Skip to main content

Scaling Data Access with Databricks Unity Catalog and Immuta

Summary

  • Vizient built a data access framework combining Databricks Unity Catalog with Immuta's policy enforcement layer to reduce thousands of manual access policies into one scalable, attribute-based governance model serving both human users and AI agents.
  • The framework's four pillars — speed, transparency, consistency, and compliance — enable self-service data access with automated request workflows, eliminating access bottlenecks while maintaining comprehensive audit trails.
  • Agentic governance extends the same policy-driven access controls to AI agents, ensuring that as AI workloads scale in healthcare, every data consumer operates with consistent, auditable permissions.

Scaling Data Access with Databricks Unity Catalog and Immuta

Watch: Scaling Data Access with Databricks Unity Catalog and Immuta
As healthcare organizations scale AI, the biggest barrier is often not the model, but governed, scalable access to trusted data. this video explores how Vizient built a data access framework that treats data governance as a strategic capability, combining Databricks Unity Catalog's centralized governance with Immuta's policy enforcement layer to support both human users and AI agents with consistent, policy-driven access.
Learn how to reduce thousands of manual policies into one scalable framework using attribute-based access controls, automate access request workflows, and implement agentic governance. Discover the four pillars of governance (speed, transparency, consistency, compliance), how to achieve fine-grain access controls at enterprise scale, and practical strategies for enabling self-service data access while maintaining security and audit trails.
🤝

Chapters

FAQs

What is Immuta and how does it complement Databricks Unity Catalog?

Immuta is a data security platform that adds a policy enforcement layer on top of Databricks Unity Catalog, enabling attribute-based access controls at enterprise scale. Together, they allow organizations like Vizient to replace thousands of individual manual policies with a single, scalable governance framework that consistently evaluates who should access what data and under what conditions.

What are the four pillars of Vizient's data governance framework?

Vizient's governance framework is built on four pillars: speed (making data available quickly without friction), transparency (ensuring users understand what data they can access and why), consistency (applying the same policies uniformly across the organization), and compliance (maintaining audit trails and meeting regulatory requirements). These pillars guide both the technical architecture and the organizational approach to data access.

How does Vizient handle governance for AI agents?

This video describes Vizient's approach to agentic governance, which treats AI agents as data consumers subject to the same policy-driven access controls as human users. By extending Immuta's attribute-based access controls to AI agents through Unity Catalog, Vizient ensures that agentic workflows operate with appropriate, auditable permissions as AI adoption scales across the organization.

What is Vizient's 'right data, right user, right time, right tool, right way' principle?

This five-part principle reflects Vizient's core philosophy for data access governance: ensuring that the correct data reaches the appropriate user at the right moment, through the appropriate tool, and in a compliant manner. The principle drives both the technical implementation of access controls and the organizational culture around data stewardship at Vizient.

Full transcript

[00:08] Good morning everyone. Thank you for being here at the data and AI summit, the largest we've ever had. Uh hopefully you all are in the right session. We've got the title up here on the screen. Some folks that showed up earlier for a Lake Flo declarative pipelines paid training which has been relocated to room 2011. So just getting that out of
[00:23] the way. For the folks on the corners and in the back, if you want to move to the front, I would love that. We're going to be running microphones around later at the end of this for Q&A. Um, so just save any questions that you have until the end. But again, thank you for being here and I'll introduce you to the speakers today. We have Tyler Ditto
[00:40] who's a principal uh success manager for Ammuda and we have Charles Chaz Hatfield who's a senior director of data governance at Vizant. Both fantastic partners at Data Bricks. Excited uh to have them take it from here and share a better together story on what they're
[00:56] doing with data bricks. Um, so I'll hand it over to you guys. Sounds good. Thank you all for coming today and let's get started. I'm going to take a step back here and get to our forward-looking statement. I am a senior director of data governance and I support my legal
[01:12] partners in my organization. So, please pay attention to this. Let's move on. Um, then we have the complete your surveys. Please do this and provide the feedback for the speakers and all of the attendees. Giving that information means
[01:27] the speaking get better at data bricks year-over-year. So the more feedback you give, the better the speakers are going to be at data bricks and those things. So we're here you're here for how vizant AI transformation starts with scaling data across with Amuda.
[01:44] If you want to know about our life histories, this is here and then let's get started. Okay. Free the data. That's one of our mantras at Vizant. We don't want to be the blocker of the data. We want the data to be available
[02:03] and usable by the organization. That's the whole point of having data is to make good data decisions because you have it available to you. How do we do that is what we're going to talk about a little today. and Ammut is here with us
[02:18] to talk about how now that we've set this framework in place, we're going to be able to accelerate in the AI world. So the things I want to point out here is this mantra of right data, right user, right time, right tool, right way.
[02:34] Those are the four things. Right data, right user, right time, right tool, right way. That allows you to do all of the things that we're going to need to do in this new world. If you don't have those things solved for, if you don't have those things accounted for, if you
[02:51] can't measure those things, you can't go faster. People will want you to go faster. and they're going to make you and drive you and push you and do all of those things. But if you don't know the data you have, if you don't know the users
[03:06] who are accessing it, if you don't know what tools they're using to get it, and you don't know the time frames that they need access to it, then you can't govern it correctly. And if you can't do that for a human, you're definitely not going to be able to do
[03:22] that for an agent. So, we have stickers somewhere. I have stickers that we hand out. So, if you want a sticker, I've got one for you. Anything? Uh, did that. Okay.
[03:38] Yeah, sounds good. Okay. So, let's talk about our journey here. So, those things that we said in the beginning about we're in healthcare at Vizant, so patient rights is a pretty big thing to us. And so that's our kind
[03:54] of data rights that we kind of use in that context. That is the things that we define through our scale. So if we're going to transformation in terms of using data providing clinical outcomes improvement imply complying and making
[04:10] the ability to have our go segment provide better products in a more timely manner then we going to need that data and we need to know what's in our data and what's available. So if it's there we'd start up this chain we start
[04:26] climbing the mountain. So we want to be able to trust and govern at scale. We get a lot of data from a lot of health care systems. And if you know who Vizian is, you know who Vizian is. If you don't, part of that is that's how we
[04:44] maintain our trust with our health care systems is we're not selling our data to anyone but the health care systems themselves. We're giving it back to them so they can make better outcomes, improvements. our data is for them to make patients lives better
[04:59] or it's there to make sure that the GPOS the supply chains have access to the materials they need without being stopped. So we have to have that trust. Then we move on to data access as a business capability. They don't want to know that
[05:15] they can't access that data. They want to know how they can access the data. And then AI amplifies that cost. If you're inconsistent in those things, AI will give inconsistent answers back.
[05:38] So those four pillars in this in the center, you're seeing probably more of the data governancy language, the AI governancy language that goes with those rights that we talked about earlier. And those are the things we really have to have in place to have the ability for you to have a user journey or an AI journey.
[05:56] And those paths are very different in terms of what context the user is bringing to the use of that data. Let's talk about humans because we're all humans.
[06:12] And when we see data, when it gets put in front of us, when we get access to it, we can interpret context. We actually have built inside of ourselves our own meta data catalog. We have that information. We've trained ourselves on
[06:29] that stuff. Someone doesn't necessarily need to tell us something because we have years of experience with that data. Someone could ask us to go, hey, will you define this column for me in this table, in this schema, on this catalog? We probably could. We might not get time
[06:45] to do that, but that's another story for another time. Then we have the we tolerate delays. We understand that the timing of it is negotiable, right? We don't want it to be. We want everything to be real time. What this is doing is really saying,
[07:01] okay, I understand why this data is coming in at a different point in time than this other data. So I can negotiate and figure out how okay I can make my valid outputs my outcomes can be targeted from a period of time because I can make sure that the data lines up in
[07:16] that piece and then we work around issues. We have shadow it. It happens, right? That's what's going on in the world because we have to deliver the data to get there. And hopefully what we've done with our
[07:31] framework is allow those things to slowly be less risk to the organization. Right? On the AI side, we have it requires precision. You have to train your models with that metadata because they don't
[07:49] have years of experience with it. They haven't lived the dream of the data. They haven't delivered it to people. So, you have to start accelerating that. It requires real-time access. So, it'll start returning decisions that aren't
[08:06] timely because it'll say, "Oh, well, this is really good date, but I didn't realize that this other data was three months old. it might not catch that unless you've trained it and made sure the models are looking for those things. And the scalability issues really there's not a
[08:22] scalability issue, right? It scales. We've seen it scale. It's scaling right now. The metadata that it needs to scale is what needs to be applied, what needs to be there for that to happen. And just a call out here that I'll come
[08:38] back here to a moment. the the the main themes here are speed, transparency, and consistency. And having that in a framework and tools that can actually go deliver on that is so important.
[08:54] So what's Vizant's answer? There are some people in the room that work at Vizant and they'll tell you that well, you can ask them how this framework is really going and they'll give you some good answers and some bad answers because that's what happens with any framework. you go through the growth of it. But what we're trying to do is
[09:10] centralized policy enforcement, right? We are a shop that uses a lot of data bricks, but we also have a lot of older systems that are there that aren't being brought up to data bricks, but we still need the data in them. So, we need
[09:27] to be able to think outside of that. And we need a place where the policy is applied. If that data exists in data bricks and that data can be grouped with other data outside of data bricks, we need to be able to centralize that policy so the same policy is applied to that data no matter where it resides.
[09:44] Automate access workflows. We're going from a world where we need to grant access based on risk, not based on a person hitting a button to say yes or no. Right? So that means you need to have the framework in place. You need to have the flow in place so that you can
[10:02] define that risk and you can say you know what CHAZ has approved access to this the tables in the schema 97% of the time it's been requested. Well do you really need Chaz to push the button anymore or can AI take that over because
[10:19] it says 95% of the time you're saying yes. Why don't why are you stopping it with human in the loop? And then consistent access patterns. And that goes back to if you're consistent for humans, you can be consistent for AI agents. You can be consistent for
[10:35] whatever's coming in the future. All of this is changing. What is a user? You got to answer that question. What is a What is the data you're looking at? You've got to answer that question. What are the tools you're
[10:50] going to use? You got to answer that question. Let's see. Oh, I'm I'm right on time. I'm doing good. Okay. So, this is what it enables for us. More
[11:06] self-service access. Right. The framework exists. Teams come to the data governance team and says, you know what, there's a new schema that's been developed at our organization. Can you put the governance around it? Yes, we have a framework. We follow the rules. We set it up. We put it into the systems
[11:21] for how those requests happen. Our teams can go and now make that request and it goes to the right people and it goes in a timely manner. And yes, do we want it to be faster? We do. But we're healthcare, we're conservative, we like
[11:36] someone, a person saying yes. But when we're comfortable, when we've matured and we don't need that person to say yes because we've identified the actual risk of them needing to say yes or no, then we'll be able to take the step forward because the framework is in
[11:51] place. more consistent governance. Well, we really have spent most of our time in the governance world on exceptions, not on the rules. Right? Someone agrees to the rules day one and then day two,
[12:08] well, but I really wanted to do it in this case this way, right? Well, we have to steer them back to the right way. That doesn't mean the first initial setup was right, right? they now have a more context to say what they want their workflow to be. So great, let's go
[12:25] change the workflow. Let's not make exceptions. And this leads to better scalability, right? If you can do this once, then you can do it twice. And if if the second time you have new
[12:40] parameters into what you need to think about, then you change the framework. You don't change what you implement. The framework is not something that stays in place forever. It grows, it lives, it adapts to your organization
[12:56] and then that becomes a repeatable enterprise capability. Next one. Yep. Perfect. Okay. Yeah. So, working with Chaz and the Vizant team, they provided the the paintbrush and the art for what they wanted to do. They
[13:12] laid down a framework for how they wanted to govern their organization. Uh, and I I can't stress how critical that is. So, over the course of my career at Ammuda, um, I've done delivery for over 70 70 organizations. And I will tell you like that's the most critical part. Someone coming to the table with a
[13:27] vision for what they want to do, how they're going to govern access to data, how they're going to comply with their regulations in the rules and legal that they have in their both their industries, but the organization they're in. Um, we provide the medium. That's what that's what Ammuda is here. It's a
[13:43] policy enforcement layer that uh Chaz and uh William who's not with us today and and Kevin Crawford were able to work with us to go to deliver on this framework and actually put it into action for many consumers, governors, uh DBAs, data sewers, engineers, all the
[13:59] people that interact in an ecosystem to be able to make this whole thing possible. So I'm going to talk about some building blocks here for how they deliver on where they're going in order to get to this this northstar which is really this AI transformation. How do you unlock agentics in your organization
[14:16] put the power of data into your consumers and people's hands so they can go do their job create market differential in organizations and succeed in their industries and especially when we're talking about uh Vizian's vision here with how they work with uh healthcare organizations around the world.
[14:33] So the first thing I'm going to talk about here is what we call policybased access controls or pback. Um this really amounts to three different things. Um many people in the in the crowd I'm sure you've heard of arbback or rolebased access controls. I see some head shaking out there. Um more recently we heard uh
[14:50] over the course of the past couple years databicks released what we call attribute-based access controls. That's embedded in what we do with Ammuda. That's how we started. uh immune as a policy engine from the base layer and that foundation is what really allowed
[15:05] uh the Vizian organization to move so quickly especially when we're talking about rules and I I just want to comment there is we had to take the organization from an arbback strategy to an aback strategy before we could apply policy
[15:20] right so the journey wasn't that the tool couldn't do it the rule the issue was I had nested groups that I had to break in En intra so that I could actually apply the data and use policy. Exactly. And that piece is what
[15:38] allowed them to reduce a significant amount of overhead and scale so quickly. So we took thousands of policies that Bizant had that delivered access all over the organization, a variety of different rules that deliver different iterations of essentially the same data
[15:53] set. Maybe it's it's one schema and a couple tables from another. Maybe it's a catalog and half of another, but you start to see these like differences that kind of patterns that come across the board. And being able to go deliver on that uh without the tooling that we that we delivered with with Vizant was really
[16:08] really difficult. So what do we do? We took we took thousands of policies and boiled it down to just one. They used variables to do matching with attributes and metadata with users to in order to deliver access across the organization
[16:24] in a very scalable way. And just to talk about that is that one policy is not one type of entitlement, right? We have five different types of entitlements that we use through that engine. Because what we're thinking
[16:39] about is that the entitlement is really how do you match a user with the metadata of the data? And so we could take all those thousand
[16:54] policies, identify our data with the metadata that how we want it to be consumed, identify where we're going to get our list of users and how we group them together. And that's the one policy. Take this person identified with this
[17:11] data that's been identified and now they can see each other. Exactly. So we talked about the rules part of it here. Uh, but Chaz also mentioned exceptions and while we want to get away from that, every organization we see the exact same thing. Um, again like I' I've seen this
[17:28] over 70 times before. Every single organization that's out there today has request workflows and working with Chaz, we heard it come in a variety of different places. Maybe it's a phone call, uh, maybe it's a team's message, maybe it's an email. Uh, none of those are ideal because it's all diverting a
[17:44] person back to another place. And what what I hear when I see that is I hear I see desperate processes and disperate processes means time wasting and thinking about coming into an organization and really have no clue where you go. How do you get access to
[17:59] data? Do I call Tony that's three seats over? Do I call Chaz? Do I call William? He's in our governance team. That's a really challenging process to go to especially when you're trying to ramp people but more importantly like get everyone to be rowing the same boat and moving the organization the same place
[18:15] and that's why we release request workflows and are deploying here with with Jazz and the Vizant team today. What work request workflows do is it brings consumers and governors closer to the data instead of someone going to say service now or sale point or giving
[18:31] someone a call or sending an email to get access to data in likely due to uh asking for a role because Steve across the row said that's what role I need to access this data. So really challenging circumstance to be in and I can't count how many times I've heard people say I
[18:48] requested access to this role and I got it but I still don't have that actual data I need. So what do we do with this? We we allowed individuals consumers to request access to actual data assets and data products. So those assets may be tables and views could be schemas and
[19:04] cataloges. But also, when you have curated products that are out there today for specific missions, people can go request access to those exact products, fill out data use agreements, and then ship them to the stewards or data owners that are out there. And I'll let Chaz talk about like
[19:20] how critical this part is because I'm seeing some head shaking here, and I know a lot of data owners out there in many organizations today deal with the struggle of trying to determine, am I actually allowed to give someone access to this? I have a very little information. How do you make that
[19:36] determination? So part of that journey for us is who actually owns the data. So we we have kind of three arms at Vizant. We have a go, we have um our clinical outcomes, the benchmarking that we do and then we have consulting arm of those
[19:53] things. So imagine as we're going on our data bricks journey, all of the pillars, all of those different data spread out across the organization starts coming on to data bricks and we're going to move into that place. What happened before
[20:09] was oh well you're not even in that domain to request access. So that was how we were controlling access. You couldn't even get to it. But now we're putting all the data in one place. So who owns that data now? We have to go find those owners. They have to tell us
[20:25] what they own. They have to tell us what goes with what. Then we go to our Aback. We start tagging things. We say, "Oh, this is a strategy around a data set." That's an entitlement type we have. And that means
[20:40] that we are tagging tables across the whole catalog and schema range. Now, is that the most mature way to do it? Not necessarily. But it is a way that we've done it in the past within our organization and we need to be able to onboard that have a strategy to bring
[20:56] it into data bricks have a strategy to have that control on top of it so you can bring it on and then start sharing it with the organization instead of the one pillar who could see the one thing. So we talked about peback and we talked about request workflows and that really
[21:11] builds the foundation because this entire thing are building blocks to get to this AI transformation and this agentic journey that we're seeing so many organizations chase after and that's where like viz or vizette and chaz and I really started to work on this is how do we go deploy uh GPT or
[21:29] cloud or layer genie rooms into organizations and allow them to take advantage of it allow consumers to come to a single place and gather insights without having to run SQL. This is a really really big deal. I can't count how how often I hear organizations with
[21:44] really really large citizen an analyst groups with teams where some people know SQL pretty well. A lot of people are pointing and clicking and trying to gather insights where peback and request workflows come into place at with agentic governance on top of it is we know who the person is.
[22:00] We know what data they're allowed to have access to and we're we're able to deliver that in one place. Someone can go and ask, I'd like to know what my top five customers are in this area by spend or maybe it's tell me a little bit about the efficacy of certain drug, whatever
[22:17] that may be. They can gather those insights and request workflows are extremely important in this in this aspect as well because someone can go ask for an insight and maybe they're delivered back a part of a table. Maybe there's some rows redacted. Maybe there's some columns masked as well that
[22:33] they're seeing in this in this chat interface. That person can then interface with an agent and request access to the data that they're not seeing or the insights that they need. And that agent on behalf of them can go make a request of a human approver. The
[22:49] human approver can see whether that's an email, maybe they get a notification in Teams and they can view the information around why they need access, the purpose for what they what they're going to use it for. and approve it right there on the spot and that agent notify that consumer to
[23:04] be able to do it right there. So what do what do we all talk about here? We talked about uh data being being delivered via access on rules, us dealing with exceptions and a consumer getting insights to be able to do their job all in one place without
[23:20] having to switch context. And I can't explain how big of a deal that is. I see a lot of people's head shaking in the crowd right now. Context switching is the largest cost in any organization today that goes completely unnoticed. And all these things allow you to go
[23:37] deliver that. And if that wasn't enough right there on on your own, every organization, especially companies in HCLS like Vizia need to be able to prove compliance. What happened? Did a user access data in data bricks? Did they request it through
[23:52] quad? Were they in a genie room? Did they run a query in PowerBI? No matter where that is, we can tell you exactly who did it, did they use an agent for it, what rules or policies were in place, who approved it, who denied it, all in your audit logs. And we layer a
[24:07] chat interface on top of that. So those compliance people can go ask questions. Who had access to this data at this time. What queries did this person run on this day? What agents were used during this week? So my my Vizian, I'm sorry guys, we don't have that yet, right? So like but that's what we've been doing
[24:24] this the journey that my my the people that have been going on for the data access framework means I can now start doing that because I've set all the foundation so the only thing I need to do now is start turning things on because I can control the grains for those agents because I'm treating the
[24:40] agents like a consumer right so on our that journey is that was our goal we've cleaned up the things we need to do. So, we can do that and we can do that today. And my my boss is
[24:55] here and I'm I keep pushing on him that we need to do this today and we're going to get there. So, I think we're good for the next one. Next one. Yeah. And and and how do we do all this? So, um we'll talk a little bit about Unity Catalog as the the data foundation
[25:11] here. So thanks to a really really strong partnership between Ammuda and data bricks we created a very tight integration that allows us to embed those security rules and those policies in the access security layer of unity catalog. Additionally where those scenarios where
[25:28] we're writing um what we call fine grain access controls or data policies policies that mask and row redact data we're writing those UDFs for you. So you can do it at scale. You can write one policy that maps or redacts all the data
[25:43] you need based on someone's region or who they are, maybe what job function they have or what they're using it for. All those things are very very critical because it provides a pane of unity and glass across the entire organization. No matter if you're a DBA or an engineer,
[25:59] you can see these rules in data bricks. If you're a governor or a data owner, a steward, you can see this in in Amuda. And it goes far far beyond just that. And I think that's where there's so a lot of power at at the Vizian organization as well. So yeah, if we take that what you just said and what we want to do and we go
[26:15] back to what we were talking about in the beginning, right data, right user, right tool, right time, right? What we're doing is we're taking the right data and we're using Unity catalog to build tags. So we can put that and
[26:30] attach it to the data. We're going to Entra. We have octa but we're focused on entra with the data bricks. So we go there to identify who the people are, who the service principles are because we're treating them like a person
[26:45] and getting that lined up and then that comes back around. Then Ammuda Ammuda just takes those two points and says, "Oh, these two things match based on your policy. I it sends it to data bricks and says here's the grant and if I take someone
[27:01] out or put someone in, I don't have to go to data bicks. I don't have to go to Ammuda. I just have to make sure that the identity has the permission and has been associated with the right tags. And for those organizations out there that are still going through that
[27:17] journey of migrating data to data bricks or to a lakehouse or a new warehouse, we integrate with legacy data data stores as well. Whether that be SQL server or Postgress SQL, we empathize with the situation a lot of organizations are in. They're also being a requirement and a need for those tooling. So you can write
[27:32] one policy, any platform, anywhere, any time to deliver the exact theme that that Chaz started in the beginning only if you put the metadata on it. Just make sure we're all on the same page there. Right. Next one. Yeah. And then lastly, like how this
[27:49] architecture works together like I said like the the partnerships that we built with data bricks and a variety of the tools in the ecosystem. uh immuniz uh we what we need is a is a data access framework like Chaz and team put together that's what makes it really
[28:04] valuable we provide the medium to go apply that we integrate with the tools the cataloges the metadata and the users that are out there today and then we work with them and consultate on how to go do that successfully and again like we have a
[28:19] team that's seen this very often we've had the pleasure of working with Chaz to go deploy this and uh they've been able to reap the benefits. We're really looking forward to continuing this AI transformation with them.
[28:35] Do we want to I think you covered this or do you want to skip it or Yeah, I think we covered this. Okay. Yeah. Okay. So, business impact, right? Where we've accelerated today is only a part of the journey. So what we've accelerated today is when our dev teams
[28:51] build new things, they don't have to worry about implementing the security around it because they know, oh, I'm going to put it in this schema or this catalog or this place and that is already a place recognized by the data access framework. So I just need to inform the data governance team that
[29:08] that is data I want to make available to the enterprise. We put it through our framework. We work with the owner to say how do you want to identify that your data is unique to the other data sets so that you can own it and then you can approve it.
[29:24] You basically build what you want to build and then we make it available based on how you want those rules. Do you want them to go through some training? Great. We'll go check if you taken the training in the framework. Do you want to go and say, "Hey, have you do you have permission for this client?"
[29:41] Great. We'll do that. That's the place where we're have some pieces in place, but that's where we're maturing because the framework allows us to do that and we don't have to think about how am I going to do the policy later. The policy is taken care of. I
[29:58] really need to get my data owners to own their data and tell me and that's a whole another conference to talk about. Yeah. And to simplify it, it's it's all about speed and creating frictionless access to data. Create a user experience
[30:13] that people want to go back to. Don't make it challenging. When people find a a place and in a process that's easy and simple and transparent, they want to use it. Where we see people divert and go do different things and call people is when they're experiencing challenge or lack
[30:30] of transparency. And that's what we're trying to provide to the to the organization today. Again, free the data. You can only have free data if you're transparent. Sometimes you don't have good answers. Sometimes it's not nice what happened and you and what you have to fix. But if you're transparent,
[30:47] you're open to it, you explain what you're having to go through, and sometimes it's because our framework failed and we've got to fix what our framework was doing. Sometimes it's something else, but we're transparent about it. And then the trust is there, the improvement of the framework happens,
[31:04] and you continue to move forward into this new world with AI. Okay, call to action. So, I'm going to go really fast on this slide so we can get to some questions if you have them. So, clear access
[31:19] framework. We are pretty transparent. We know what it is. There's users, there's data, there's tools, there's time data bricks platform foundation. This works really well on data bricks because of all the different pieces. It has
[31:36] tagging. It has Unity Catalog to support it. It has the ability to start grouping things in. It has the ability to make workspaces that make that even crazier than we want it to be. But again, we're trying to focus on the cataloges themselves because we care about data
[31:51] access, not compute. And that's one of the things that we do limit some of our scope to. And then Amuto policy enforcement. I know in the future I'm going to have more than one policy in Amuda. It's going to happen because I'm going to start doing things that are
[32:08] dynamically happening at the data query time and those are going to be different rules and different reasons for those to happen. We have one because we made the transition to a thousand down to one to establish the framework. And now it's
[32:23] well clean is not the right word, but we've got a good handle on it. So when we grow back up and we start adding policies and we get very specific on what we turn on and off, we're able to do that. Uh and then the last message here, just we're going to step pause here and go
[32:40] through a little bit of Q&A. So uh we'll answer some questions as well. Um after this, uh Chaz and I will be just across the street at the Howard. We have an Amuda lounge and we'd be happy to answer more questions around how they built the framework, how we implement it, how we deployed it together, what it looks
[32:56] like. So, if you guys have questions, if you guys are dealing with some of these things in your organization today, we'd be happy to talk with you and walk walk you through what we did and maybe there's some similarities we might be able to help you out with along the way. So, um, with that being said, I think we
[33:11] can open it up for QA here. Yeah.

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.