AI-Powered Data Discovery and Curation With Unity Catalog

Summary

AI-generated metadata: The Discover Page Experience automatically enriches data assets with AI-powered metadata for faster, more accurate discovery across Unity Catalog.
Natural language exploration: Users can query data lakes without deep schema knowledge by using conversational, natural language data exploration in Unity Catalog.
Certification and deprecation workflow: New certification and deprecation features bring clarity to complex data environments by marking trusted assets and outdated data.

AI-Powered Data Discovery and Curation With Unity Catalog

Talk by Hongyi Zhang, Staff Software Engineer, Databricks ; Peter Wang, Staff Product Manager, Databricks
In today’s data landscape, the challenge isn’t just storing or processing data — it’s enabling every user, from data stewards to analysts, to find and trust the right data, fast. this video explores how Databricks is reimagining data discovery with the new Discover Page Experience — an intuitive, curated interface showcasing key data and workspace assets. We’ll dive into AI-assisted governance and AI-powered discovery features like AI-generated metadata, AI-assisted lineage and natural language data exploration in Unity Catalog. Plus, see how new certifications and deprecations bring clarity to complex data environments. Whether you’re a data steward highlighting trusted assets or an analyst navigating data without deep schema knowledge, this video will show how Databricks is making data discovery seamless for everyone.
Here’s more to explore:
Unified and open governance for data and AI: https://www.databricks.com/product/unity-catalog
See all the product announcements from Data + AI Summit: https://www.databricks.com/events/dataaisummit-2025-announcements

Chapters

FAQs

What is the Discover Page Experience in Unity Catalog?

The Discover Page Experience is an AI-powered interface that automatically surfaces and curates key data and workspace assets, using AI-generated metadata and lineage to help users find trusted data without manual schema documentation.

How does AI-assisted lineage improve data discovery?

AI-assisted lineage automatically tracks data relationships and transformations, eliminating the need for manual lineage entry and helping data analysts understand data provenance quickly.

Can non-technical analysts use natural language to explore data?

Yes, natural language data exploration in Unity Catalog lets analysts query data using conversational language without requiring deep technical knowledge of underlying schemas or SQL.

How do certification and deprecation features help manage data quality?

Certification marks trusted, quality-verified datasets while deprecation warnings flag outdated or problematic data, helping analysts make informed decisions about which assets to use in their workflows.

Full transcript

`0:00` Introduction to data challenges

All right, welcome everyone. Um, I know folks are still trickling in from the main keynote and uh, yeah, so uh, we'll get started. Um, but you know, hopefully can folks will continue rolling in. So, welcome everyone. Let me start by posing a question to the audience. How many of you guys have gone into a data lake and had trouble discovering or finding a right data asset? Just raise your hands. Okay, pretty much everyone, right?
Uh what about how many of you guys are in the position where you know you put data in but it's just like such a headache to document the data put the like you know lineage information in and then you know and maybe you even document it really well in another platform but then your data analyst still comes to you and be like hey what's in this what's in this data all right quite a lot so today we're going to be talking a lot about how um AI I powered data discovery and curation works in data bricks and how those capabilities can help you solve some of those challenges in your workflow.
Today before we dive in a quick disclaimer we'll be discussing some road map items and also forwardlooking features that is obviously subject to change right so um we always try to make things better based on customer feedback but that also means some of the things that we plan do change so just a heads up quite a lot of legal ease there but so hi everyone I'm a product manager on the unity catalog team focused around discovery and lineage.

`1:20` Data curation and discovery

Hi everyone. I'm a staff software engineer from data bricks and I work on kalora data uh discovery linear generates AI integrations. Happy to be here today and after the presentation we'll be sticking around. So feel free to come to come talk to us. We'll answer some uh questions. So let's begin by imagining the perfect data lake, right? Um, ideally it has well doumented data and it's clear exactly what that data is about, where it's coming from, and how you can use it.
Um, you want to have good data quality and that there aren't any issues in the data pipelines powering those data. The environment should ideally provide enough information to provide easy self-service data discovery instead of relying on data gurus or information silos or even be dependent on some other separate documentation platform. And finally, the entire thing should be secure and well-governed. But reality is often pretty far from that ideal, right?
As the data volumes grow and the complexities of that grow, um things tends towards entropy. Data documentation coverage starts to decline, data quality starts to degrade. You start to have a lot of information in there that's like not really maintained. Um discovery gets harder, signal to noise ratio gets worse over time and data governance can either become too loose, too permissive or too tight where essentially a new user code goes in and they can't find anything.
So today we will be exploring two key areas where data bricks AI capabilities are transforming how you work with data and curation. First we'll be looking at data curation challenges facing data stewards and how AI is helping them scale their efforts. Next we'll be examining how data discovery experience from the perspective of uh users and consumers change with how these new AI features and how it's simplifying their journey.
Finally, we will see some of these features in action in several demos. Let's start with the problem of data curation that data stewards face. First, there is a significant amount of overhead in maintaining high-quality metadata. As data volumes grow exponentially, keeping that documentation up to date and comprehensive becomes increasingly difficult. Second, monitoring data quality is inherently complex. It's not just about varying the uh verifying the presence of data, but it's also about understanding freshness, completeness, and ensuring the upstream jobs are functioning as expected.

`4:01` Data curation challenges

Detecting anomalies and preempting issues are often, you know, quite manual and timeconuming. And in fact, even if those issues are detected, a lot of that information is not really shown to the users at the right time. It's often sil and when documentation exists, it's often siloed in platforms like Confluence or SharePoint disconnected from the surface in which users are actually interacting and using and finding the data.
Essentially, they miss the critical context uh right as when they need it. Finally, the data stewards are constantly struggle to a challenge of data management. Right? You must strike a balance between securing data and also enabling good self-service data discovery, granting the right level of access without introducing risk or slowing the teams down. It's a nuance and really delicate effort. Let's walk through how data bricks is solving some of these challenges.

`5:31` AI-powered metadata enrichment

To address the challenge of curation at scale, we now offer AI assisted metadata enrichment, automated data classification, and automatic capture of insights and behavioral metadata based on usage. As a part of Unity catalog's capabilities, AI generated comments and descriptions can help you automatically document Unity catalog objects at the table, column, schema, and catalog level. It draws on existing metadata to reduce the manual overhead and help users quickly understand a data set.
This can greatly reduce the manual burden in both documentation and curation for data steward and data engineers. We also provide automatic data classification using data bricks profiling. Data bricks can intelligently detect and classify sensitive data across your catalog. This eliminates the need to manually tag sensitive data and then review the access to that sensitive data. So let's take an example of how traditionally this will work.
You load some new data into data bricks. It contains some sensitive information. So some data steward needs to go in and tag that information and then also manage the permissions on who can access that, who can query that and you know what is offiscated. But with this new set of capabilities and how data bricks uh automatic classification is integrated with attribute-based access control, what you can do is define your policies once and have that automatically enforced for all the new data coming into your platform.
So for example, now your new workflow is you load data into data bricks. This automatic classification feature will automatically tag those data as sensitive and because of an existing data policy enforcing who can access the sensitive data, it will automatically be offiscated when certain users are quering that data set. Another important layer of information that is helping users help uh understand and contextualize data is the behavioral metadata.
So, insights generated based on how people are interacting with data, including things like lineage, popularity, query patterns, and related assets. These insights really help users assess whether a data set is relevant and trustworthy and how it's typically used by others in an organization. With Unity Catalog, we automatically capture and surface this type of behavioral metadata. So users can see how that data set is being uh how that particular data set being is being used across the organization.
Like all Unity catalog features, this is fully fully permission aware, meaning users only see the insights that they're per uh per uh allowed to see. And because behavioral metadata is captured at the query and compute layer, we can even generate these insights and capture them from federated sources into Unity Catalog. As long as the queries are routed through unique catalog, you will be able to automatically document those insights.
Now, let's talk about the challenge of maintaining informing users about data quality at scale. With data bricks automatic data quality monitoring, you can monitor tables for freshness and completeness uh on data quality issues. It detects it not only detects issues, but it will actually pinpoint the root cause and perform some root uh root cause analysis for you. When upstream jobs fail or when periodic refreshes to a particular table fails to materialize, we will automatically tag that data as being uh has having a data quality issue.

`8:39` Data quality monitoring

By integrating data quality insights right into the catalog, we can ensure that the users and the data consumers are proactively alerted to data quality issues without needing to context switch in a different platform or subscribe to some complex notifications. We also provide a central dashboard for reporting data quality issues. This helps teams move from a reactive posture to a proactive large-scale monitoring of the data quality across your enterprise.
Issues that we detect are prioritized by impact using some of the behavioral uh insights that we shown you earlier like lineage usage and um and asset popularity that's captured in Unity catalog. So for example, if a source table feeds many downstream assets, core dashboards, we can flag it as a high priority and surface the likely root cause so that you know exactly what to prioritize and how to fix them. Another common challenge that we hear from customers is that while Medallion architecture allows you to separate data by refinement levels, it's still not really clear exactly which data sets are truly vetted and ready for broad consumption.

`10:12` Certification and deprecation

To address this, we are introducing certification and CL deprecation features designed to improve the signal to noise ratio and make it easier to highlight trusted data and flag outdated or unreliable assets. With certification and deprecation, data stewards can now clearly mark trusted assets as certified and outdated assets as deprecated. These tags are prominently surfaced in the catalog search and seamlessly integrated into both exploration and authoring surfaces.
Uh so all users can benefit. For example, certification deprecation directly influences search ranking and data set recommendations across the data brick service. Deprecation signals are also prominently displayed in authoring tools. That means users are alerted even while working in their own environment. Similar to getting a GPS warning about, you know, traffic on the Bay Bridge or on your way to work, you get a heads up of the information right when you need it so that you can steer clear of problematic data.
So in example here, you can see, you know, we select a table, but that table is deprecated. So we have a very prominent strike through to say hey maybe you shouldn't be quering this table. These signals are also integrated in with our attribute based access control. So you can even implement a soft deprecation policy for example and any data set that you know I deprecate will have restricted ability to query that way you can deprecate asset and before well before you actually remove it from your system.
Finally, how do you balance the need for user self-discovery with the need of securing enterprise data from unauthorized access? Metadata browsing and uh access request capabilities are designed just for that. Giving users visibility into what the data what data exists without exposing the underlying data itself and providing a path to govern access via a request for access when it's needed. So over the past year, we introduced uh browsing and request access capabilities to help broaden data discovery without compromising security.

`12:01` Access control and security

Users can now view rich metadata, including the AI generated context and other insights without being able to query the underlying data. Once they determine a data set is relevant, they can then request access directly to the relevant approvers. those access requests then are routed to the right individuals and we can even integrate that with various systems. So for example if you're already using approval flows in Jira or or service now we can integrate with those external systems.
This approach combines the security with the ability for users still maintain self-discovery uh in your enterprise data environment. So we talked a lot about challenges facing data stewards. How about the other side of the coin? How what about the challenges uh faced by analysts and data users during their exploration journey? For many users coming to data bricks, one of the big challenges is finding the right data. We recently introduced smart suggestion system to address some of that problem.

`13:17` Data discovery for users

The smart recommendation system is designed to streamline your workflow uh either by helping you to return quickly to the data you care about or recently been working on or uh surfacing relevant assets based on uh things uh that you might find useful. It uses signals like recents, favorites, uh and usage patterns to generate personalized suggestions. Think of it as picking up where you left off uh without having to dig through complex technical hierarchies.
Beyond just smart recommendations and rankings, we're also introduced a new rich detailed page on each data bricks asset to surface those signals that we captured earlier. All that work that goes into curating a high-quality data asset, adding documentation, applying certification, tagging and governance only works if the users can find it at the right time and understand it. That's why we redesigned our catalog pages to showcase some of the most meaningful insights right on the overview page right when the users need them.
These insights include certification deprecation status, the data quality indicators, the descriptive tags, owners and steward and steward information as well as a concise summary of the top insights that we captured such as query volumes, top users, related assets. By bringing those cues together all in one place, users no longer need to hunt down each bits and pieces of information across multiple documentation tools or different platforms.
And you can assess whether a data is uh trusted and fit for your purpose. Those social and contextual signals are critical in building a uh building trust and usability for your end users. And they can help users un answer questions like is this data actively maintained? How widely is it used? and can I trust the results if I query against this data set? With the richer overview, we're not just surfacing metadata. We're empowering users to make fast, confident, and responsible choices on what data they can trust.

`15:44` Natural language exploration

We also recognize that many users, particularly analysts, want to quickly get an answer without having to write complex queries. So, we've integrated quite a lot of natural language capabilities from the data bricks assistant right into the catalog product. This is exactly why we introduced natural language data exploration on the sample data page. Once you've identified the right asset, thanks to an improved object detail page with rich metadata details, you can immediately start exploring it using simple natural language queries.
Powered by data bricks assistant, the experience delivers intelligent contextaware responses. The assistant doesn't just look at the data, it understands the context, right? It draws from all the previous behavioral insights from other queries, column descriptions, table metadata to deliver relevant answers and generate the right query. So instead of writing SQL, you can ask a simple question like what are soft uh what are the top selling products this quarter and you'll get the results instantly.
So it even suggest follow-up questions based off of how this data set is being used elsewhere. With this capability, we're not just making data more accessible. We're also making it much more conversational and intuitive. The assistant isn't also limited to just exploring data or generating SQL. Um, it can guide you through how to use the data platform uh more effectively offering offering best practices on data bricks on data governance and even help you discover relevant data asset in natural language search.
So now we walked through those capabilities. U metadata enrichment, smart recommendations. Um let's bring it all together and show you how it works in action. Yeah. All right. Thanks. So I will take it from here for the demo. So Peter just went through a bunch of cool improvements we have done. Uh let's see them in action for the whole picture and to see how that can help you make your data discovery journey and curation journey easier.

`17:25` Practical product demonstration

So first I want to take a look at from the admin perspective. So imagine I'm a data admin working in an e-commerce company and managing the data pipelines as a data admin. My job is to curate and maintain my data so that everyone in my work can easily consume it under security and governance control. Gladly uni catalog already offers it. So let's look at this catalog I maintain. So first you can see this catalog has a really good description that I already worked with AI together to come up with.
It's in good markdown style h has a lot of rich context. I kind of borrow it to advertise a little bit about the catalog features but you can see its capabilities right how rich it can offers to my uh underlying consumers to read about and let's take a look at the schema page. So opening up one of them I kind borrow the descriptions to highlight all the key tables um with some overviews and to see what tables does which so that my uh my data users they don't need to you know go through all the tables look at again and thanks to the data quality uh monitoring capabilities uh that I enable at the schema page uh now all the data quality is automatically captured and then you can see I have quite a few table that's healthy all my key tables looks good I'm pretty happy about it and I can see everything here on the schema page without you know uh going back and forth about it.
Let's look at take a look at the products table that's very important for my company. The products table as here you can uh document it. It has a lot of key um informations. It has product ID name the metadatas for my product lines right inside my company and also uh again healthy quality states. has some quite good usage I created before the demo and now let's check out the full insights as you can see I created before the submit so the usage pattern kind of picked up write a few queries uh generated some Q2 analysis here and it's already automatically captured here so I feel good about my product table let's certify it so that all my consumers will see it with higher confidence uh I also want to kind with you guys together on another table the transaction table.
So as you can see in the description it has a core fact table that captures all the consumer purchase record all the uh record in good healthy state as well um and also has all the good column comments that are I I already created and some uh insights pattern that already surface up one thing I want to you know introduce here is that the erd diagram so within this diagram I think that's the best way kind to help you understand the key relationship ship between these tables with within this data sets.
So my transaction table has a foreign key that points back to the product table that we already see that's how they are connected together. So again feel good about this. Let's certify it so that my consumer they can view it with high confidence. And then I want to show you guys how I created those tables with good documentations with AI. So with this new session raw table I just created uh then the AI suggested description kicks in and then I accept it like immediately you have data documentation but that's not enough right?
I want to add some like my own context and make it you know uh even more rich with like my data admin feelings about it. Let's ask AI to add like a how to use section within good style. So immediately it has some like border plan uh play that I'm pretty happy about. Um but there's a lot thing more things you can do with AI. For example, let's say I want to add some sample query code so that you know my uh customer my users they can take a look at it.
So can AI takes the mission execute pretty fast. Um yeah and all the we want all the u AIdriven comments to be reviewed and accepted by the admin so that this is not something random right this is human reviewed but powered by AI so I made some edit then let's save it again um there is actually one more very high uh uh asked feature that is about translation so a lot of things were done is here in English I'm not a Spanish speaker, but with the help of AI, I can translate like everything, but retain the code in English for sure.
That helps all my like Spanish speaker users to be able to read this table as well. So, I'm pretty happy about this saved back to UC so that everyone can view it and reviewed approved by myself. And let's take a look at the column side as well. Uh, which just showed, we just introduced this bulk generate and bulk safe capabilities. So previously if you're familiar with this feature you needed to do like one by one per column.
So now we you are able to generate for the whole table within two two clicks. One click to generate one click to save it. And that works for all your columns. And of course you can make more uh edits towards it. But we're hoping with this approach we make your uh data admin uh journey with a lot more efficient. Okay. So now let's transition to the consumer side. Uh let's keep it simple. As a data scientist, I just want to find the right data, right?
And I just want to show show you a few ways to start the data discovery journey. So first uh smart recommendations on the landing page. I land on the catalog experience. Here are all the things like we we track all the uh signals and think that's relevant to you. You can as you can see I work on these like demos a lot already. Um so it's it already able to surface you know all these default catalogs uh default schemas uh right after that so I don't need to dive into the whole trees hierarchies to go for to go for towards it and another thing is that inside your sidebar we have the recent and favorites builtin so you don't need to go like a lot of context switch you can just find it to the left and open it to the right you can see I can easily jump back to my catalog demo catalog pretty uh easily And on top of that, um, we would just talk about the request for access, right?
We introduced this new filter that you can control like to what extent you want to find your data. Sometimes you want to expand to see more stuff. Sometimes you only want to limit to things you can query, right? On top of that, if you see like for example this customer data set, I find it really interesting and it's certified and you can even request access to it. You can add a message to your admin to explain why you need it and then this request will be sent in your way like your admin set in set certain ways and then you will be able to um get the permission like under all all the governor's control.
Uh lastly u of course one important way is a search right you always find the stuff via search. So we now have the certification integrated with search. You can see that my product table is boosted because like it's certified and has a lot of usage on it. Okay. So now I find the data uh thanks to the rich text uh information uh rich detail information here like you know as a consumer I can read through how other people are using it feel good about it now let's see like in live for the sample data exploration right without even writing a code now I can ask question about it I have this product table I just want to see something and then the suggest a good question what are the bestselling products in revenue right let's try it out and see how it works So let's run it.
I think the system responds really fast with a SQL to you know um to explain the like you don't need to even write it and it just generate. So you can see that it runs very successfully and one thing surprised here is that it was even able to join with the transaction table we just talked about even though I'm like just querying this table. So I think this served as a good foundation for me. So let's you can just use the generated SQL and continue your um analysis journey like in your notebook uh SQL editor as the way you can go deeper analysis.
So we're hoping with the help of AI we can pave your way to set the like initial discovery with a lot of ease. Yeah. Okay. There's actually one more thing. So additionally uh that's not everything we talk about data data bricks assistant. I'm not sure if everyone uh familiar with it can always help through the journey as your data copilot. So start your data bricks assistant on the top right like bring a panel you can set where you like it.
Uh so you can actually also find your tables via assistant without the context switch. So let's say let's find the product table again and then it is actually be able to show up on the right hand side as well and there's like a suggested few actions you can act towards it. For example, like I always go with just open this table for me and maybe I'll check out a a few more impro uh a few more inspirations. And the other things you can do is that you can always always attach to this message.
So then the system knows the context and then we'll try to answer based on the attached object. A few other things for more inspirational usage is that you can say suggest some questions for me, suggest some queries for me. So we're hoping the data assistant as a AI copilot is not only like solving a problem but sometimes more inspirational sometimes it tells me if I'm new to a data set right it can quick me quickly help me understand the data sets um in natural language as well so you can also do show related insights you know it will grab the insights for you I think that's pretty cool and then you can explore full insights if you want and um there's actually I One um more thing about it's not only working for your table right like all your catalog objects like uh catalog uh schema all the stuff so you can ask questions it has the context so try and sometimes I even use this as inspirational and use this like generated result to contribute back to the description right so I personally find that pretty useful but I'm pretty sure you know you guys also have a lot of innovative ways to interact with the datab bricks assistant that's integrated to the product.
Lastly, there's like a little trick I think it's going to quickly show is that you can use this at uh command to tag and use the object. So then it attach to the direct assistant. It will be able to kind answer questions for the object you're tagging without actually you know you need to open it and all these things. Yeah. Cool. So I think that's it for the um demo part and I know it is a lot to consume to recap for all the features improvements that we have shown.
So from the engineering perspective right I like to show you guys how uh uni catalog builds the foundation of the data intelligence platform for you and for your company. So I think the number one takeaway is that uh context matters and we already have all the context built into the UC uh as knowledge base right for the large language model. So with the help of UC metadata content large language models are able to understand your needs seamlessly.

`28:41` Best practices and engineering

As you can see in the diagram um we have like your your usage patterns, lineage activities, data bricks and all these contribute to the data intelligence pl uh platform. So that um um for example with the AI generate comments we use like your related table schemas metadata as context to help you to enrich your uh unity catalog context in a fast way and on top of that once you save back the stuff to use it to the data intelligence platform uh it also help with your all your downstream applications like genie once you're in notebook writing and dashboarding as a result the whole ecosystem become smarter. uh yeah so what we improved for the for the AI capabilities so some of you guys might notice there are a few principles we try to improve and following so for one we have invested into speed and quality for the AI generic comments and all the cap AI capabilities um for generation speed we have improved by faster model faster serving and overall the speed now is 2x faster and we also improved quality by a lot the second point is that we want to make your AI assistant contextual.
For example, we put your catalog contact integration uh in line with data assistance assist assistant so that it save you back and forth copy paste like you don't need to do this and help you boost your productivity. And the last point is that we believe the best AI should be easy to use and integrated into your system. As you can just see, we make the quick data exploration with natural language super fast. you don't need to context switch going to different surface just for starting your query.
Uh so lastly I would like to give some tips about how to better utilize AI capabilities and make best of unity catalog and help you build into the future. So first uh I think the number one takeaway is that we strongly recommend you guys to add more descriptions to uni catalog objects specifically for tables and columns and we actually expanded the AI capabilities offering to or unity or unity catalog objects including catalog schema and even clean rooms for example and also we hope with the new bulk column generation and save it help you to make the journey a little bit faster and with that it will help you uh help all the downstreams to be AI powered like genie editor assistant etc.
And as a second, as you just saw the inline edit description, I think that's pretty cool, really powerful. It's now integrated with AI and supports like translation markdown that I already show, but I'm pretty sure like it can cater your need properly. Uh, and the third thing is that we have this little button that can execute it directly within the system with the sample data exploration. So you don't need a context switch to go back and forth.
You can get some quick answers. And yeah, we're looking forward to your feedbacks. Uh, please send your feedbacks to help us improve. Yeah, back to. All right, so I think there's one more thing that I'd like to share which some of you might have seen already announced uh during the keynote. We've heard a lot of customer feedback around the how our current Unity catalog interface is very technical, right? it's it's really built around hierarchies and it's not really designed for um exploratory search but rather you kind of have to know what you're looking for to get to the right things.

`32:34` Discover page and conclusion

So we have introduced this the discover page to try to solve that problem. Again in many of our conversations with customers we've heard one common theme you need catalog is really powerful but it really feels too technical for many of our business users. And while the data hierarchy is really useful if it's clean and structured, it's not always intuitive for somebody who isn't familiar with exactly where things lie in the schemas and and cataloges.
In fact, some data leaders told us they've been really hesitant to open the data bricks um catalog broadly, not because the data isn't ready for consumption, but rather the experience isn't accessible enough for users, even even SQL analyst users, right? um who are not familiar with the technical schemas. That's exactly why we introduced the discover page. It's designed to be welcoming front door uh to your data estate.
It's an internal marketplace-l like experience where users can browse, search, and explore the most valuable data assets and insights your organization has to offer. Whether it's certified UC assets, key metrics, or dashboards, or even Genie spaces, discover page brings together all of that into a single unified page. You don't users don't need to know exactly where that data lives. They can simply explore what curated data your data lake has to offer.
Through a combination of algorithmic recommendations and human curation, discover page helps highlight the most relevant and trustworthy content. It blends data intelligence with human expertise to uh to scale data curation across your organization. And most importantly, the discover page lets you organize and present data in a way that mirrors how your business works by domains, functions rather than technical schemas.
This shifts from a system centric view of the data to a user centric view. Um, which is going to be key in driving adoption, boosting trust, and really making data truly self-service. So, let's see what that looks like. Here I have the discover page. Um I can see that right now there is a set of domains that is aligned to how my business breaks down either organizationally or functionally. There is a set of feature dashboards, tables, genie spaces and metric views that has been curated either h by humans or you can see a little sparkle here by algorithms.
If I go into my set of feature dashboards, I will see that there is a set of dashboards that's been uh uh certified in my organization. Those will show up first in this experience and then you will also see the ones that are suggested algorithmically. We use various signals to approximate you know whether or not this data is widely used and should be fit for consumption. And so that way you don't necessarily need to have the human curation um to especially if the data state is super uh is is sprawling but you still can get the benefit of finding the most critical asset for your organization.
Similarly, all of this is built on the same permissive governance um a certain permissions and governance off of your Unity catalog. So, this isn't just a separate um surface that you have to control with a different set of permissions. If your user has the browse permission to see an object, it will show up for them. If they don't have browse permission, it they won't be seeing those assets. And similarly, this is all integrated into the request for access.
So you can have a set of certified objects that is shown broadly to your organization but if the user need access to them they can request access to a central authority which can grant them that permission. So um that's uh what we wanted to show um for today. Uh and again thank you guys for coming to the presentation. One quick call to action is that you know we will be launching discover page um in private preview in select regions starting next week.
So if you're interested please reach out to your um data bricks account team and we would love to hear your feedback on that. So thank you very much and uh we look forward to your questions.

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.