Sponsored by: Replit | From Lakehouse to Production App: Building Governed AI Apps

Summary

This video demonstrates how the Replit and Databricks integration lets any team go from a plain-language prompt to a production app running on Databricks Apps, inheriting Unity Catalog governance and enterprise authentication without writing infrastructure code.
A live demo shows a scenario modeler built entirely through Replit's AI agent, which pulls lakehouse tables and models from Databricks to construct, iterate, and publish a governed app in a fraction of the time traditional development requires.
The video covers Replit's infinite canvas for parallel prototyping, multi-user editing for collaborative app development, and the one-click publish workflow that deploys apps to Databricks with scoped service-principal permissions already in place.

Chapters

FAQs

How does the Replit and Databricks integration work?

The integration connects Replit's AI-powered development environment to Databricks Apps through a service-principal connector with scoped permissions, so apps built in Replit can access Unity Catalog–governed lakehouse data and models securely. This video walks through the full flow from building in Replit to one-click publishing on Databricks Apps with inherited governance.

What is Databricks AppKit and how does Replit use it?

Databricks AppKit is the toolkit that enables apps to run natively on Databricks Apps with Unity Catalog governance, enterprise authentication, and observability built in. This video shows how Replit's AI agent leverages AppKit to automatically wire up governance and permissions when generating and deploying an app from a natural-language prompt.

What kinds of applications can teams build using Replit and Databricks together?

This video covers customer examples including revenue scenario planning tools for sales teams, real-time operational dashboards, AI-powered internal companions, and mobile applications—all built from lakehouse data without requiring a dedicated engineering team. The integration is designed for business users and non-engineers to build production-grade apps on governed Databricks data.

How does one-click publish to Databricks Apps work with Replit?

When a team is satisfied with their app in Replit, the one-click publish action deploys the app directly to Databricks Apps with the service-principal permissions and Unity Catalog governance already configured through the integration. This video demonstrates that the published app inherits all enterprise authentication and observability settings without any additional setup steps.

Full transcript

[00:07] Hello everybody. Uh, how's everyone doing today? Uh, kind of last day of the conference, so I'm guessing people are a little worn out, but thanks for being here today with me. Um, my name is Axel. I work at Replet as a field engineer working with our enterprise customers to
[00:20] basically in implement Replet at their organization. And today we're going to be talking about the integration between Replet and uh, data bricks. We have a really strong integration and partnership with them. So we're going to be talking through how you go from uh
[00:33] you know your lakehouse to a production application uh in in data bricks. Okay. So uh clicker it does work. Sweet. So uh forward-looking statements you can
[00:45] read those at your own pace I guess. Uh but I'm just going to move on and uh talk about what we'll cover today. So today we're going to talk about uh the problem uh which as we see it is getting good data into your business users's hands is still very difficult to do
[00:58] still very hard to do uh how we see and think about the solution uh combining replet and data bricks apps uh to uh change what it what it takes to go from you know an idea to a initial prototype
[01:11] to an MVP and all the way to production in data bricks apps in a governed way. I'll give a uh live demo uh which is going to be fun and then uh we'll also talk a little bit about customer examples and then hopefully at the end
[01:24] we'll have a little time for Q&A. So feel free to save all your questions as we go through here uh for anything that we uh didn't cover that you want to cover. Cool. So uh getting back to the problem.
[01:36] So getting your data into your business team's hands still takes months still takes a very long time to do. So even if you've built something that's awesome, that's incredible, you know, we're all here at a data bricks conference uh
[01:49] building on amazing data platforms that can help us centralize our data uh make it uh super easy to kind of create those production grade enterprisegrade governance over it using you know things like unity catalog. Uh you have you
[02:02] might already have reliable pipelines, AI models, a lakehouse that can power real applications. Um, but people and businesses are still struggling to actually put that data to work. It's a very difficult problem to solve. So even
[02:14] if you have built something that is, you know, incredible, it's a beautiful data ecosystem, it's hard to actually get that uh value out of it. And even if you haven't built something, maybe you're still in the process of putting together your uh your your data ecosystem and
[02:28] you're you're still doing a lot of migrations, you don't feel like you're not ready. Even then, uh you want to have incremental value from your data platform and data ecosystem. Uh so so why is this? Well, the current
[02:40] state of building on top of your data is pretty brutal for business users and nontechnical users. So your business uh user, you know, someone on your finance team, someone on your sales team, uh
[02:52] somebody who's, you know, maybe on your marketing team, people who are nontechnical would have a problem or an idea uh for for a solution to uh you know, a business case that they they want solved. So they need data to do
[03:04] that. They need something that's data powered um that is maybe a dashboard but often goes beyond a dashboard and into something more of a application that they're interacting with, working with in their day-to-day. And so they have
[03:16] this need and then they uh go to your engineering teams, your data analytics teams, your data science teams and ask for it. Uh sometimes it's an official process, often times it's not. Uh but then that often gets put in a backlog.
[03:28] We hear stories about this all the time. someone has a great idea uh and then it gets put in a backlog and is lost forever. There's a ton of lost ideas because uh people, you know, as data professionals, it's hard to actually
[03:41] prioritize things. Uh and so it it does make sense that things get lost in the background because of this. Then your business users start doing things like doing workarounds, right? So they might have a ungoverned spreadsheet that
[03:53] they're working off of. uh they might be um like pulling data from sources that are not actually you know sources of truth. um they might be uh trying to spin up you know uh shadow IT
[04:06] applications right like buying things that they shouldn't be or using applications or software that they shouldn't be and that is not good all right uh it's it's not good for everyone right you you don't have the governance that you get uh from your data platform
[04:19] the business users needs are actually not met because they're restricted by the tools that they have available to them and and the uh access that they have to to usable data uh and so it's it's like a tension for everyone it it
[04:31] hurts everybody in this situation. It isn't just one person that's uh that's going wrong here. And then even after you get out of the backlog, so maybe you you did go through the whole process. Um business users maybe put together, you
[04:44] know, a business case, everyone agrees with it. Um often times those tools that are created are the wrong fit. There's a lot of different requirements uh that maybe changed over the course of them even requesting it, right? They might
[04:57] have requested one thing, took three months to build the tool and now they need another thing. They needed something to evolve. What what they asked for previously uh is no longer what they need. And this is like a huge problem across uh to all enterprises
[05:11] that we encounter. Uh and and it's kind of like a fundamental reason why we uh you know prioritize this partnership with data bricks and replet. Okay. So
[05:23] the solution is I kind of already uh teased it a little bit. Um you know what if you your finance lead could describe the app that they need and have it running on data bricks by the end of the day, right? So so not just um not just
[05:36] your technical users building tools, not just your uh data engineers, data analysts, BI developers, data scientists building tools. It goes beyond that and lets your finance people uh your sales
[05:48] people your non-technical users uh build on top of the great data ecosystem that you've already built. Um and that's what Replet is. It's a platform for uh nontechnical users though we do have
[06:00] many technical users I will say as well to build on uh with natural language uh any type of software that they they want. Uh so you can describe your app in plain language. Replet agent will in the background create that application for
[06:13] you and then you can actually deploy it straight into your data bricks environment as a data bricks app. So about 75% of the users building with this integration right now are non-technical. Uh and the nice thing is
[06:25] that you have full ownership over it. So even if you are in a technical role and you still want it to be governed, you have access to all to under to all the underlying code. You can push it to your git provider and we'll we'll show some
[06:37] of these things as well. uh you can push it to your git provider and then once it's actually deployed it's not deployed in some uh thirdparty service it's deployed right directly into your uh datab bricks environment right into your uh datab bricks tenant
[06:51] okay so this is now the workflow that that we see so instead of that previous mess uh that we were working with uh instead you connect your admins set up the link between your data bricks
[07:04] workspace uh replicate can be granted only what you want it to see so you can limit permissions to uh just see schemas for example or uh only see masked data and we have a multiple different ways of connecting. We have both a uh M2M or
[07:19] service principle connector that you can set permissions and also a U2M connector. So it's by user uh then after you have that connection set up your users will be able to use it to create uh and create applications and solve
[07:32] their business problems. So then they'll be able to describe what they want in plain English. They don't need to to look at the code though they can if they want to build a full stack stack application with both a front end and a back end. Uh they can pull all the data
[07:45] from datab bricks they can even pull in genie workspaces AI models from data bicks etc. Uh and then get it up and running and actually view what they're creating uh real time
[07:58] from there. It's a one-click deploy to your data bicks environment. Uh so after you are after your users are are done with it or ready to to rock uh they are able to deploy it and publish it right
[08:11] to your data bricks tenant and then the app is live. So every single user uh would who would want to access it would be logging in with their own data bricks identity. All the governance rules that you've already set up in you know unity
[08:23] catalog and data bicks apply to that application. So we natively support things like on behalf of authorization. Uh you can set different uh permissions for uh you know across the the application and make it uh just just
[08:36] what you want it to be. So this is the new this is the new workflow as we see it. Uh I'm going to go ahead and move on to the demo here. There we go. And I'll start uh grabbing
[08:49] my laptop screen. Perfect. Okay. So, this is the replet environment. Um, it's very easy to use.
[09:01] It's very simple to use. I've lost my mouse. Give me one second. Uh, but what what you basically your users will do is they'll describe exactly what they need in natural language. And then from there, they'll be able to um build their
[09:14] application. There it is. Found the mouse. Boom. Uh so what I'm going to do here is go through like what the typical app uh making process might look like uh for
[09:27] you or or other people using Replet that are want to use the data bricks integration. Uh if they want to deploy to data bricks we have a predefined data bricks apps template that they can select from. So if you click up here and click the data bricks app template um
[09:40] they're able to use that and that will come out of the box using appkit uh also supports OBO or on behalf of authorization uh and has additional uh agent skills that are built in to make it particularly good at dealing with uh
[09:54] and creating appkit uh applications. Then I'm going to paste in a uh little prompt that I pre-made here. So create a simple reactive forecasting and scenario planning dashboard for sales pipeline
[10:06] bookings powered by my data bricks data. So in the background I have like a little demo environment that's running that I've included some uh data in. So you're able to uh specify with natural language. You don't even need to include the exact tables though you can if you
[10:19] want to. and Replet agent will be able to look through your data and find and explore it uh and see exactly what uh should be used to create the the data. So I'm going to send this off and it'll start working and what's happening in
[10:31] the background is again it's pulling together a full uh stack application. So it's pulling together both a front end and a backend that allows you to go from uh you know your initial idea to that initial prototype and then all the way
[10:43] to a production app in data bricks. So, first it's creating the it's grabbing the integration. I set up a service principle integration and gave it the exact permissions that I want my replet users to have. So, I limited it to a
[10:56] couple of uh specific tables that I want them to have access to. Um, and then I also gave it access to a genie space. So, we can even pull pull that in uh and put it in our application as well.
[11:10] Okay. So, while that's running, I did do a little bit of uh you know demo magic. will this will take you know between five and 10 minutes to run as it's exploring the data and looking for different things. So I already have a
[11:22] different version of this pulled up uh that will pull up here. Um and then after you've kind of created that initial uh you know application what you can see is a live preview of exactly what's going on uh in it. So in this
[11:36] case we created the scenario modeler. Uh you can see that uh I have like a reactive um kind of initial dashboard that lets me decide what I do and don't want to look at. I can simulate
[11:48] different growth paths and I can even scroll down and see some additional data that has been populated. From here your users can specify exactly what they want to change, right? So they don't need to um there there's no like needed dragging
[12:01] and dropping. Um they can instead just specify what they want in natural language. So I could say something along the lines of uh I' I'd like some top level KPIs
[12:13] uh for this dashboard and you can specify that and as uh replet agent is going through and working it'll basically uh render live uh those changes that you're making so that your users can get exactly what
[12:26] they want right we don't want situations where users are um are asking for things their needs change uh and then once once you deliver for them is no longer what they actually want. Uh you know, because
[12:38] they their needs have changed. Uh so this is a way for them to get kind of exactly what they want. It's easier to to maintain change. Uh uhoh. Little live demo thing. Maybe I'll
[12:52] give it a refresh here quick. Uh okay. Well, it happens to the best of us. I'll try one more time.
[13:04] Uh if that doesn't work, that's totally fine. It might be like an internet thing. We'll see what happens here. Um okay, but one thing that that we have seen more and more of with uh with
[13:16] applications that are being built, um there we go. Uh so one thing that we've seen more and more of uh with applications that are being built is they they want to go beyond just initial dashboards. So you can see that this is like a nice initial dashboard that gives
[13:28] people a view into their data, but it doesn't actually let them take actions. it doesn't let them like interact with the data in the same way that a full web application can and does. Uh so one of the things that we see uh very commonly
[13:40] for users using the replet and data bricks integration that they're moving beyond just like static dashboards. I was hoping this would finish by now. It ran into a little error. So I'm going to actually just skip over to another thing. Basically what I did was I
[13:53] specified to replet agent, hey make some additional changes to this uh dashboard and uh make it more um you know fitting for the workflow that an actual salesperson that's using this would want and make it so that they're actually
[14:06] able to take actions uh and and uh things like that. So then another little demo magic. I'll go over to this uh preview here. And what I did was I spun up a bunch of different agents which Replet supports through uh parallelism.
[14:19] Um, and you're allowed you're able to create a bunch of parallel agents and work on different parts and tabs at the same time. So you can see that this is kind of the same broad initial dashboard, but now we've added a whole bunch of different tabs for uh different
[14:32] things that the user can take. We've also made it reactive. So we can actually write back to your uh data bricks environment. You can even you can connect it to a lakebase uh to write back to a lakebase instance or you could uh connect it to other resources that
[14:45] you want and interact with them there. as mentioned previously. So, uh we fully support parallelism, basically being able to run multiple agents at the same
[14:57] time. If I go up to this top left hand, uh place right here, I can click the plus button and then spin off a fully parallel uh task that will create and modify the dashboard as I wanted to. So,
[15:10] if I uh bring off this side task, I could say something along the lines of make me a version of this with um you know, more, I don't know, earthy tones
[15:23] if I'm just uh doing uh something completely visual. And then what it'll do is it'll create an alternative version for me to review. Um I think the Wi-Fi is being a little slow for me right now. Um there we go.
[15:40] And then it asks me a couple of questions to follow up and it'll start bringing together a plan for me to um actually start creating an alternative version of this that I want. You can actually run up to 64 agents in parallel on replet if you're uh using the the
[15:54] enterprise uh you have an enterprise account uh which makes it great. You can also spin off agents from your main tab as well. If you go back here and click the plan mode button, uh you're able to spin off multiple agents in parallel uh
[16:07] to to finish your work and work on a bunch of things at the same time. You can even see on the sidebar here that uh I had a couple of tasks that were already queued up. Um
[16:20] you'll notice that it turns green when you're ready to merge it back into your main application. Uh and also if you have uh conflicts or things that you want to that uh require one being finished before the other one does,
[16:33] it'll also cue it. So uh this one it has to be wait to be merged before uh before the next task is actually updated and pushed back to main.
[16:46] Give it another refresh here. Taking a little little while to load. One of the things that parallel parallelism supports uh is also uh very collaborative sessions. Um and oh yeah
[16:59] actually one thing before I get there uh another thing about full web applications is it makes it again very customizable. So earlier I added uh in this task uh the ability to have like a walkthrough of my dashboard. So it can
[17:11] give a guided tour to new users of exactly uh what they're going to see or what actions they can take in this particular dashboard. Again, we're moving beyond uh we're moving beyond dashboards into full applications. We're
[17:23] making it workflow driven and making it uh exactly what users want and need for their uh particular business case. The other thing that uh parallelism
[17:35] allows us to support is extreme collaboration. Uh so you can have multiple people working on the same project at the same time uh because they're able to run these parallel agents over your dashboard and
[17:49] application. So you can actually see there's this little ET section here that is actually my brother sitting right here. He's joined in. Uh we both actually work at Replet as field engineers. And so you can have multiple
[18:01] people working uh on the same project doing uh different parallel tasks. So maybe Ethan, you want to spin off another task while while we're working here? Um, and then I'll even I'll spin off another
[18:13] task as well. So maybe I'll say um make me a version of this um application that allows me to uh you know run an ARMA
[18:28] model over my time series. So create like a you know a more interactive um or more like uh statistical model over the
[18:40] uh the time series data that I have instead of just a kind of simple forecasting one. And you can see that Ethan if I zoom in here has spun off another task. Uh he's he's asking for a page that gives an executive summary on
[18:52] a weekly basis of this dashboard as well. It does feel like a video game if you have like two, three, four people working uh in in a replet app at the same time spinning off tons of parallel
[19:04] agents. Uh we have companies with you know 10 people working in the same project spinning off you know 64 agents at the same time uh working in parallel and it's a great way to move faster and uh and get get things get things done in
[19:18] a in a very cool cool way. Yeah. So, as I said earlier as well, uh this is all looks like there's another question. I'll let Ethan uh decide what to do there. This is all supported uh
[19:32] and and managed like all the the code is accessible. It's all managed via git. If you go uh to this plus but tab right here, you can click uh you can type in git and see all of the different agent commits uh that have happened and any
[19:45] commits that you've done. You can connect it to your remote repositories as well. If you go up to the top right hand corner here, uh you can access all the files and all the underlying code. Um and you can even interact with it in a uh you know ID IDE like fashion if you
[19:59] want to edit the code yourself by hand.
[20:11] Another thing that replet uh supports is a uh infinite canvas that the agent has context over and is able to access and understand. So, if I go to this uh top lefth hand button here, uh I'll zoom in a little bit so that you can see. If I
[20:23] go to here, uh I can click this and it'll pop out uh into an infinite canvas. You can see I've already taken some actions in here. Um I've laid out all of the different pages that I have in my application. So, I can look at
[20:35] them, see them all in one place. I can interact with them. This is a fully interactive canvas. You can see Ethan's cursor is right here. We could take notes. uh you know maybe I don't like this particular uh page or something
[20:48] like that. I could take notes. Um I could uh add additional context here. Uh you can see we've even had agent lay out kind of an architecture diagram of what this an initial architecture diagram of what this uh application looks like and
[21:01] is doing. You can also use this you know one of the main use cases for this is spinning off additional designs. So if I uh if I click on a particular um page or application, I can uh page uh I can ask
[21:15] for changes to it. Uh so I could say make me some alternative designs for this uh page and then send it off and agent has context as to what I'm looking
[21:27] at and what actions I'm taking in the canvas. I already did this earlier uh with this particular forecasting page. You can see that there are these three uh little mockups over here and these
[21:39] are just front-end uh mockups. They're not actually including the full front and back end, not the full end to-end application in Replet. Uh which makes it easy to spin off variations. Again, making it so that users get exactly what they want out of their applications and
[21:52] it's not something that's um not conforming to their exact needs.
[22:07] Okay. So then if I go back to the uh main application um yeah we can make those changes. We can do them in parallel. Uh we have full access to the code. We have an infinite canvas that lets us um make changes. And
[22:19] now after we've kind of got it to the point that we like, we've made the exact application that we want. We've pulled in all the governed data from datab bricks. What we can do is now publish it. Uh so if I go to this the publishing
[22:31] tab, I'm able to publish it directly to my data bricks instance. Uh and you can see that uh normally with your when you're replic Oh, it looks like uh my Wi-Fi is a little broken here. That's okay.
[22:46] Uh normally when you're normally when you go to the publishing tab, you're able to publish to a replet.app. Um, but when you're connected to data bricks, you're using the um you're using your actual internal
[22:58] data bricks environment. Oh, no. That's okay. I'll just talk over it a little bit. Uh, you're actually using your uh actual data bricks environment. Uh so when when you're publishing instead of publishing to a place that is um you
[23:12] know in replet or outside of your data bricks environment or ungoverned you're publishing it directly to uh your your data bricks tenant and you're able to have those same controls and security uh
[23:24] that you have already configured in your data bricks uh instance. Oh man. Okay. We'll see if we'll see if this
[23:37] loads in a second. Uh if not, I'm going to go ahead and just go back to the slides. The Wi-Fi has been kind of bad for me all week. Not sure about you all. Uh but
[23:49] we'll make it work, I guess. Okay. Well, um once you actually are, we'll we'll just show you what it what it looks like broadly. So once you are
[24:02] uh actually publishing it uh to your data bicks environment uh from there it's hosted in your data bricks apps. So if you go up here and you can click on this data bricks apps. Uh I won't just because of the Wi-Fi problems. Uh but
[24:14] once it's there again it's fully governed. We're not actually writing any data back to your uh replet environment. It's fully contained in your uh data bricks environment with all of your controls all of your governance there. And again, you can connect to any of
[24:27] your the services that you have in uh data bricks. Uh so if you have like a a genie space that you'd like to use or an ML model or anything like that, you're able to access that through your replet
[24:39] application. Okay, we're back. Um so I'll go back to that publishing tab just so that we can see what it looks like and then we'll also access it in here. Uh so you can
[24:52] see that we have uh data bricks apps domain doesn't like that name for me. So I'll just call this dashboard sales forecasting dashboard 3. And then when I'm ready to I can publish it directly to my data bricks environment.
[25:04] One other thing to note is that we do support um a wide variety of artifacts that you can publish uh to your data bricks environment. So oftent times you'll be making like a web application or a website uh that you want uh hosting
[25:18] your you know workflows and applications. But we also support creating slides uh and then mobile first web applications as well. So if I go back to uh my initial uh project, I can click on this tab and I've also created
[25:31] a couple of slides to go with this. So one thing that we see pretty commonly is people don't like uh copying and pasting the the data from their BI reports. Um and they prefer to have something that's
[25:43] live updating. So these slides are being uh live populated with the data bricks data. So as it updates, as it changes, uh it's it's updating these slides and once you publish it uh or have these slides handy, you don't have to
[25:56] constantly um you know go back, change the data manually, uh things like that. So it can be it could be very useful way of using uh this this integration. The other thing is that Replet is great at
[26:08] making um mobile first web applications. So if you do have use cases where you people you want people to be uh you know in their data in your data bricks tenant uh but not um but uh you know maybe not
[26:21] on their laptop they might be using their phone or something like that. Uh you can also create you know more uh mobile friendly versions. So if I go here I can change the screen to an iPhone. I'll do the iPhone SE or maybe
[26:34] 17. Uh you can create web applications that are uh mobile first so that people can uh view things on their mobile phone uh without actually um being you know uh
[26:46] outside of their data bricks tenant. Uh it's all still governed and accessible uh for them there. Okay, now that the Wi-Fi is back, I'll go ahead and also take a peek in our
[26:58] data bricks environment at what the uh data bricks apps looks like. So I go to the data bricks apps. I published another version of this a little while ago. So I'll look up sales forecasting. I can go to the sales forecasting
[27:10] dashboard published app. Click on here. And then uh we get that application that we created in replet because we have the app template uh prepopulated in replet. It it's it's uh containing the kind of
[27:24] architecture that's required the packages that are required and the uh and skills to kind of help replet agent work with data bricks apps. So you get the exact same application once you publish it uh back to your your data bricks tenanted environment
[27:42] and then again you can still take those same actions uh that you you took uh in your in your replet app makes it very easy for your users to make exactly what they want.
[28:00] Okay. Um, so I'm going to go back to the slides then, uh, and talk a little bit about, um, you know, teams that are already using this. We were in private preview until, uh, pretty recently. We're now in public preview. So, anybody can access this integration and take
[28:12] advantage of both the integration uh, piece that pulls data in to their replet applications from data bicks, but then also publishes those uh, apps as data bricks apps to their data bricks
[28:24] environment. Uh so we have a a bunch of different use cases that people are using across uh industries. Uh something that we've seen is you know revenue scenario modeling. So something kind of similar to what I showed here today.
[28:36] We're seeing real-time operations dashboards. Uh building type uh internal tools that are helping their operations teams um kind of manage complexity. Uh and then we're seeing internal AI
[28:48] assistance. So again we do integrate with uh you know genie spaces but also any type of AI models that you have in your data bricks environment and then replica can also integrate with them uh itself. So you can integrate with those
[29:00] and create actual AI agents in your applications. Uh and then a very very common use case is self-service analytics apps. Uh so either client-f facing or internal applications uh that
[29:12] uh allow people to you know create data exploration tools that fit exactly what they want to do and empower people who are you know maybe um technical adjacent uh and uh empower them to explore the
[29:25] data that they have access to and use all the governed data that they have uh and uh and yeah get a lot of value out of it. Okay. And then we we had kind of early
[29:37] access with teams at Bane, Zillow, Asenture, and Abacus. And again, would love for all of you to try it. Um it's a it's a great integration. Um if you are interested in learning more, you can go to replet.com/partners/databicks.
[29:53] Uh we also have a booth here. So if you're interested in learning more, you can swing by the booth. It's uh 651. It's near the puppy park, which is a great place to go. Anyways, the puppies are very adorable. Um but would encourage you, yeah, to to give it a try. we could do a free we could do free
[30:06] enterprise trials and things like that. Uh so yeah, appreciate it everyone. And uh I did save a little there is a little bit of time left for a Q&A. So if there are any questions um I think I'll have a person walking around with the
[30:18] microphone. That's cool.

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.