Build Production Apps with Databricks Apps: Scaling, Governance, and Citizen Developers

Summary

Databricks Apps has grown to over 5,000 customers building more than 150,000 applications since its general availability launch, making it the fastest growing product at Databricks by nearly every metric.
App Spaces provide permission isolation and governance for citizen developers — non-technical business users — enabling them to build and deploy applications safely within production workspaces without exceeding their authorized data access.
Genie App Builder leverages Unity Catalog permissions and Databricks environment context to let users create governed, production-ready applications without writing code, with skills, MCPs, and plugins extending integration to external tools.

Build Production Apps with Databricks Apps: Scaling, Governance, and Citizen Developers

Building enterprise applications has historically required months of development time and complex infrastructure management. Databricks Apps changes this by enabling both technical and non-technical users, citizen developers, to rapidly build, deploy, and govern production applications on the Databricks platform.
Learn how Databricks Apps powers 5,000 customers building over 150,000 applications. The session covers serverless infrastructure with horizontal auto-scaling, app spaces that provide permission isolation and governance for citizen developers within production workspaces, and Genie App Builder which leverages your Databricks environment and Unity Catalog permissions. You'll also discover how skills, MCPs, and plugins integrate external coding tools while maintaining production governance through Unity Catalog permissions and app spaces policies.
📂 Dev Hub: https://developers.databricks.com
🤝

Chapters

FAQs

What is Databricks Apps and who is it designed for?

Databricks Apps is a platform for building, deploying, and governing production applications directly within Databricks, designed for both technical developers and non-technical citizen developers. It features serverless infrastructure with horizontal auto-scaling and App Spaces that isolate permissions so business users can build safely within production workspaces.

How many customers use Databricks Apps and how many applications have been built?

Over 5,000 customers use Databricks Apps, and more than 150,000 applications have been built since the product launched in general availability at the previous year's Databricks Summit. According to Justin Durivage in this video, it is the fastest growing product at Databricks by nearly every growth metric.

What is Genie App Builder and what does it enable?

Genie App Builder lets users build applications by leveraging their existing Databricks environment and Unity Catalog permissions, with Genie providing the AI context to generate the application logic. This enables users who cannot write code to create governed, production-ready applications that inherit the security and data access controls already configured in their workspace.

How do App Spaces protect production environments when citizen developers build applications?

App Spaces create permission-isolated environments within the production workspace where citizen developers can build and deploy applications without accessing data or resources beyond their authorized scope. Policies defined in App Spaces ensure that self-service development scales across the organization without compromising the governance and security controls already established.

Full transcript

[00:19] All right, welcome back to the Summit Live studio. Uh for everyone joining now, I'm Ari Kaplan, the global head of evangelism at Databricks, and I wanted to welcome two new people. Uh You're Jason, you're going to be a co-host for the next couple segments. next couple segments. I've been this
[00:35] I've done this for a couple years now, so I love I love doing it, and thank you for having me back, Ari. Yeah, looking forward to having you. And Justin, people you probably seen him on stage today, but please introduce yourself. Yeah, uh Justin Durivage. I'm a director of product here at Databricks, and I oversee apps and customer like.
[00:52] Excellent. How was it being on stage? It was like an insane number of people. It's actually now the largest AI conference in history. Yeah. As well as the largest data and AI conference, however you define it. 31,000, I think, right? In-person. Yeah. And then over 100,000, including the
[01:08] folks 160 countries. Yeah, I would say uh it feels like it. I mean, there's a lot of energy here. Um it's crazy, you know, the keynote, there's a lot of energy. There's a lot of good uh feedback from the crowd, right? You There's a lot of jumping in and clapping and cheers, and it just
[01:24] feels like a really fun environment. Uh people are really excited, very positive. It's just It's always great. A great experience to have. So, what products are you uh in charge of, and tell us about your team? Yeah, yeah. It's a great experience to have. I'm also glad I'm past the keynotes.
[01:41] Congratulations. You're great. You're so calm. good. Thank you. Uh yeah, so I oversee uh Databricks apps um and customer like. I think we're talking customer like a little bit later. Um but yeah, on the app side, it's just been uh tremendous year of growth. We launched apps last year actually at Databricks Summit in GA.
[01:57] Mhm. Um and you know, for those of you who didn't catch the keynote, uh we have over 5,000 apps customers now, built 150,000 apps. Um it is the fastest growing product at Databricks by pretty much every metric. Yeah, in the in the history. In the history, yeah, it's just crazy.
[02:14] Um and you know, people are just really excited to use it. They're building cool things. Um quite honestly, it's just a really fun product to be on because it feels like it's unblocking things that people didn't know were possible, right? Yeah. And one of the things uh that was really exciting whenever I was hearing you talk
[02:29] on the keynote was you're talking about, you know, the persona that you guys are targeting or I don't know if you intentionally targeted them or you discovered it, but can you tell me a little bit about about the the citizen app developer and and what that looks like? it it's it's definitely intentional and it's not where we started. Um so, when
[02:44] we started with apps, it was really about empowering the data teams to build. Um so, the vast majority of app developers were still on the data or like central engineering and platform teams. Um actually, something like 96% of apps are built by the data engineering teams
[03:00] today, even though 60% of users are coming from non-technical teams, app users. So, you see that split, right? Um and we looked at that and if you look at what's happening in the broader world, everybody's low-code, right? Um citizen developers are getting it's like a renaissance, right? It's explosion of
[03:17] um tools and platforms that that cater to to citizen developers. And so, um we want to unlock that next wave of innovation. Um I really feel like at this point, we're not going to be constrained by building solutions. We're going to be constrained by understanding
[03:33] problems and who understands problems? It's the people that are living those problems day-to-day. True. Um so, we want to empower them to be builders. And you know, the Databricks over the course of the last 2 years, we've been talking about turning those people into users of data and AI and that's amazing. But we want also turn
[03:49] them into builders. So, you know, a couple of ago, we started on this journey of like, well, what would it take to empower everybody to be able to build an app. Um, and you know, we've been working on this for a while, so it's exciting to be able to, you know, unveil this this uh this week. And it was uh it was interesting like so
[04:04] last year when we first came out with uh apps, it was uh you know, it was a pretty early product, but even then we were seeing a lot of adoption. But at early product, it it kind of like it could only scale so much, right? Like it was pretty much just like one single machine, but I think we've changed that since then, right? have, definitely. Yeah, and that was
[04:20] kind of the first dimension we got pushed is um okay, we built these apps, it's limited to a single machine, the app becomes successful, what do we do, right? Um, and so we've been working a lot on kind of the core platform capabilities. Um, we added larger instance sizes. Um, we
[04:36] just released horizontal auto scaling. Um, so now apps can be multiple instances. So it automatically scales out as your number of users Exactly. And um And not just number of users, like what they're doing, so we look at the resources, we look at the CPU and memory, and we scale out automatically.
[04:51] And what's really cool is scaling out an app where there's stateful sessions um is actually tricky, right? So have to make sure that the next request that comes in knows about the previous. Exactly. Exactly. So, um we actually had to implement a lot around the scaling to
[05:07] um maintain session state, sticky sessions. Um, you know, there was a lot that went into that. Um, but we're super excited about it now, and we see apps now with tens and tens of thousands of users on a weekly basis. So, it's incredible just the amount of growth that we've seen.
[05:22] I see I see Yeah, the growth is everywhere. So, first of all, if you haven't tried this, it's something you absolutely have to get into, but you know, 2 years ago we were talking about people were dipping their toe in, and now the whole citizen developer aspect is growing, but with that growth, you
[05:37] know, the the costs could go unchecked and governance when you start scaling. So, how does Databricks help help uh make that all right? Yeah, yeah, I mean, it's funny when you talk about citizen developers to some of like our core platform teams, there's always this like tension between, you
[05:53] know, cost and governance and innovation. And you know, these platform teams, they want to unblock innovation, but when we went and talked to them, their biggest concerns was well, first of all, our governance. Um it's really designed around maintaining a single app today, right? Um so again, the
[06:08] assumption was the data teams are building this, they have the right permissions, they go in, they grant those to the app. Um and that's great for that persona, but what about the personas that aren't in the production workspaces and don't have the grant permissions on these resources? Um and and also, it's different to govern a
[06:25] single app versus a thousand or ten thousand, right? So you need to kind of change the construct, and that's why we released app spaces. Um Yeah, I was going to ask you about that. So app spaces is a new concept. We didn't have this last year, and it's um it's serving like a lot of different purposes. And so maybe you can talk a
[06:40] little bit, there's like the security aspect, but also just um you know, what's the concept in general? Like how should people think about spaces? Yeah, so app spaces So first of all, apps are still deployed in a workspace, but a lot of what we heard is, "Hey, like I can't give these citizen devs,
[06:55] you know, full access to this production workspace." So it's not separate from the workspace. What it allows you to do is kind of curate a sandbox in that workspace and pull together different permissions. And not just workspace permissions, also, you know, UC permissions, app sharing permissions, um
[07:11] network permissions. So for example, another common issue um is, "Hey, my production workspace does have access to these external resources and external APIs, but I don't want these apps to have access to that. So I need to lock down the workspace even further." And that's where the concept of app space
[07:27] came is, you know, you things within a workspace. Within a workspace, yeah. And curate permissions, like things that are governed by UC that aren't necessarily workspace centric. So how do you collect all those policies together, define that as a container, and then let people innovate in that container. And so I
[07:43] think platform teams are excited about this because again, now, you know, they feel that pressure. They want to unblock the teams from innovating, but it doesn't have to be the wild west, and they can define these policies, they can define the app space, um and let people innovate within that.
[07:59] That's awesome. Yeah, so uh one of my favorite parts or the best part is customer successes and customers getting value, uh whether it's from the keynote you gave or your own personal experience. What are some favorite uh apps that you've seen deployed? Yeah, I mean, it's I I I said this in
[08:15] the keynote, but I'll reiterate it here. It's just it's my it's the favorite part about my job is um I get to go in and, you know, get demos from all these customers that are building. And the breadth of use cases is just it's crazy. Um Yeah. And what we've been doing now, and I
[08:31] think this is really cool, um we've been doing more of these hands-on workshops. So basically app hackathons where we go in, um you know, we have a structured program around this. We do some training and enablement, and then we kind of encourage people to ideate and build. Um and what's amazing is like they'll
[08:46] usually come out of that with 20 different ideas for how to improve their business apps. Um and it's really quick to go from these ideas to prototypes and all the way to production. So I just I'm inspired by kind of the breadth of what gets built out there. Um it's it's truly amazing. And I think what's cool is this could
[09:02] only be done in Databricks because these are not just apps like that are just static hosting, right? These use data and AI and models is bringing together the best of the platform. I mean, so personal plug. So I I founded a Databricks Labs a number of years ago, and I still manage the pipeline of all
[09:18] these new Databricks Labs projects. And before I was using spreadsheets, and I had my different bits of data in three different places. So I I vibe coded an app, I hosted on Databricks internally, and now I can basically just manage that whole pipeline, and then I I plug the the backing store of that, which is Lake
[09:34] Base, into my downstream pipeline to help manage metrics. So So thank you because that's made my life a lot easier. But, when I built it um I didn't have I I only had to use like an external coding agent, but I think now I I could do that all within Databricks with uh with Genie App Builder, right? Genie App Builder, yeah. Yeah, we're
[09:49] very excited about that. Um and and really kind of the the special sauce of Genie App Builder is the fact that it does have all that enterprise context. So, it's built on the Genie ontology. It understands your data. It's not just the data. It understands your Databricks environment that you're building this up
[10:04] in, right? Um so, it understands the permissions and the policies of the app space and really kind of helps you build within those confines, right? And that's where we think it's um different and very complementary to external coding tools, right? You can still use external coding tools to vibe code up prototypes in the
[10:21] front end and kind of get that initial structure there, and then you can use Genie App Builder to really vibe code that to production, right? So, is that like lovable replica personal assistant? exactly. Um you know, those tools are not designed to vibe code in the production Databricks environment. Um
[10:38] they're fantastic at building these apps and they're very complementary to what we're doing. We uh we came out with um you know, last year with partnerships with all those, right? Exactly. there's they've still been a big part of the whole ecosystem. Absolutely. And um and really the way to think about it is Genie App Builder is
[10:53] targeted at that kind of citizen developer that needs this vertically integrated solution with permissions, governance, and vibe coding all-in-one. Um we still want to make it super easy for anybody that wants to use their preferred coding agent to be able to do that and deploy these apps. And so,
[11:08] we've built skills and MCPs and plugins and um we've invested in this ecosystem and we really want we just want to make it super easy for anybody to build an app, whatever tool you want to use. Yeah, yeah. That's awesome. Yeah, I really like that fact that you can basically just stay within the the Databricks environment now uh using Genie and then auto create
[11:26] this app, which before would have taken, you know, months to to build an app. And now it's just literally within minutes. Yeah, I think it's going to I think it's going to open up the floodgates in a good way. Yeah. That's why that's why we package these things these three things together where you have the the native vibe coding with
[11:42] Genie app builder, you have the governance with app spaces, and now with this new serverless micro apps, you have the infrastructure for really low-cost apps. They they love what you just said. Yeah, they love it. They can't get enough of it. But uh Croatia just tied it against you.
[11:58] Yeah. Oh. Or they were cheering you. I don't know. I will take the cheering. Yeah, yeah. Um But yeah, all three go together. They're very complimentary. Um you need to you can't sacrifice budgets and governance just because you want to open up to citizen developers. Yeah, so we have like a minute or two
[12:13] left. I think importantly is you we've just excited a lot of the audience. Like how do they get started? Are there resources? Or like where do you go to start building? um first of all, we launched a new dev hub. Um so developers.databricks.com. Check it out. It's incredible. Developers, developers, developers.
[12:29] Developers. I won't go that far, but yeah. Uh but uh check it out. There's uh prototypes. I think at this point we have um over 30 35 prototypes already up there um that you can immediately copy a prompt into your coding agent and deploy it. We
[12:45] have all these prototypes are open source. It's just a great way to go from zero to one, learn, and then build some momentum from there. That's amazing. Well, thank you so much for your time. Um I really appreciate having you here, and I'm I'm really excited about all the new functionality coming out with apps. Um
[13:01] and I can't wait to get my hands on it. So Yeah, thanks for for coming on board.

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.