—improving on-time performance and sustainability metrics across terminals.

What role does generative AI play in automotive product development?

Automotive companies use generative AI on Databricks to accelerate design iterations, simulate crash tests, and optimize vehicle architectures—reducing development cycles for autonomous and traditional vehicles.

How do companies balance profitability and sustainability through data?

Data intelligence enables manufacturers and airlines to optimize routes, reduce energy consumption, and minimize waste—proving that operational efficiency and environmental responsibility are mutually reinforcing.

Full transcript

All right, welcome everybody. What's going on San Francisco? roll in it and you'll find out very very soon. Um I happen to be a musician in my other life. You know when I'm not uh working in AI software and trying to do cool things at data bricks uh I actually am a musician. So I'm a big rock and roll fan. So I try to keep uh a level of uh healthy dose of rock and roll in everything I do. So uh if it's too cheesy just you know just move on with your lives and we'll go have a beer later tonight.
But uh you know I try to incorporate rock and roll in everything I do. So first of all for this session uh who definitely didn't endorse the rock and roll version of uh the industry forums are deote and AWS. So where are my deote and AWS people? So a big round of applause without them we would not have been able to uh to hold this session and thank you for your support over the years uh one of our strongest partners of our strongest partners uh and help us innovate um for our customers.
Uh another uh set of uh really important partners that I wanted to really thank is our brickuilder partners. For people who don't know about the brickuilder program, it is one of the most successful most important things we can do at data bricks. And what does bricku builders represent? What brickuilders represents is really solutions that are built on top of data bricks gamechanging technology. It's solutions that are focused on use cases, value accreative solutions that are actually going to be used by end users, business users.
So if you're building a digital twin solution, if you're building predictive maintenance into your operations, if you're building a dynamic pricing engine, all of these players are helping, all of these partners are building the next generation of AI enabled solutions in all of these functions. Whether it's supply chain, whether it's connected customer relationships, whether it's connected products, warranty, you name it.
So big thank you to Avanard, Solable Technologies, Data Centex, Ernston Young, Fredence, and Deote who are countlessly uh building and innovating with us on a number of solutions and we're very excited uh to be deepening these partnerships so that we can bring more and more innovation for customers uh sooner. And also yesterday for some of our partners uh were able to be um recognized for their amazing contributions.
Uh there's all our partners are winners. I will just say that. But at the same time there are some who win bigger uh than others. And for them I'd like to congratulate Capgeemini who's manufacturing partner of the year. Cable technologies that's energy and utilities partners of the year. And then on the ISV side uh Aviva who I know is in the room. uh partner of the year on the manufacturing side. So, big round of applause for all of these amazing companies.
All right, so we have about uh 56 minutes together. So, we're going to go through uh some of these amazing sessions. I uh happen to have the luxury of slotting what I believe uh some of the most amazing voices in our industry for AI and how do you actually lead with AI not just the technology aspect of it but actually making things building things uh that are adoptable by the business and there's more than just technology in that.
So we'll cover a lot of ground today. So I have with me um who's joining us from applied materials who'll be talking about um what it means to accelerate AI innovation in semiconductors and high-tech. I'll have uh Dr. uh from BCG and BCGX uh talking about how they're building uh the next generation of pricing and sales uh AI solutions uh using data bricks and how that's helping customers in the automotive industry. And I also have uh the unique pleasure which of working with my two favorite Brits uh which is Dr. which is Richard Masters from Virgin Atlantic as well as uh from Heathrow Airport which are going to talk to us about how delta sharing is really transforming this space of aviation data collaboration.
And it's cool because it's something that you and I everybody in this audience uh if you fly frequently or if you fly at all uh you actually observe the impact of data collaboration in the industry. So this will be really cool use case that we'll be sharing today. So uh I did talk about rock and roll. Uh so the topic of this that the driving theme behind this industry forum is really one of relentless innovation. And one such relentless innovator is a guitar player that I hold in the highest regard.
Somebody who's influenced me and countless other musicians around the world. That guy's name is. Any Eddie Van Halen fans out here? So Eddie is one of those relentless innovators that truly pushed the boundaries of what was possible with the guitar. And relentless innovation like that is what sets true Titans apart. Whether it's in music, whether it's in technology or it's in manufacturing or transportation, didn't just play the guitar.
He he reinvented it. He he took it to places and the sounds he created were just not heard of before. And he built new tools and sounds that changed music forever. So Eddie could have played the same guitar or the same amplifiers as everybody else. You he could have shown he he was not satisfied with what was off the shelf. So everybody in time was using Gibson Les Paul. So for the guitar nerds, taking a Gibson Les Paul and plugging it into a amp, you know, pretty standard.
You have a Stratcaster and you plug it into a Fender amp. Pretty standard again. And a lot of legendary music came out of those those tools. They are amazing tools in their own right. But he didn't settle for what was offtheshelf. He wanted a sound that was unmistakably his. And he tinkered. He tinkered relentlessly. He built what is called the Franken Strat, which uh looks something like this. It is uh not not the prettiest guitar on the planet.
It has a few things missing. But this guitar was made by combining parts from many different guitars of the generation and it created a unique sound. But then he went a step further. That thing underneath the pickups right there is what is called a tremolo system. It's called a floating tremolo system. He actually patented that in his guitars. So it allowed you to bend notes beyond octaves that you could have ever thought.
And he patented this. And he was the first person to actually go do that. And these inventions enabled him to truly push the boundaries of what guitar playing was about and created sounds that no one else could replicate. It was truly an eruption of sound and a new benchmark of music. And he also created what is known as the 5150 amp. So he didn't stop at the guitar. He kept tinkering. He kept tinkering with amplifiers, different tones, and he created something that was unmistakably his.
So when you hear any song by, you immediately instantly recognize that's playing. You can never tell. Nobody else can replicate that. And Eddie didn't stop at hardware. The EVH or the gear brand and the 5150 amp just became were not just products. They became platforms. They got re-imagined as software adopted by generations of musicians and integrated into digital music production tools. This opened up a whole new ecosystem allowing others to build on his innovation and shape new sounds in industry.
I feel titans, the transformation agents in in each of these industries, whether it's in manufacturing or transportation or anything in between, they're making the same shift. They're no longer content in selling a product with an episodic relationship. They're continuously moving to as a service models that are actually pivoting and adapting to customers needs in real time. and they're embracing services, they're embracing interconnected ecosystems and digital engineering at scale.
So in this AI gold rush, companies fundamentally have the same choice in my opinion as did back when he was playing. What's offtheshelf? Offtheshelf is what is general intelligence systems built on the vast data on the internet. And if I had to take an analogy, if the data on the internet were a library, then it's safe to say that everybody is reading from the same pile of books. But then there's the path that took. Just as gear became such a unique part of his sound, your data is the foundation for your unique competitive advantage.
And that is the essence of data intelligence. It is building AI on your organization's own data, your customer interactions, your operational history, your deep product knowledge, your proprietary processes in a way that no generic model can ever match. And just like a good rock and roll concert, this is the part where audience participation is required. What I would like you all to do is take your phones out. Get your phones out.
I know some of you are already taking pictures. Get your phones out and I'd like you to point it at that QR code over there. This will take you to a survey link. So, let me give you some context on what it's taking you to. It's taking you to a survey link. This is what is called the data intelligence business outcome map. And if you access this, this will take you to a survey link. Well, this is the part where you help us help you.
So, if you tell us what are the areas and use cases that are super important to you, we will put investment behind it to build more solutions for you alongside ourselves and our amazing partner ecosystem. So, I'd encourage you to take a few seconds and go through the QR code, go to the survey link, pick the top five use cases that come to your mind that are most important for you and your organizations to create value.
And as you do that, I'm going to introduce somebody I'm super excited about to come up on stage and do a fireside chat about what it means to accelerate innovation in semiconductors and high-tech. So, quick round of applause. All right, thanks for coming. Thanks for having me. All right, so, you are in an interesting industry. Um, so you know, semiconductors was like a huge topic when the covid pandemic was happening and there was a bunch of short you know disrupt, you know, disruptions and shortages and so on.
So you sit at the heart of semiconductor industry. I think the whole world got exposed to how important this industry is and the impact it has on your lives on our industries at large. Tell us what can you share some specific examples of how AI is changing the very nature of work at applied materials and what this means for the skills and roles at the company. Absolutely. Um, so everybody loved the kickoff, the portion.
Super amazing, right? I think there's one portion. You got it wrong, Shiv. The sequence. You set the bar so high after and then you got me on the stage, right? So maybe I think you meant your friends from Heathrow. Uh, you wanted them up here. But if you take that out of the way, uh, we we look at AI and we begin to look at AI because we've been playing in the traditional machine learning space for a while now, right?
We've been using it in our tools. we've been using to improve our services. Uh but when Gen AI hit a few years back, the first thing was is this really a tool or should we think of it as a team member as teammate, right? Because if you ask a tool a question and you don't like what the answer is, you can toss it. Is that what you would really do if you were asking a teammate that question? Hope hope most of us won't do that, right?
We would nudge it. We would guide it. We would give it that context. We would encourage it to move in a direction where the domain expertise the data asset of the company lay and give us grounded either augmentation capabilities because RPA has been around for a while but if you think of what AI workflows can do what agent AIS can do it can up the game these models now have reasoning capabilities short-term memory long-term capability the possibilities are infinite is automation the only aspect of it.
Can it augment as well? Think of a research job. We've got materials embedded in our name. We most of our new college graduates, 50% of those and above are PhDs in material science. The time and effort that goes into conducting those researches. Can we give them these research tools for them to up the game, for them to give a leg up? Think of the software engineering community, right? the code acceptance rates, the generation, the optimization, the documentation.
So huge huge impact out there as far as the augmentation is concerned. So some of the key areas that we've explored, you know, there are some really critical tools that we have the bill of materials goes 18 level deep. Post assembly, these tools take weeks, if not months to calibrate. We now have a machine learning model that significantly converts those weeks and months worth of effort down to ours with no human in the loop.
Think of some of the some of the challenges that we've got as far as different defects. As these nodes get smaller and smaller as you can print more circuits on the smallest size possible, the number of defects, the different types of defect can get really complex. and access to those back in the day if you went about the supervised learning route access to those with clearly labeled images and train an AI model those have gone significantly down but some of these capabilities we using to really go about provide our engineers in the field that capability to go look at you know really really complex defects on these dice associate the root cause to it and come up with solutions that would have taken them very very long time as example yeah So one of the things is like I think there's no shortage of good things to do in AI, right?
And you lead data and AI for applied materials, right? So like as as companies are are looking through this and it's almost a leaders it's not almost it is a leadership challenge, right? Where you have to manage the hype around AI but also have to show results and B and then you also have to manage the risk because you know some of these models are doing things that you can't explain. there are some you know you don't want your data becoming part of the public corpus like help me understand like how important is this at applied like to b to go into this eyes wide open but also be able to control the risk because you're in a unique industry where you have to balance the innovation and the risk part interesting the timing of this couldn't be better so this is the level of importance that we're giving to AI again as a teammate not as a tool in our ecosystem u I'm presenting later in the evening out little too late than I would prefer to the board of directors because they really want to in Europe because they happen to be in Europe.
So that's that's the reason why we've got to do it really late. Um but they want to understand you know as Gen AI hit the world almost 2 and a half 3 years back now uh we identified an enterprise level risk the risk of using it from a legal from an IP leak from a contamination and the security aspect of you know are we going to expose our networks are IP to external not just external but internal threats as well where access to information the threat of access to information get amplified right so we had to factor those in legal security and IP and then the flip side of it is what happens if we don't use it what are the competitive advantage I gave you a few examples earlier what happens if we don't use it can we give our tools and services that leg up right so as we were thinking about these AI solutions we fundamentally bracketed into three categories we we thought about how do we use these capabilities to make our existing services and operation erations better, faster, cheaper.
How do we give a leg up to our products and services? And ultimately, how do we go about discovering new materials or you know enabling the research of it? So we had to build these guardrails. The first thing we did is we took a super step back and said we're going to block all things which is Aai. So those folks in the room, you have gotai in your name. Please don't take this personally because there was this concern that we didn't want to be one of those guys.
You know there's this really popular case unfortunately where an engineer uploads a code and says help me optimize this code. That code became part of in your industry. Exactly. Right. We didn't want that. We didn't want our IP to be part of the public corpus out there. Right. But again we didn't want to lose the competitive advantage either. So we had to go build this, define these guardrails, define these swim lanes where then we can really accelerate to the approach that I called earlier from an automation, from an augmentation standpoint and really let our researchers, our engineers have at it to throw the data set at it and see what they can get out of it.
Interesting. So with that like the urgency from business leaders or functional leaders or engineering leaders is is only going to increase around how do I put AI in engineering, how do I put AI in customer support, how do I put AI in product support or whatever function you name it, right? So a lot of these functional leaders, these business line of business leaders are probably knocking on your door every day, can when can I have this in in my business?
So help me understand like how is like that urgency this like because we're seeing this in almost every customer and I'm sure everyone in this room is facing this in many ways because they manage data and AI. How are you dealing with you know this new set new just urgency that you've not probably seen before on AI and what strategies have you implemented to ensure alignment and business value being delivered? Yeah, I I think uh what really resonated uh to me with the slide you showed up there is how important data is.
The fact that you can now bring data together at vast scale, do it efficiently, guarantee the quality of it, right? So that's where for us the setting up of the platform moving from a legacy on premise environment to move it to the cloud where we can do this at scale. we have a 4 pabyte system for us to put all our data sets together. How does one know what's in this reservoir of data that we have? So that's really you know solutions like unity catalog that stood out for us because what we were seeing is there was a problem of plenty that was kicking in. you know they some of our users engineers developed the kid in the candy store syndrome where the moment you gave them access to all of this there was this deluge of uh engineers coming and for a lot of it was creating redundant capabilities so with the likes of unity catalog now we can sitting on the data intelligent platform not just do it at scale but really have the capability to go explore data assets get the semantic layer information about it get the metadata information about Right.
Once that is out of the way, now you've got your ML flows. That's what we're in the process of implementing. We got the AI factory which allows us to use these data sets, bounce the latest and greatest models, have the capability to experiment at pace at speed and then really using the agentic frame deploy them or create you knowledge assist and those kind of solutions. M so that's how we've gone about it to make sure foundationally we can ensure we can do that at scale govern the information put the processes around it give them the capability explore and then you know experiment rapidly and deploy at scale very interesting so tell me like of all the communities at applied materials all the functions people who are most like who's the most engaged community like functionally like who's who's been like let's just say knocking on your door the hardest Well, um I I I wouldn't call the squeaky wheel getting the most oil here, but there there were different groups.
I mean, we've got engineers spread ac all across the company, right? And you've worked with you're an engineer yourself. You know how it goes. When they see something like this, a lot of them, so you got two opposite spectrum. One is I tried it once, didn't work it, let me check it. Second is I think I can go go solve every problem that I ever encountered. I'm going to go solve it with this part particular solution that's available to us.
Right? So we've got uh our engineering community which are in the business of doing the research. Think of the corpus of white papers, experiments and everything that we have in our ecosystem for decades that we've collected. Yeah. They were search engines on top of it. We are moving out of the business of search engine and moving into an answer engine. Mhm. not just have the capability to go search mine information that lies in this vast corpus of knowledge that we've compiled and collected over years.
A lot of it which have been untapped and untouched. We now these people took it, they embraced it. Now we've got a solution which gives the engineering community a leg up and says has somebody conducted this experiment in the past what were some of the parameters that they used? That's how they've gone about thinking right very happy to report our HR community were equally excited we have solutions we've deployed uh where depending on the region you know applied is all over the map uh we've got different regions which have different policies all of those are now made available local to them they can query that in their local language which will give them specific answers which then could get tied to an agent AI to actually go fill out a leave application a relocation application and all that kind of stuff so you We've got roughly around 10 plus AI gen based solutions deployed with roughly around 13 to 15,000 users.
Wow. Okay. And how long did that take you if you don't mind me asking? Um in the past 2 and a half years the first year year and a half was building these guardrails and then we just took off. Very cool. Very cool. So what so all of these people here are like kind of your peers and you know different industries. What would be like one piece of advice for them to be successful in their roles? It can get overwhelming with all these things that is hitting us from every direction, right?
There literally is never a time when you don't see something breaking and think this could be of importance to me. I think you've got to stop trying to dip your toes in the water and take the you got to small. You got to experiment with something that you think is relevant for you. You should try it with your own data set. Roll that out to pilot users. Here's the critical portion, right? What we felt was when we went all in and we took longer time, it didn't really by the time somebody saw it, it's not what they had in mind.
Do shorter releases, roll it out to pilot users, incorporate iteratively the feedback that you're getting, and then what made generally available, then focus on adoption because we all get asked the same questions. You want us to make multi-million dollar investments. What is the ROI? The ROI is not going to come through week 1, week 2, week three. Adoption, increasing the adoption, capturing the feedback, focusing on AI literacy and adding solutions to it is when you start to see ROI coming to fruition.
Yeah. And when I met with, he had this really interesting uh comment. He's like uh AI progress is not measured in weeks or months anymore. So if you have a plan that is 18 to 24 months or 18 to 24 weeks, please shorten that to 18 to 24 days. So that was his that really stuck with me. , thank you so much for amazing insights. Thank you. Appreciate you sharing with the group here. All right, everybody. Perfect. Thank you.
All right. So, uh, now without further ado, uh, I'm going to invite somebody who's, uh, had the one of the longer commutes to get to San Francisco, and he's braved, uh, transatlantic flights. So, without further ado, I would like to invite Dr. from uh, PCG. So, glad to be here. Hope the mic works. Awesome. Um, what an event. Um, what a growth story. Um. I'm partner with BCG. And what I would love to share is, um, a bit of a challenge when it comes to AI, about how to create impact because, um, we will speak a lot about these days about how great data is, how important infrastructure is.
But we need to remind ourselves whatever we do here it is there to make corporates and the businesses more healthier in EBIT in the money they create and there's an a link we need to think of and I would love to share a bit um how we see this being an automotive BCG you know we're working on strategies we're a consultancy um we have built up over the last years um an amazing powerhouse around some of the smartest data scientists data engineers around the But again to keep and to eat our own medicine to bring this all to value I think the most important uh sorry the most important for us is how do we embed all this intelligence around data into domain knowledge.
So what is for us most important is that someone who really can code how how someone who is a top class engineer also understands deeply the automotive value chain deeply the automotive industry so he can combine whatever he thinks he is doing to what really makes sense uh to automotive clients. Over last years we were thinking about where can AI really kind of disrupt and really kind of not only be a cool gimmick to what AI is doing to automotive players but where is it it really can be visible into the P&L and the financial success of companies and we were working a lot on pricing you can imagine the last year's pricing um just recently with the whole tariff situation but also demand patterns the supply chain crisises the whole stock management the whole supply chain management a lot of those things were so disruptive to automotive that um they are really kind of differentiated some of the automotive players to be winners and some of them really become the losers of these days and AI and the way of managing this complexity is in huge important lever.
Let me share one of the most provocative um slides we have in BCG these days and let me guide you through a bit something we call a maturity curve where we stand with AI at the moment and this is us looking in the automotive industry but looking at so many clients out there and there's an anality we are facing a lot of the AI and the data projects we are seeing everybody's pushing they are not real value they they look like value they smell like value but they are not real EBIT.
Let me explain. Um we just did an self assessment of CEOs around the world and 60% of CEOs and CFOs are self assessing themsel in a category called the showman. The showman the liberty provocative is a category where you do a lot of things right. You invest into a data platform into infrastructure. You even build it for scale. You make it secure. You make it accessible. You do a lot of things right. You even create use cases or even process transformations.
You drive adoption and you adjust the status quo and you are trying to succeed. Taking an example of an big automotive OEM we have seen they have done all those things created an amazing AI platform. They kind of went through all the things around infrastructure the risk teams the work accounts. They took everybody on board. They took a large process in their company and fully transformed it. Not one use case but a real process end to end.
They transformed it with AI and they did for three months an AB test to look how 300 of the old team or old process with the teammates were operating versus the new the new process with um AI enabled. They have proven more than 50% efficiency. They were running 50% faster with the new process. And they were celebrating this in their board in the financial community. We were celebrating. Everybody was super happy about that.
It looked like impact. You had the benchmark. It smells like impact. But if you put on the glasses of the CFO, where is the money? The thing we found is yes, the process was running faster, but the free up time of the of the team were not used efficient. They were just doing other stuff. They were doing backlog. they were doing longer breaks. Nobody really took care about restructuring the freed up time. The second thing happened.
This process indeed runs significantly faster just that the organization realize it created the bottleneck somewhere else in the value chain. So it's a bit um an example but what we see a lot these days is a lot of input a lot of um effort on the data and the AI side but not everything which looks like impact is really EBET and there are many reasons for that technology maturity maturity of the organizations but one of the biggest reason is the color code of this slide this is the pant um we see and this is what this room mostly represents and what also this conference represent the tech community is on their toes.
The tech community is pushing and is having ideas and is driving all this change. What we very very strongly observe and this is in European and US thing the non- tech community is too passive. We see so many efforts being pushed by techies but then if you think of the things which will make the strength to be money the restructuring of the team the change or the deep bottleneck of the full process this is often with a mandate which the tech community doesn't have.
So at the moment one of our observations early days and things are changing fast but one of our observation is that the non- tech community is too much in a cinema seat observing challenging asking prove me that AI can work they are not leading enough actually if we look at where really AI really made it to money maker so really made it beyond all those steps this is where we have seen non- tech leaders leading the head of Bus the head of procurement the hat of function those ones who then invite attack to kind of drive change and together make this happen.
This is where we see real EBIT transformation. So AI changing the financial reality of large corporates. This is only happening when those two work together. Let me bring us back to what it means for automotive and why this is important for us to think of in the whole ecosystem of AI at the moment. So when we think about how to make into money maker with AI and automotive um and we think a bit they take a domain of sales one of the most important thing is to better understand reality and for years in the community in automotive the reality is so complex there's so many things which not perfectly are there the data is not perfectly there the understanding of the competition is not perfectly there the supply chain is super complex so a lot of efforts we have seen they have given up in conquering complexity and they went for something they were always calling smart simplicity.
Let's find a pragmatic way um to solve the problem. It didn't really fly. What came out of this whole effort of simple or simplicity was that we have seen a lot of tools being generated and they were interesting. They were insightful but given they are not solving the full complexity users and decision makers were just taking them as an input looking at them often ignoring them and then did their own thing. So the big question for us is in this new era of AI, how do we use this technology to accept the world is super complex but we are not giving up on the complexity.
We're trying to really conquer it and in automotive understanding one of the most complex thing to understand is around what is the consumer doing at the moment. What is kind of the implication for our optimal price? You see how many things are at the moment influencing price. What is the question around uh optimal stock management? So where to put how much stock, how to organize your supply chain. A lot of questions to raise.
Let me start with pricing or I will only do pricing given time constraints. But um pricing you can imagine is extremely interesting these days. Um, and especially if you think of the starting point of many of automotive automotive manufacturers, automotive OEMs, as harsh as it sounds, but they don't know the price they're selling their cars for because most of them are selling cars to dealerships and dealerships are selling them to individual customers.
And dealers can do whatever they want with the price of the car. They can use their own margin. They're using financial services to discount. They're using a trade and discount to take in your old car. you're using procedure value discounts to kind of subsidize a lease lease product or whatever you want. For OEMs, it's amazingly difficult to understand what is the actual transaction price. Um there are some markets like the US where it's more transparent, more mature, but in the rest of the world almost impossible to get to keep track.
The second um the second big thing we are missing we are seeing is um a real lack in automotive is the understanding of elasticities. The question like how a price change would impact volume change sounds pretty easy in theory but there are so many things impacting volume which have nothing to do with price. the stock situation, your brand awareness, your competitive set, like regional micro market dynamics, the performance of the dealers.
There's so many things which have nothing to do with price and distilling the real price signal and then putting this and this is the last one. Putting this in all the real world constraints, tariffs, um um CO2 emission taxes, um there are so many things which even then will kind of calibrate if it is a good or bad decision. So a perfect example of if you go only for one of these pieces and try to solve a bit you could create something which looks like impact smells like impact but it's not real impact.
So our finding is you need to really go the full whatever problem this just an example whatever problem you take you need to think it's through end to end to really prove is there money to gain are you able to influence the decision or are you just giving a nice to have nice dashboard type of insight or are you really influencing that so conquering complexity is what we were we were on this is an logic we are now built up for um uh couple of the big OEMs one of the public example we can share.
I think we we we're working for the BMW group. We have kind of launched a pricing engine in many markets for them. Um bringing this all together into one workflow um is is probably the most exciting journey we have seen over the last three years to go after. Um bringing really the complexity together of what is happening in market, what is happening in the OEM itself, how do you connect this data points and how do you do the interpretation.
So sometimes you will have KPIs turning in one direction like your stock is aging but your competitor is getting aggressive, your supply chain is drying out. There's so many conflicting signals. How to create a logic um to bring this all together. The magic here is around the data. It is not so much about the UI or UX or it's not so much about the steering process. The magic about conquering complexity lies in managing the complexity in form of data and obviously bringing together this data in an organization like the BMW group or any other OEM is not kind of located in one place.
They're serving more than 100 markets and every market is an individual small kingdom. The sales department has very different infrastructure than the supply chain departments. There are huge corporates which have huge silos in infrastructure in governance. Bring this information together that if you only look in the OEM itself, you'll be having heavily self-fulfilling prophecy type of data because whatever you get out of this OEM is somehow influenced by the decision making of this OEM in the past.
So you need to bring in the new world, the external world, the competition, other signals like bring this all together um takes so much time and once you did it you want to maintain it, you want to make it scale in 100 markets or something like that. So making those pipelines being efficient, being re scalable um is something we see in modern state-of-the-art technologies and data bricks is an extremely great example to help us to really prepare a lakehouse um in a way that it can really kind of scale fast.
I think the estimate from the team is that we can run it to 50% less effort in the overall effort just by organizing that. Second one is there's not one um machine learning model you need to do that here. There are couple of elements in machine learning we need to take into account. Maybe demand forecasting models, maybe kind of the elasticity models we were talking over and other simulation capabilities. Reinventing this over and over again, always hyperconiguring that is an extremely important task.
But having um a notebook to have it already kind of prepackaged, having it a bit more standardized to accelerate the development process is key. And if we talk about pricing, it's an so sensitive function of the uh OEM's topline being a bit wrong may kind of influence heavily the success of the company. It need to be totally enterprise grade secure. It need to be managing. Obviously we are talking about very sensitive margin data and things like that.
So it happen to be in very very strong access right management behind that and again think of that not in one small department think of that at in an organization with more than 100,000 employees and um everybody in every silo has an different kind of roles and rights management. This to orchestrate is something um which is data bricks is perfectly built for. But to conclude with the point, it is about conquering complexity, not going for the simple one, but moreover, it's how do we bring this to money?
And our learning of the last years is to really kind of think to end to end really until the point you see the money in the bank and not stopping earlier because there's very likely that if you kind of conclude the full journey, it might not conclude at all. Cool. Thanks for your attention. Have a good conference. Thank you. Clearly, this one doesn't work. Does work. Okay. So, thank you. First of all, um amazing to know that the BCG team is now building out solutions on data bricks.
So what that means for every customer is uh you get the domain expertise that PCG brings to the table with the best technology that's out there in AI development. So super cool excited to see more. Hopefully we'll do more uh of these things in in the future and be launching a lot more solutions that customers don't have to reinvent the wheel on when when you already have something that can be brought to market much faster.
Why would you recreate it? So thank you so much for. One more uh round of applause here really quick. All right. Now, my favorite, two individuals from the UK who are now sharing data with each other and are transforming what aviation data collaboration looks like. So, Rich and, time to get up on stage. we go. So, um first of all, thank you to data bricks for letting us um come and share this story today. This is a a combination of um two pieces of work we pulled together and it's really talking about where the future of travel is going to go to.
So um I'm the um uh head of cloud and technology but also moving into a product role at Heathrow and um Yeah, I'm the VP of data and AI at Virgin Atlantic. Starting with me, starting so a bunch of stats, lots of stats up there on on us at Virgin. It just gives you a sense of who we are as an airline and hopefully people are familiar with the brand with us an airline. Hopefully people have experienced it as well. Um, but we're long haul only.
So we we have our hub in in Heathrow and we fly out of Manchester and Edinburgh as well. But we connect the states with the east as well. And a big thing I want to just focus in here is this this connecting passengers side of things. So we're collaborating with Heathrow and collecting passengers is is a big part of that big part of our uh of who we serve. Um, and just another stat on there is around our size. We got 45 aircraft, so people might not be too familiar with that, but we're relatively small for an airline as well.
So, making those experiences really key is is is important to us. Cool. And a little bit about Heathrow. So, um, we're actually 84 million passengers last year, which was 5 million before um more than we were at 2019. So, uh, not only have we, um, gone through CO, but our passenger numbers are higher now. um than they've ever been before. Um 46,000 flights a year um of which Virgin's our second biggest carrier. Um we are a 3.
6 billion uh revenue organization and we have 75,000 people working at Heathrow keeping the place running. Um Heathrow is 4. 7 square miles. For those that are British in the audience, that's the size of Winchester. Um so small city. Um and we have all sorts of different weird and wonderful things at Heathro. So uh two fire stations. Uh we have an animal reception center. So all um Marati uh raceh horses and polo horses that will ever compete in the UK will fly through our airport at any given time.
Um and we have a two rivers um which we had to redvert around um terminal 5 when we built it. um and making sure all of the ecology and everything in place was still in place to make sure those rivers who flow into the temps uh still exist. So I'm going to show you next um a little bit of a video um if I can find out where the button is for it to work. And this video shows um uh Heathrow through data um from when we started collecting data in 1991 through to um uh the end of 2023 when this video was created.
And what you'll see um when we run this is within our data you can see global change. So let me run that for you now. This was always going to be a risk. Which one we on? How do we get it to play? One second. Flost the mouse. That one. No, we go back one. Go back one. Yeah. And then go. Here we go. So, this is moving now. So, what you can see right at the top 1991 is France was our biggest uh thing and it's shooting back 2425.
This is the opening of the channel tunnel and Euro Star. we start seeing the USA grow um and grow quite quickly as a market. Um in a second you'll see the impact of 911 when we get to 2001. So this is the shrinking of global aviation traffic because of terrorism. Um there's a small glitch here with liquid bands and laptop bands. We grow back again for a bit. Um at this time we're starting to see United Arab Emirates grow as a um as a destination. uh at 2007 we get to 3 million passengers uh for the US and then um the recession comes from the credit cunch we can understand where we are now just in a second we'll see a small glitch where we go up this is the London 2012 Olympics and the growth of passengers which eventually does tip us over 3 million but we can see um the United Arab Emirates um impact on um investment in aviation here um at this point we think that United Arab Emirates might overtake USA as a destination.
But what we're going to see in a minute is the impact to the aviation sector for COVID. And this is the impact it had on all of our businesses and where we use data was this growth all the way back and this is where data bricks has been supporting us as an airport. So to go from 19 million all the way back to about 4 million and then back to 83 million is the um is the journey that we've gone through and all of our forecasting is run on the dice bricks platform.
Brilliant. So I I love that I love that graphic. It's it's yeah can't compete with that. I've not got a slide quite as good as that in any of my decks actually anyway. It's it's really fantastic use. Yeah. Yeah. Thanks. I I'll take that on. Um so look for us at Verge Atlantic, as I talked about earlier, connecting passengers are a big part of us. Um I've talked in some other forums today around what we've done around building our customer identity using data bricks and our partnership with Emperity.
Um and understanding more and more about the types of customers that are that are traveling on us, that full plane aspect, whether you're part of our flying club or not. Hopefully you are. um and what what we can do with that data to understand more about how we can reduce the seams on those journeys. How do we know what's going on with who is who is connecting? What are those connection times they've got going on? Are you actually going to are you going to be fine and make your connection?
Are you going to be really tight? So, we have to help you through the airport there or are you just going to miss it and we can deal with and helping you out during that disruption. So the seamless passenger journeys for us is is kind of really key in how we can reduce those seams with our with our um whether it's our sky team partners and our joint venture partners with Delta or with AFKL connecting onto our onto our flights or through with our airport partners as well with Heathrow.
So how we can kind of share some of those core statistics can really help with those scenes help the operational teams prepare for disruption and events during the day. So um all the optimization that's done by the airline is fantastic, but it's no good if we don't have any operational efficiency on the ground. So this is where we come in with the data and we're starting to drive how do we get better resource allocation?
How do we achieve that turnaround time safely and make sure that everyone um uh gets onto their plane safely and securely. Um, and then how do we u make sure that all of those 75,000 staff that work at Heathrow are coordinated around um the operation of our airport so that we can get that next flight out um back um to to Virgin to um lead with passenger experience. Yeah. and and that seamlessness really optimizes that travel experience for you because you know you're going to go on to your holiday or to your or your business trip as well and you're going to remember you're going to be in a in a certain mood depending on how that how that journey went how that travel experience kind of kicked off for you as well and that kind of stays through and we see it in the NPS results in the surveys at the same time.
So having that kind of through helps us keep improving that and optimizing that when you come back through on on your next trip. So here are some of the challenges that we're facing. So data integration is the first problem and we're coming from a place where um APIs take an incredibly long time to build. They're very bespoke. Everything has to be very tightly integrated. So how we work and how we integrate our data within uh between Heathrow and Virgin Atlantic, we've got a significant advantage being on the same platform um and being able to start thinking about how to share um from data bricks to data bricks and that's really important to us.
Um go on. No, I was I was just going to say the um I was probably going to skip straight onto the semantics part, but um it it really it fits with that platform alignment there with Unity catalog with having those ability to have different definitions and ability to share across there is really key for us. Um um do you want to talk about some of the tech misalignment first and then I can come back to that? Yeah, sure.
So um the challenge with the tech fragmentation fragmentation we find is where does the data privacy uh piece come in this piece. So how you know by being um unfragmented when it comes to the technology that we know that we are sending information from one place to another knowing that it's going to be received in a similar ecosystem in a similar um in a similar vein and um and then pass back to us in in a way that um we can ensure what we're receiving is is the right information to be able to act on.
Yeah. And the way we'll share it is is around aggregate statistics really. It's it's around those those overall statistics on connecting times as I said the types of passenger overall and we're not exposing sensitive information there. Um but the platform lets us make sure that's not happening as well that we can share at the right aggregation. Um we we can really kind of understand and tag the data with the right use cases around that and control that right through the stack at the same time.
And that's where as I've just mentioned earlier Unity really helps us do that sort of thing. Um and things like delta sharing and the technologies there can help us kind of split that across across platforms. Cool. So, so if I if I carry on with my semantics point next. So, sorry. This is this is really this is a this is a passion project of mine when it comes to the knowledge graph that people may hear about as as kind of a key layer here when it comes to the data connecting things through to AI and governance as well.
Um but for the airline and travel industry and with our airline part um airport partners as well, the graph is key. The way we talk about these ontologies, these taxonomies, how things relate. It's really going to help us move towards this kind of future retailing state where we really want to get to as an industry. Everyone's thinking about this across the industry. How do we make that more of a retail type experience for people so you have more choice, more ways to kind of build your basket and have these experiences?
The semantic alignment there across us all and how we really manifest that within our data sets is going to be critical to that. So that's that's that's a really key point for me to take away from this is is us all kind of trying to get our definitions right across the piece. That's me. So, um, so if we don't, if we don't get it right, then this is what this is what can happen. You get these unhappy passengers when it when it comes to the seams.
If you got a carrier carrier A connecting onto carrier B, they could be pointing the joint venture with us. They could be Virgin to Virgin. They could be um other competitors out there. We don't own the full journey. This is this is a really important point when it comes to thinking about retail and this kind of end state we want to get to with kind of breaking down a lot of these barriers in in the way you may see some of these seams when when taking a trip.
Um so actually having that collaboration in various forms you know um or um what is it um competation some someone coined a term earlier around how you can have competitors still kind of collaborating with data in the state. can do that through clean rooms, through ways of just looking at these statistics that can help and have Heathrow in the middle there helping us kind of all collaborate and give to get in terms of that data to help really kind of reduce the seams and the friction going on um from carrier to carrier.
So, how does this all come together? Well, I mean, this is where we're so excited about uh Delta sharing and being able to move that data from it because it's going to allow us to be able to um use this technology to map the entire journey. So from the arrival of the aircraft in the first place um through to the operational process of getting that landed um handing that process back to the airline to be able to uh service their passengers, provide that passenger experience um and then um and when that plane and passengers are ready to go, we can then take that information back on and we can focus on how do we get that plane uh on the way and then uh finally back to Virgin to be able to um serve their passengers on routes.
Brilliant. So this is this is some extra stats going on here. So this is where these are some great stats as well like 50% reduction in misconnections and missed baggage rates and a 16 point increase in satisfaction. So you know connecting passengers have a tough time of it sometimes, right? you you you've got so many different scenarios, so many different things that could happen and starting to kind of reduce those misconnections to increase that satisfaction and find find ways of improving your your your experience along the way is really key.
Um and we've we've got co-shares going on with various different partners I mentioned earlier as part of our Sky Team Alliance. We just announced something with Indigo as well um recently and that's going to drive a lot of great kind of connecting passengers through Heathrow on on that. So there are great all the logos up fantastic. Thank you. Um um so you know having those stats really manifest for our partners and for our connecting passengers off our alliance there is going to be really key for us.
So being able to collaborate there with Heathrow to get those stats in the right place at the right time is is really going to kind of Yeah. 16 points is is pretty pretty big. That's going to help a lot. Yeah. So what are the next steps? Um it's great that we've got this partnership. Um but this partnership needs to extend from airline to airline. needs to extend from airport to airport. Um so the um the trade bodies that exist around aviation including IATRA are a great place to be able to put standard reference data.
So we're working on minimum connect times between between airlines and airports so that we've got a standardized set of rules and then with those rules can get played out to other third parties. It might be travel agents, it might be booking sites and other areas where we can um really start sharing the data not only between the airlines but within the whole ecosystem, the community. Um I think we need to break down those silos.
We need to start talking about in terms of value, in terms of um what it means to each uh with each party. We're trying to move away from how much does data cost, how much can we sell data for and more about how we work better together in order to be able to do that. Yeah. And and getting those definitions back to the semantics point is key. It helps us have that common language. Let's explain it to our other stakeholders in our organizations as well.
When you talk about some of this data sharing, people get a bit nervous saying like, oh, we can't share individual customer data at this level. We're not doing that. We're sharing statistics. We're sharing aggregate views that you can still take a lot of action off the back of. And I think kind of bringing everyone along for that as it manifests in your own data and your own insight within your organizations helps you then talk about it when you want to collaborate with others and communicate that real value.
Um it's it's a it's one of those ones where without the tech being so kind of seamless and easy to do this now it it made it even more of a challenge because you might get someone convinced around it in terms of oh this might be the right thing to do but then that tech challenge of being able to then implement it means you've lost that audience again because someone else has come in or a new group's coming a whole new process has changed but we can just turn it on right we we can create those aggregate views we can start to share those it's now the story it's now that value piece to talk about how you can think about this differently yeah and that's what you're going to for Heathrow and from Virgin over the next year or so be a deeper collaboration really using the data bricks technology to be able to share that data between us in a seamless and trusted way and um hopefully that will continue driving our um our value to our passengers up because they're uh they're both our passengers.
That's us. There we go. All right. Uh, thanks again, Rich and. Quick round of applause. One more time. All right. We're almost at the end. And, uh, I believe you guys were missing something before we leave, so I'm going to bring it up for you guys. I think I went through this pretty quick. So the people who couldn't scan the QR codes, apparently I was very quick with this. So people who couldn't scan the QR codes can now do so.
So again, just to remind all of you, help us help you. These are the kind of solutions we want to bring to market alongside our partner ecosystem. But without uh without further ado, thanks so much for joining us today. That's a wrap and see you outside at the reception. Thank you very much. Just look for the Rivian truck right there.

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.