Enterprise AI Leaders: Moving Pilots to Production at Scale

Summary

Senior leaders from Etihad Airways, Munich Re Group, and RaceTrac share what separates organizations generating measurable AI outcomes from those still building pilots, focusing on architectural decisions, organizational obstacles, and how Lakebase and Genie accelerate the path from insight to action.
Etihad Airways reduced data engineering timelines from weeks to days using agentic tooling, lowering the skills floor so analysts can build end-to-end data applications without deep engineering expertise.
RaceTrac replaced executive dashboards with Genie-powered store intelligence apps in 80 days, cutting a projected two-year rebuild timeline down to three to four months.

Enterprise AI Leaders: Moving Pilots to Production at Scale

Every enterprise has a data strategy. Far fewer have turned it into measurable, board-level performance. In this executive panel, senior leaders from Etihad Airways, Munich Re Group, and RaceTrac share what separates organizations generating real outcomes from those still building pilots. The conversation covers architectural decisions that worked, organizational obstacles they didn't anticipate, and how Lakebase and Genie accelerate the path from insight to action.
Discover how Etihad reduced data engineering timelines from weeks to days with agentyc tooling, lowering the skills floor so analysts build end-to-end applications. Learn how Munich Re implemented governance as an enabler rather than a bottleneck in regulated financial services, enabling agents while maintaining control. See how RaceTrac replaced executive dashboards with Genie-powered apps in 80 days, cutting a 2-year rebuild timeline to 3-4 months. Understand the four differentiators for moving from pilots to production: context, control, choice, and cost.

Chapters

FAQs

How did Etihad Airways modernize its enterprise data platform with Databricks?

Etihad moved to a data-as-a-product model on a cloud-native Databricks platform, which this video describes as the foundation for the airline's broader AI transformation. By adopting agentic tooling that reduced data engineering timelines from weeks to days, Etihad lowered the skills floor so analysts—not just engineers—can build end-to-end data applications.

How did RaceTrac replace its executive dashboards with Genie?

RaceTrac deployed Genie-powered store intelligence apps in 80 days, replacing static dashboards with a conversational interface that gives executives real-time access to operational data. This video describes how this approach cut a projected two-year rebuild timeline to three to four months by leveraging Databricks capabilities rather than building a custom reporting layer from scratch.

What is Munich Re's approach to governance in enterprise AI deployment?

Munich Re implemented governance as an enabler rather than a bottleneck, allowing AI agents to operate within regulated financial services while maintaining control and auditability. This video describes their approach as finding the balance between enabling autonomous action and meeting the compliance requirements of a highly regulated industry across more than two decades of IT leadership.

What are the four differentiators that separate enterprise AI leaders from those still piloting?

This video identifies context, control, choice, and cost as the four differentiators. Context means grounding AI in real business data and processes; control means maintaining governance over models and outputs; choice means preserving architectural flexibility rather than locking into a single approach; and cost means demonstrating measurable ROI that satisfies executive and CFO scrutiny.

Full transcript

[00:09] Hopefully you're all here for beyond the road map. What separates enterprise AI leaders from everyone else? If not, that's what you're going to see. Uh so So So hopefully that's what you're here for. Uh I am Craig Wiley. Uh I lead product for AI and ML here at Databricks. Uh I
[00:26] joined from Google Cloud a number of years ago where I led AI for GCP and built Vertex AI and then previous to that I was uh inside Amazon for 9 years including uh so if any Amazonians or post-Amazonians in the room, yeah, we've
[00:41] we've bled on the same soil. Uh and uh I I was there for 9 years of which the uh my that included my time as the founder of Sagemaker and kind of the builder of V1 of AI and machine learning over at AWS. Um I've got a great panel today and
[00:58] and I'll just say given this is a panel discussion, I want to apologize in advance. Because it's a panel discussion, I have to get my lines right. And I can't just improv this. And so you'll see me looking down and reading at times and apologies, that's why that's happening. But I wanted to
[01:14] introduce our panel briefly um and then we'll do more we'll do deeper introductions as each of them uh share some of their insights. So uh we'll start uh Evan Digney. Evan is the head of enterprise and data and AI platforms at Etihad Airways and was responsible
[01:31] for leading the airline's move to a modern uh cloud-native data platform on Databricks. Anna Tureba is the executive vice president and head of IT at Munich Re Group with more than two decades across IT leadership and digital transformation
[01:48] at Mu- Munich Re, Siemens, and BMW. And lastly, Francisco Macedo is the executive director of enterprise data and infrastructure at racetrack where he leads the customers the company's data platform and infrastructure teams.
[02:04] And so if everyone if I could just get a big hand for the panel that we've got here today. So this is a a peer level panel where we're going to explore what separates organizations generating real outcomes
[02:21] from those that are still building use cases and ideas and what have you, right? How do we actually translate into the real value that we want to see from these technologies? We'll get into the architectural decisions that paid off the organizational obstacles that they
[02:38] expect and how lake base and genie are speeding up the path from insights to action. So with that, we will we'll dive right in and Evan I will start with you if that's all right. So
[02:54] Etihad modernized its enterprise data platform onto a data as a product model on Azure data bricks with unity catalog as the governance backbone while the migrations continue. It runs real-time streaming on Kafka and flink for
[03:09] operations and is displacing legacy BI business unit or replacing legacy BI business unit by business unit with genie. Evan's own data and AI summit session covers the move from traditional BI to AI BI starting with a
[03:26] catering pilot then binding the binding constraint is engineering capacity not data. So we'll start with for a lot of leaders in the room the constraint on AI stopped being the data and became people. Not enough
[03:42] engineers to keep up with demand. Is that your experience at Etihad and how are you closing that gap? Hiring, low code agent tooling, reorganizing, and the skills it takes. Give us a sense of how that works. So, there's a few things happening at Is
[03:58] this on? Can you hear Um so, there's a few things happening at Sentry Head. So, we've definitely seen um a an outset whereby the number of data-related products and AI-related products are outstripping the supply that we can give from the business for
[04:14] those. And that is still the case today at Sentry Head. We have a very hungry business. We have over 15,000 people in the organization, and we have a lot of use cases that we're able to serve. And so, hiring alone for us isn't going to solve that problem, essentially. Um and
[04:29] we found out that very quickly. We have a saturated market. We have new skills coming out all of the time, and to find people with those skills is becoming more and more difficult and more competitive. And so, as well as trying to find that skills, we're trying to leverage new tools and technologies to
[04:44] solve that problem at the same time. So, using some of the Agentyc tooling, um such as Genie code, for example, um is something that we're exploring. And what we're seeing at the moment is that if I compare some of my data engineers in my team now to 18 months ago, in some
[05:01] cases, they're are three times more effective um at ingesting data. We've seen that a very live example with we would in board onboard a new vendor, we would perhaps explore ADF jobs as a result, but because of some of the Genie code we can use and Lake Flow jobs, we're seeing that go from a couple of
[05:17] days uh sorry, go from a couple of weeks to a couple of days um with that as now. So, that's the probably the first lever that we're using is to pull on to technology to be able to do that. The second thing that we're looking at doing is we're trying to lower the skills for, as well. So, again, with the
[05:32] Agentyc tooling here, we want to make sure that as many people as possible within our business units are able to build what they want to build when they want to build it. And lowering that skills for is key. If I can get a sharp data analyst to be able to build something end to end without involving a
[05:49] data engineer. I think it's a very big win for the organization and that has seen a lot of speeding up in the delivery. And we proved that with our catering pilot as you as you touched on. That was our first real sort of jumping off point where we took a very small team, we took a platform, we were using
[06:05] some of our genetic tools to build and within a couple of weeks we had a MVP product that our CFO could use and that was really the the big you know wow moment for us to say we can do this at scale. We don't need to be able to steer this big ship in many directions, we can just be very agile in doing that.
[06:22] Uh that's fantastic. Uh so Unity Catalog is your governance backbone with migrations that are still in flight. Uh you know, Ali yesterday talked about enterprise context and your AI can actually understand, continuously learn, is permissions aware and all of these
[06:38] kinds of things. You know, in your world what has to be true about the business context before you trust a Genie answer? And kind of how have you ensured that? Yeah. Okay, this is um something that does keep me up at night a little bit when we were doing Genie
[06:54] migrations because I think everybody in this room knows that when you build trust, it takes a long time to build trust and it only takes a second for you to lose trust in a Genie answer and so you know, what we end up doing here at at Etihad is we make sure
[07:10] that before we actually end up deploying um a Genie space, there are a few things that need to be true. I think the first thing is definitely Unity Catalog needs to be enabled. So, by the end of this year, I talked to my yeah, we will be fully Unity Catalog enabled across the entire um organization and that being
[07:28] our lineage and our backbone of our organization needs to be true. We don't release a Genie space unless we are fully Unity Catalog enabled, not partly, fully done. Um and we're doing that in waves across our business units at the moment. So, the first thing I think is unity
[07:45] catalog is very important. The second thing um is probably semantics. So, we need to make sure that my engineering team don't own the business semantics. It's the business that own those semantics. It's the business that validate them. It's the business that sign them off. And as
[08:00] those change, we need to make sure that the semantics or the definition change with the semantics change with those definitions as well. It can't be a case of my data engineers or or product owners saying, "Yes, that looks right to me." We need to go and validate that with the business. And so, we often have
[08:16] what I call warranty periods whereby we release our genie spaces to a certain group of users for extensive testing before we then release it to a wider audience. And what we've had with our um human resources department recently is
[08:31] we did a big uh release of genie and we were waiting for some thumbs down on our questions. I think we had over 2,000 questions asked in a couple of weeks. We had zero thumbs down on the answer for it. And that was because of unity catalog, semantics, and then the testing
[08:46] we put in place. The other thing that I think needs to be true whilst we're doing migrations, you know, the the the HR one was from Tableau to uh AI BI and it's using personal information as well. And so, using sensitive data, we need to make
[09:02] sure that our access controls are A-back and our R-back are up-to-date and they're living breathing, you know, um uh within the organization. We need to make sure that I don't stale data can be as bad as the wrong um access controls in my opinion. So,
[09:19] when you provide an answer uh to a user, especially if it's pulling back personal information to them, we need to make sure that it's live and up-to-date as well. So, we put a lot of work into governance from that angle as well. And again, it needs to be collaborative, but it also needs to be very much business led and business owned in my view.
[09:35] Nice. I uh the the two no thumbs downs on 2,000 questions. I'm glad we got that I'm glad we got that video taped. thought we had we thought I would I thought I had a problem for about 4 hours and then I was like I didn't do such a good job. turned on. Yeah. Yeah. Um well, thank you. That
[09:52] that was tremendous. Um we'll move on to to Anna. Munich Re is the world's largest reinsurer and a heavily regist- uh regulated financial services group. Its data leadership has presented publicly on AI and data management in and across the group. Data Bricks is
[10:07] positioned as the standard data and AI platform spanning financial reporting, risk, natural language across business users, uh and strong lenses, regulated industries, and governance. So, Ali made the case that AI governance has become a whole company problem, not just data,
[10:25] but agents, tools, their identities. In regulated reinsurance, what made governance an enabler rather than a break while keeping the control on on regulatory aspects? Where have you seen security or compliance teams accidentally become the bottleneck? We
[10:40] hear so often, you know, uh from companies, you know, "Hey, uh regulatory and kind of governance is slowing me down." And then we hear from other companies, "Because I've got the governance, it's speeding me up." And we'd love to get a sense of that from you. Well, I I think both holds true, right?
[10:55] So, um just to put it in the context, Munich Re is very big um a group and has in different divisions. I'm representing and my perspective is coming from the investment asset management division. Uh in that sense, we are also highly regulated with Kite, which is similar to
[11:11] Bite in financial services. And um where we turn it around is in in things like when we're really reporting to the Bank of Germany, for example, Central Bank of Germany, we have we really have a standards where we need to meet all these um where we are um
[11:27] reporting on solvency or in IFRS. Those data are very structured and account book of records has their own rules and that helps us centralize solutions that we build in the past years with streamlined and highly automated
[11:43] processes for the business. And this is where we're in the future we'll be able to implement and evolve the agents with the appropriate controls, of course. Uh where we can see bottlenecks more is on the space of where we need to deal with vulnerabilities, where we need to
[12:00] do with new requirements coming up like Dora, for example, where our teams need to be much more on that and there's a lot of effort involved. Therefore, we're not bringing business value on that sense and could be seen as bottleneck, but at the same time with this we're really making sure
[12:17] that we protect our assets from the Minigroup perspective, which is data, the main asset. So, uh we've seen these vulnerabilities as as well in in many different keynotes how these can really be exponentially now and we really need to make sure that we can use
[12:32] the old tools. Data bricks is great in that, but also the everyday everyday AI tools to minimize that that impact. That's fantastic. Um you know, when AI moves from giving advice to potentially taking action
[12:48] autonomously and what have you, you know, the meaning of trust changes, right? In financial services, what does that have to look like from an audibility, lineage, human oversight perspective before you let a system kind of, you know, act as opposed to just recommend?
[13:04] Yeah. I mean, what is key is that the human in in the loop is a still defined in the new processes that we will have to co-create with business both in IT and business. We'll have new processes and the human in the loop and the break
[13:19] points would be crucial. So, we need kind of a parts of our organization that define where the guide rails will be, where the break breakpoints need to be uh from the internal control system perspective, and also because of traceability and accountability is key for those areas.
[13:37] That's where also tools like Databricks could help us. And yeah, this this all organizations and agents then could be put on more centralized scalable processes like calculating KPIs, like analysis market analysis and
[13:54] all these where you could put agents, but these agents could evolve as well and could be much more trusted. So there for the human in the loop could go out and then you could replace that with an agent going forward. Absolutely. Absolutely, that's fantastic. Um and we'll we'll move on to
[14:09] racetrack. So racetrack runs on Azure Databricks with a mature data warehouse and Unity catalog footprint and a common data model program. It's flagship app, store intelligence, demoed publicly at Data and AI Summit, is built on lake based Genie and Unity catalog metric
[14:25] views. It replaced the senior executive's most used daily dashboard, reached production in roughly 80 days, is in daily executive use, and is being extended into a shared platform other departments build on. Public reporting also covers earlier data set rationalization, about 2/3 cut,
[14:42] and data ownership across business domains. So let's talk about store intelligence since you're demoing it here. You replaced a senior executive's most used daily dashboard with a Genie enabled app on lake base and got it into daily use
[14:58] fast. What made that the right first bet and what did building on lake base let you do that you couldn't do before, especially as far as kind of making it actionable and informative? Well, I guess as they say, right?
[15:14] Innovation is really driven by necessity. And one of the issues that we had with our old dashboards and in our previous architecture was that at point we were kind of running out of time to to process our data sets in Power BI and it
[15:30] our reports at that point were you know fairly unreliable. A lot of these they just wouldn't even be done on time and it became such a problem that we were like let's let's let's rebuild this from scratch and and just to step back a
[15:45] little bit we we as most places I'm sure I'm sure most of you guys will relate to the story. We had a data warehouse a prior data warehouse that when we decided to move to data bricks at the time we were new at data bricks and I think most people
[16:01] were this was three four years ago. We lift and shifted pretty much all the issues we had with our old data warehousing environment. We greatly replicated in data bricks. And and that was also you know a huge bottleneck for us. So about six months
[16:18] ago we got together and decided you know if we had to build this right and and in the proper medallion architecture how long it would take and and to the desperation of our finance team and business combined it would have been
[16:34] just about a two-year effort to rebuild this from from the ground up. And this was as early as six months ago when we we sort of came up with that with that timeline. Fast forward two and a half months ago when we we
[16:50] truly began to dive into Genie we realized that we could have actually achieved that in probably three to four months. And that's what we're working on right now. We're working on building all the metadata that's going to power Genie going forward for to rebuild our data
[17:06] warehouse and our first sort take of that process was the the store intelligence dashboard which we we used Genie to reverse engineer a Power BI report into Databricks apps and using metric views, we were able to
[17:23] denormalize a lot of the data that we we had before that just was not getting processed at all in our old data sets in Power BI. And essentially what this this dashboard does is it allows, you know, executives to ask questions to to to to the data um
[17:39] and it shows margin we we can actually I wish I you know, I showed this yesterday. Uh essentially the idea is anybody can walk into any of the RaceTrac stores, well, anybody with with right access. You can pull up your phone and it would geolocate you and if you
[17:54] are an executive, you can look at every single metric about that store in terms of what, you know, are you what's the margin? What are the best sales? What the the best sellers? What you're not hitting target. So, it gives you just a huge host of
[18:11] uh KPIs and metrics. Um and it's it's, you know, you can zoom into the data as much as you want. You can ask any questions about your data. So, that that was just a a game-changing uh experience because, you know, dashboards gives you static data. With with this new
[18:28] dashboard, uh now we can actually create profiles. So, if you are a store manager, we can create a profile that allows you to ask certain types of questions that are related to your specific role. If you're an executive, we can expand that role and you can ask questions across the entire um uh
[18:45] product line. So, it's been it's been um just a game-changing experience for for all the, you know, RaceTrac in terms of access to the data and how quickly we can we can now uh have insights that we we just didn't wasn't able to have before.
[19:00] That's that's fantastic to hear. I I often, you know, when hearing from customers who are getting ready to to move things over to Databricks, it's like, you know, you can lift and shift it, but we'll just make your problems faster. Right. You know, I'm like uh you know, really that that refactoring and kind of
[19:16] rethinking things can create tremendous incremental insight. Sure. And that actually exposed, right? With with the Genie um you you know, before it would take a much longer much longer to find issues with your data because somebody had to create a
[19:31] dashboard, somebody had to do something like somebody was an active process. Right now with Genie, users have uh you know, the ability to ask data questions that now exposes issues that that were really hidden before. Yeah. Yeah.
[19:47] I love that idea that the down votes might just be issues. You know, um so uh you're a practitioner leader and and so I'll ask you um what buyers actually worry about. You know, we've we've heard the kind of no lock-in and we've heard the, you know, lowest
[20:04] TCO as both kind of driving factors for decision-making. When peers in this room are are weighing a turnkey platform versus kind of stitching their own pieces together or something like that, I'm curious, you know, how you evaluate
[20:19] it as far as kind of choice of cloud, choice of model, you know, what kind of lock-in you're signing up for and and kind of the real cost of the engineering overhead involved and what have you. With with our with our with our Yeah, I mean, just kind of with with some of these decisions around the data platform and and what have you.
[20:35] Well, it it's certainly I you know, I I liked the analogy that I actually made yesterday. It you know, if you if you scuba dive, leveraging Genie and and everything else that comes with that platform, it's kind of like, you know, going from from snorkeling to scuba diving with a tank.
[20:53] It's that it's that drastic. Um the and you know, in in in a situation where um it would take three to four weeks to create something, truly it's it's just mind-blowing that if you properly use Genie, you shorten that time
[21:09] in in you know four or five times easily. And uh so it's it's been really easy for us to to justify uh ROI on a lot of those efforts because we there were several projects that we
[21:24] we actually um priced out and they would have been you know three to six months projects with four to five resources. And and now with Genie, it's like it's even hard to to believe that you know I have SOWs that I wrote six months ago
[21:41] for you know $200,000 that we were able to build just about what we wanted with with internal resources and and Genie. So it's been it's been a pretty easy process. That's fantastic. I That's that's great. Well, so um this next question uh we'll
[21:58] do as a bit of a round robin. I'll ask the question and then I'll ask you each for an answer. And this is my last question. And so if you're tired of listening to me ask questions, not to worry. Uh we will be looking to you should you have any questions after I ask this one. So think up of whether or not there any
[22:13] uh crazy impending questions that you have for this group. So um you know, throughout the keynote over the last two days we heard four words repeated. Uh context, control, choice, and cost. Context, control, choice, and cost.
[22:31] Uh 18 months from now and I 18 months from now always worries I used to be able to predict 18 months into the future, but now I say something I think that's going to happen 18 months into the future and it happens two months later, right? Uh but so as you look 18 months out, which
[22:47] one of these, context, control, choice, and cost, do you think is the biggest differentiator between the Agentyc pilots and the agents that are in production, if you will, and and why? Which of the four and why? Go.
[23:02] Oh. You saw me? Sure. You saw me. So I just remember the four now. So, I think it's going to be control for me. Um I feel like this is a bit of a trick question, but they they're all equal, right? But, we'll go with control, right? So, the reason I don't think it's going to be context is because I think
[23:18] as organizations grow, we add data, we see more and more data. We've seen some of the releases yesterday as far as the keynote that's going to help us with that data, and context will just become richer and richer as a result of that. Um and so, I don't think it's going to be context. I also don't think it's
[23:34] going to be cost because I think cost causes you to take action anyway. I don't think that's something that is going to be a bolt-on. I think it's going to be something that organizations fundamentally think about from the start. I think there was a the McKinsey study that said you 70% of AI programs
[23:50] failed because of spiraling costs. Um however, I think organizations are now getting wise to that as a result. Uh contact well, we've done context, cost. Um What's the other one? The choice. Choice and control. The unchartered choice. Um again, I think platforms are going to
[24:06] converge to very similar offerings, especially hyperscalers here. And so, I think choice is going to be a bit of a not so much of a differentiator. It's going to become very similar in their offerings, and you have to pick very similar use cases that just suit you as an organization. So, I think that control is going to become more
[24:22] important. And the reason I say that is taking your AI from a very quick and easy pilot and then putting it into production is going to involve you making sure that you manage your access permissions well, you make sure that you've got really good data governance in place, you make sure that you know
[24:38] where your data lives, it's not stale. And I think those things are going to be living and breathing in the organization. Um technology will compensate for some of them, but I'm making sure that you know, if you ask Genie a question and you don't have control over where that question is
[24:53] looking, what Genie is doing, what answers it's providing to you, is it still relevant with the semantics? I think it's going to be a an issue. So, control for me is going to be the big one in the next 18 months that organizations may just miss. Yeah.
[25:08] Well, for me is is a little bit of a different perspective. It would be for me cost because I think all the companies and the hyperscalers are investing a lot of money, including Databricks, of course, in all this AI and they will need to have revenues at some point. So, we need to be very, very
[25:25] careful having a super cost discipline within the AI tools and agents we will be implementing as well. That's where I like really the Databricks new approach of really having even cost cuts and limits to the budgets and the tokens. I think the conversations around 18 months out would
[25:42] be more on 20 this agent and this LLM, can I do it in a latest in a smaller scale using less tokens and can I do instead of an agent a normal workflow? So, in financial services we bring working in automation and ROI and
[25:59] and profits in in the past years. Now, the conversations are going to be much more transparent with business in terms of what model will you use, what token do you use and that will be available for all, not only for IT and business. And those conversations are going to be in the next
[26:14] months, I think. Um definitely cost, right? We it it's one of those things that leadership at racetrack um they really want to fund and and allow innovation to occur. So, we are at that stage that that we we
[26:31] want people to do as much as they can with with with AI and the tools in hand. But then, you know, as we're as as as the adoption has been wonderful, now we're beginning to realize that we we need to have the right governance in
[26:46] place, right? And one of the the the certainly the the challenge that we have had uh with access and control is we have had to sort of rethink as talking about rebuilding it our data warehouse, you know, define what other business domains we we need to be able to secure
[27:03] the data to the specific people who should see them. Um you know, as every organization you always have, you know, one guy that's like 14 groups. And you know, and and what what what level security that that person should have given that now with
[27:18] with AI tools, you just have access to so much data so quickly. So that's been a challenge, right? We've been really redesigning a lot of, you know, reshaping a lot of how our data is structured. Defining uh you know, the the business the business domains and and who owns
[27:34] the data in many ways. But uh you know, there's the other side of that, right? Which is innovation. We have been experiencing with um camera vision using Databricks on we know we've been try we we we need to be able
[27:50] to So if you're if you go to a RaceTrac store, we want to make sure that we have, you know, fresh pizza available every time uh customers walk in. We've been experimenting with uh Databricks in in in in in different models to make a call into our Meraki
[28:06] cameras, pull the frame and and analyze, you know, do we need to send a message to the manager to put more pizza in the oven cuz the pizza is 2/3 eaten. So but how do you how do you balance, you know, how many calls how many frames per minute do you want to pull? Which model
[28:22] to use? We went from when we began this process, we you know, each model each call was costing us about two $2.55. And we got it down to $0.42. But by trying different models that uh you know, maybe you don't need the most
[28:38] expensive model, you can do that with less expensive models. But how do you get to that maturity, right? How do you get to that point where you have that that understanding of, you know, what what you actually need. And I think every organization is going through that right now, right? Absolutely. That's It's really exciting.
[28:55] I I'll say uh my hope is that cost and controls converge, right? And that we're we're sitting there giving the controls for the costs in many ways. Um you know, I I built that capability of having the camera on for a pharmacy many many years ago, uh
[29:11] but because the the things they were selling were so much more valuable, they were willing to pay $10 or so per frame versus the 42 cents that you're driving it down to at this point. That's fantastic. So, questions from the Yeah, go ahead.

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.