Tech Industry Session: Building Collaborative Ecosystems With Openness and Portability

Summary

• Open ecosystem acceleration: Leading tech companies use Databricks' secure data sharing and partner integration capabilities to accelerate growth through collaboration.
• Built-on solutions platform: Organizations create scalable data products by leveraging Databricks' openness and portability to integrate with partner ecosystems.
• Secure data collaboration: Data mesh and Clean Rooms enable trusted data exchange with partners without exposing raw data, maintaining security and governance.

Tech Industry Session: Building Collaborative Ecosystems With Openness and Portability

Talk by Adrian Bolosan, Sr. Director Data Collaboration , Databricks ; Bob Pisani, CTO, Addepar ; Davis Matson, SVP Platform Engineering, Health Catalyst ; Matthew Houser, SVP Engineering, Tealium
Join us to discover how leading tech companies accelerate growth using open ecosystems and built-on solutions to foster collaboration, accelerate innovation and create scalable data products. this video will explore how organizations use Databricks to securely share data, integrate with partners and enable teams to build impactful applications powered by AI and analytics. Topics include: Using Delta Sharing for secure, real-time data collaboration across teams and partners Embedding analytics and creating marketplaces to extend product capabilities Building with open standards and governance frameworks to ensure compliance without sacrificing agility Hear real-world examples of how open ecosystems empower organizations to widen the aperture on collaboration, driving better business outcomes. Walk away with insights into how open data sharing and built-on solutions can help your teams innovate faster at scale.
Here’s more to explore:
Open data sharing for data, analytics, and AI: https://www.databricks.com/product/delta-sharing
A new approach to data sharing: https://www.databricks.com/resources/ebook/a-new-approach-to-data-sharing
See all the product announcements from Data + AI Summit: https://www.databricks.com/events/dataaisummit-2025-announcements

FAQs

How does Databricks enable secure data collaboration?

Databricks provides secure data sharing mechanisms, Clean Rooms for governed data exchange, and Delta Sharing for controlled partner access—enabling collaboration without compromising data security.

What is a data mesh architecture for collaborative ecosystems?

Data mesh organizes data ownership by domain, with each domain managing its data as a product—enabling organizations and partners to discover, share, and build on each other's data products safely.

Can partners build solutions on top of shared data?

Yes, through Delta Sharing and Databricks' open ecosystem, partners can access shared datasets to build applications, insights, and services without storing data locally or breaking governance.

How do Clean Rooms support complex collaboration scenarios?

Clean Rooms provide secure computation environments where multiple parties can analyze shared data collaboratively without exposing underlying datasets—ideal for cross-organizational analytics.

Full transcript

All right, thank you everybody. Um, day four, day five for some during lunch timeish. So, I appreciate everybody that made it out for the session. Um, I'm senior director of data collaboration, go to market strategy at data bricks. Uh this is one of 40 data collaboration sessions actually we had this year. So uh that's thanks to our marketing folks and everybody very excited about this especially based on the guests and partners that we have that uh you guys didn't come here to listen to me you came here to listen to them.
So they're the they're the poll in the room. Um so real quick we're gonna I'm gonna talk just a couple slides kind of highlighting a lot of the announcements from today. So, I was glad the timing of the session based on mate going earlier and kind of helping team me up quite a bit. And then we'll get into uh a discussion with from Adapar and from our friends over at Telium. And at the end, will come up and wrap us up with uh some cool stuff.
He has to talk about data collaboration specific with health catalysts. Already said who I was. All right. So, real quick, I just I pulled some cool metrics I wanted to share with you guys all for those that are not familiar. Uh, this thing in the center here that looks like a Death Star is actually all the active edges and Delta shares on the network. So, I pulled a live shot of all the connectivity of our customers and partners actually sharing data on data bricks on Delta Sharing.
So, some really cool stats. Year-over-year growth in Delta Sharing, just over 260 uh% amazing, right? We'll blow that out this next year based on the adoption. But the other metric I wanted to call out we you know part of the session is about openness. So Delta sharing we offer our customers the flexibility to share with those that you know not beyond just their different clouds and different regions but they don't necessarily have to be on data bricks right and we actually measure that in a metric we call you know data bricks to data bricks DTOD or data bricks to open.
But what's really interesting talking about the flexibility here, 40% of those shares are actually data bricks to open, right? So, um, so it's just a cool highlight I wanted to call out. Um, you guys probably saw some of this. Mate had great gave an amazing talk this morning on on the keynotes. Um, but I kind of I rewarded this today because I was like, hey, it's data and AI for all, but it's really data collab data sharing and collaboration for all.
So, uh, with the free edition, I told my dad, I was like, "Guess what, Dad? We get the data share." And he's like, "I don't know what you're talking about." Um, but I was like, "You will, Dad. We're going to share family photos and all this cool stuff." Um, but there was there was two big announcements, right? It's building on the foundation of what we already done. Today, Delta Sharing is the only open protocol for uh for delta sharing, right? uh or for data sharing but two big announcements was data sharing on Apache iceberg.
This has been a long time coming. Um and probably no surprise based on the tabular investment and everything last year. Uh so again it's going to be you know regardless of cross or cloud region regardless of format right that we're really building towards this openness. And then the last piece is the delta sharing uh network gateway. So how are we making it easier on lakehouse to share uh seamless with the network configuration.
So a lot of hard work behind the scenes with our product team and everything to make these things all G with the announcement. So I just wanted to highlight those things. And then there's this concept of data gravity and it was really in that in that uh image I showed you guys of all the connectivity. It's our customers that are sharing with their customers, their vendors, their partners, and it starts to create this data gravity.
And we really see this around use cases across industry. Obviously, this is the tech thing that we're focusing on. And what's really cool about tech, Maline and I were going around this is like, you know, you add tech to certain things. So, you got fintech, you got health tech, we got adtech. And so, we have a good range of our customers up here talking about this. But ultimately at the end of the day I just view delta sharing as the mechanism or the capability that's really powering you know business use cases that are critical to all these uh industries and when we go through the discussion we'll we'll touch on some of these um probably more than some of these but that's really underneath that delta sharing is just the mechanism it's cool and it's modern it's great but really what is it doing for our shared customers uh for our customers ourselves so with that I have special instructions to push some buttons up here.
And I'm going to invite my guest up here in one second. Just get this going. All right. And with that, uh I'd like to invite uh from Adapar. And for those that uh on the keynote yesterday, Adapar was actually f uh featured on on the apps. It was really cool. So I was just like, "Hey, I'm going to get to be with. , you want to join me up here?" And then got from Telium. Yeah, I think line with the picture. Yeah, in line.
We're going to line up with those pictures up there. All right, but I'll be in color, not in black and white this time. This is all right. Perfect. All right, to kick us off for one, thank you guys. Right, long week, day four. Appreciate it. Uh let's start off with a quick round of uh introductions, a little bit about your role and a little bit about your organization. Uh so we'll start with you. Oh great, thank you.
Uh it's Sane, chief technology officer at Adapar. Adapar is a global uh tech and data platform for wealth and asset management. So uh we uh serve industry firms from RAAS, Fentley offices up through enterprises and institutions. Uh I've been at the firm five years and uh just happy to be here and a longtime advocate of data bricks. Awesome. All right, over to you. Yeah, thank uh thanks for having me. Uh so. I'm the SVP of engineering at uh Telium.
Been here gosh, five and a half years now. It's gone quick. Yeah. Uh but Telium is a CDP. Uh we're traditionally in the marketing space. What that is is we we kind of sit between your experience layer and then your low-level system layer. Um we try to create like unified views or profiles of users whether you're coming from online or offline digital things like that. But we're a globally distributed very high throughput uh collection consented and enriched data experience.
So it's creating profiles and for marketing to drive business value. Awesome. Yeah. Thanks man. A little plug for TM. So we recently launched data intelligence for marketing. Telium was a launch partner for that. You guys are also, if I'm correct, you guys are a leader in the Gartner Magic Quadrant for CDP. We are. We actually, I think, created the category of CDP. I think we were the first U, but we were uh Gartner Magic Quadrant two years in a row.
So, yeah, good stuff. Awesome. Well, again, good representation here and, like I said, will round us out with some health tech stuff, but we got a good mix. I think that was the goal of this, right? So, we're going to start off with some of the questions. So again the topics around data collaboration uh openness uh portability uh so how have you s how have you successfully fostered culture and collaboration and accountability as you scaled across um you know your team and data exchange in general.
Yeah. So uh with both your clients and with your team internally you have to think about creating trust uh and that starts with stewardship. So this is really this idea of having an owner who is responsible for that data set for not just sharing it but ultimately ensuring its timeliness, its quality and its availability. Um also it's about getting the team to think really delta first. You know normally we often think about API first but I think delta first is becoming a standard we're really adopting is thinking what if this data has to go to a client or we want it to go to a client or from a partner.
Um and I think the other aspect of accountability is really around having very transparent metrics. Yeah. Where you actually think about how the data is being used uh who's accessing it but ultimately using that to then inform your choices because you know we have I heard one of the sessions talk a lot about uh you know graveyards for dashboards for data and all the rest. The same thing can happen with delta with delta. you know, if you're sharing a lot of things, you don't know what's happening to it, then that ultimately doesn't really inform, frankly, accountability and it's not particularly collaborative.
So, we use the metrics to inform a lot of that thinking. And ultimately, when we start to build and design, we think about Delta as an API interface. Yeah. And really think about a way to more seamlessly and frictionlessly share data with clients and partners. Awesome. Might might steal that actually in the future. Um, so, so can you describe a time when you know governance or cost control policies faced internal push back?
Probably never heard that before. Um, and how you resolved that tension both like at the executive level and also with like the engineers that uh, uh, that you manage day-to-day. Yeah, this is a this is a good one. Um, governance is very, you know, paramount for us dealing with a lot of customer data. Uh, I believe there's another question. We'll dig into governance a little more. I'll focus this one on um cost control policies and kind of dealing with some of that things.
Um I think it was important you know you need to reset expectations u with clear modeling projected cost and your actual cost. You know we had an instance where our projected was really high. So we need to go in and do a lot of cost optimizations and tunings which um you know what we'll touch on later about some tips and tricks. Um got a lot to say there but basically what we did is you know change some of our ingestion pipelines to be multi-tenant. um like a lot of the Spark parameters that could be tuned, things like that.
Um matching instance types to the jobs and the workloads, just making sure they match the things that can easily lead to cost overruns. You just really got to get in and tune. It's it's a very rich environment, so you need to need to pay attention. It can it can skyrocket quickly. So, yes. Yeah. Awesome. All right, we're going to go back to you. So what design patterns or UX considerations kind of drove high adoption amongst internal users and external consumers?
So you kind of touched on this a little bit earlier, but yeah, it's it's related to uh frankly being collaborative, but you you have to think about usability not just from a user interface perspective, but even from a data perspective. And so a lot of that comes down to making it clear what data is available. this idea of discoverability both internally in your catalog is also if you're sharing and then you ultimately want to give someone access to the delta uh tables and sometimes we even share and create workspaces for our clients and for our partners and in there we want to have them see the catalog and understand what's there.
Um we have to have we think of it as having to be very client aligned. So, and it's probably the same as with the partners, which we use more inbound and clients more outbound, but it goes the other way as well. But client alignment here is this idea that you're not just designing one sizefits-all potentially. Uh the nice part about, you know, the model of the medallion model of gold and gold tables is that you tend to think of it as more as creating a view that works for that client because that's really going to frankly drive adoption. and the easier it is for them to consume without having to massage it, without having to figure out, well, how do I actually take this data and join it to 30 other things and then have to post-process it, frankly, the easier it's going to be to actually consume, which is all part of user experience.
And so, you know, and then there's this other idea of how do we make it more like a subscription where when they get that data, they're not just really signing up for one delta table. We think of it as a set and we're constantly adding to this offering and so we want to treat it like a growing set of channels if you will that they're going to subscribe to us with and each serve different purposes. So it has to be well designed because a lot of times we think of the unit of design as the table and then the fields in there.
We have think of it as a set of tables and is really a data ecosystem that they're plugging into. So that's part of how we engage with clients what they want. They want to come to us for quote unquote everything as their data as their single source of truth. So, how do we present something that fits for works for them that's most natural? And that's really good UX. And I like what you said in the beginning. You reminded me something I say a lot often, which is a little cheesy, but I always say data collaboration starts with collaboration because a lot of times we'll get request saying, "Hey, our customer wants to delta share with each other and we're like, well, what are they trying to do and like what's the use case and what's the" And it basically opens up all these questions to your point and it's like almost responsible data sharing, right?
Um, it opens up, you know, to your point, it's like that's just the beginning of potentially larger data sets and larger tables. All right. Um, all right. We're going to jump back back to you or to you. Yeah. So, on this one, uh, so how do you manage versioning, access control, and lineage at scale in a marketplace of data products and AI models, especially in an area where you deal with so many, again going back to your point, so many customer sensitive data.
Um, yeah. So, this is kind of touch back on the governance and things. So um we rolled a lot of this before MLflow by hand. So manage versions do it in GitHub and all this and we've since you know are in progress migrating you know to data bricks using ML flow which will help and then unity catalog is a big one as you know is as you grow you get data silos everywhere and you know we deal with a lot of customer data GDPR and all these things but you need to go and delete the data well before unity catalog how do you know where where all it is you know so that's really made a big difference putting things in unity catalog and some of the features that we're looking forward to you know exploring later with like PII masking and role level access and some of those things.
So, we're we're just getting started on Unity Catalog, but it's it's in MLFlow as well, but we're um very excited about the being able to offload this to data bricks. I don't have to manage it and keep it up ourselves, you know. So, awesome. We listen to you. Yeah. All right. Um Bob, so how do how do you address security and compliance when opening live data pipelines? Kind of goes back to what you were saying earlier, right?
Uh to external stakeholders, right? Like there's some view data sharing as high risk. A lot of us know that's just part of doing business. We've done it in certain way what I would argue is far less secure ways in the past. Um and this is not unique to delta sharing. Uh it certainly opens up a new avenue or vector that you have to ultimately make sure you're you're handling correctly. We've had to do a lot a lot of this already.
Uh first via API or maybe even file sharing and the like. But uh you know the you have to believe in this idea of compliance by design and you really have to think of it from the ground up. So it's how you provision um a share. It's how you ultimately make it available to clients. A lot of it has to be part of frankly using what's native to data bricks and del sharing and permissions and access control. Um but I think it's assuring that you think about it all the way down again put it put that hat of the client on which is how are they going to consume this data?
What do they expect to be in there? um what should be anonymized, what should be masked. Um does anything have to be aggregated because they can't actually see the fine grain control which is not necessarily just a permission setting. Uh you also have have a lot of auditability and understanding and managing it's relative to managing change as much as anything else but who is accessing this but who's using it, who's changing it, um how where is the data frankly ingressing or eressing from and really tracking all that.
Make sure it's part of your observability and monitoring. Um and then the last piece I would say is this idea of you have to have kind of a a compliance enforcement model. Um it's sort of a nice way of saying it's really good QA. Uh which means that you know we do a lot of work where both as the data is um being written out and we're creating the data that's going to go into a share as well as as we're sharing it. We also run a lot of compliance checks and jobs to make sure that the data doesn't have any sort of leakage or PII that shouldn't be there. and we ensure that we're not sharing the wrong data even if we know that it's set up perfectly.
We run a lot of post QA checks and ensure that things are valid because the expectation are that whether you're getting data via delta, via API, via an uh for us and a slew of inapp products and tools, they expect the data to be consistent and they expect the same safeguards and compliance to be in place regardless of the channel or medium they're getting it from. And that kind of Q&A I assume is cross team, right? is you have governance teams, product teams, that's a collaborative effort essentially on your side, right?
Oh, yeah. Our our engineering and product teams work closely with compliance, legal, risk. So, it's uh it's it's it's a whole gang that gets involved. Awesome. All right, back to you. So, when you think about how do you measure success like KPIs around how you're measuring open ecosystem data collaboration initiatives that you guys are driving, like what are those that you track? Um, and how do you monitor that on the platform around health cost, you know, and efficiency?
Yeah, that's a good one. Um, I'll be brief on this one because I know there's some good ones coming uh kind of for the sake of time, but we look a lot of things like cost cost per DBU. Um, as far as performance, you know, we we're a very high throughput system. We process billions, tens of billions of events a day. So, um, ingestion latency is very important, error rates, just making sure we're not dropping data, things like that.
And then um for some of the models and our ML product is API latency. So it's a lot because we're real time and it's got to be fast and performant. Um so it's you know seconds down is correct. That's a lot of dropping on the floor. So you and I you and I joked around on Monday like real time in your world means different than real time and others based on the decisions you're making. Real time to some of our partners is it tells us they tell us slow down.
That's not we can't handle that speed. Like this is not real time. Real time's not five minutes for us. It's it's now. So yeah. Awesome. All right. So, I think we're going to get into ones I think are geared for this audience. I'm excited about these. So, this is I'll start with you, but you guys are both going to answer these. So, what's one piece of advice that you'd have or you'd offer organizations uh just beginning to build open portable ecosystems on on data bricks?
Okay. Uh besides just do it. All right. Uh all right. So, uh don't try and invent the wheel on governance. A lot of it's built in by data bricks which is great. Uh they've they've designed Delta and Delta sharing and the whole portability and openness from the ground up which is great. So so make sure you leverage that which means yes use Unity catalog use permissions use everything that's already built into the platform.
Um I I like this idea of curating for consumption which is the same way I said earlier about API first. Think Delta first. You have to basically think about consumption as you create data sets, as you create products, as you create different um frankly, you know, derivative offerings or whatever it may be. Think about how that's going to be consumed. Think about the information you need to include and do it upfront. Don't try and solve it at the end when you go to share it via delta table.
Um, and then the other piece which goes somewhat to versioning and and whatnot is proactively managing change. have a a cadence with your consumers and producers about how you're going to manage that change. Some of it is like API version management, but it's also about uh kind of the just because you delta share it and there's z almost zero friction. Yeah. Doesn't mean you don't have to deal with boring things like notifications and deprecation and emails and promotions and all these sorts of things.
There's a lot you have to still manage. Um so take those things into consideration and you'll get all the benefits on the uh the tech side. Awesome. , same question advice. Um yeah, I you know unfortunately we had a few missteps on our journey. So some lessons learned. Um if I had if I had to go back and start over um define some clear use cases. Data Bricks is very big. It's rich. It's diverse. You're not going to do it all at once.
Don't do that. Um but anyhow, get a clear use case. Start small. Validate. Get some early wins. Build some momentum. Uh, I think that would help a lot because the other thing I will say is if you don't have expert expertise on staff, maybe go find some. There's a lot of partners out there. It made a world of difference for us when we found a solution an SI that really helped us. Again, there's there's a lot of features and tricks in there and you're not just going to fumble your way through it.
So, it's um that made the world of difference for us is getting some help and it's helping our own staff. It's helping our own staff to come up to speed. So, um I would definitely look at that. Yeah. And there's a rich ecosystem like you said. We saw all the partners here. I think the uh those that specialize on on data bricks that audience is growing. So there's no lack of partners willing to help. Right. All right.
So looking ahead, right? It's a little bit of magic ball question I get. But um so you know what's emerging when you guys think about open standards or portability initiatives that you're prioritizing and how would you prepare your engineering organization to adopt them? Like what do you see on the horizon around that? Let's start start with you both. Um I think a lot of the announcements uh over the last two days during the keynotes and otherwise have set a clear bar as far as the openness.
Um I think whether it's Delta or iceberg, it shouldn't matter anymore. And frankly, when we think about these broad platforms and broad ecosystems, I I I I try and tell my team to just think ubiquitously that if we can get anything at the data tier, it's just a much richer, deeper integration. It removes all the friction and ultimately it just makes it more open. So I think leaning in on not only what data bricks is creating but just more broadly uh in this data ecosystem and saying all right whether it's uh Salesforce or SAP or or frankly Palanteer or Snowflake just think of it as almost that data fabric data mesh that ultimately you can just tap into and you don't no longer have to think about building the next integration.
It becomes far more seamless. Um I suppose as you know as far as portability um looking ahead I mean I I know AI is probably going to influence this space because it's a question of are you going to ultimately continue to share the data or do you want to share the insights and answers that come from that data. It'll be interesting to see how that evolves. Um that's not necessarily a a standard per se but I think it's a pattern that's going to influence this space heavily.
Fair. Yeah. No, this is a good one for us. Um, so in the CDP space, a big topic right now is what's called zero copy where where the customers maybe don't want a copy of their data to live in our database, right? It makes sense. But if we want to do something like a create audiences or create segments, we're going to need some historical data to be able to do that. So, you know, customers that put their data with us, we keep it for about 12 months, let's say, but that's I don't want to copy your data, right?
Maybe they don't either. So, how do I go? So, I'm looking at delta sharing as a solution that if they can share with us, then we can re reach in and get the query. just give me the list of users that, you know, spent $1,000 in the last six months and are left-handed, whatever. Yeah. Right. But they give me those IDs, but then we can then activate those for some type of campaign or audience, but it it solves a lot of problems and it's it's a big topic in in the CDP space.
So, that's I'm very excited about that. We're actually starting that right now. So, awesome. Uh yeah, I'm pretty pretty stoked about that. It's been a it's been a problem we need to solve for a while. So, we're getting there. Awesome. Yeah. Well, thank you both so much. We got some time left for to round us out here. Um, so with that, thank you guys so much. Thanks. Appreciate it. Okay. Um, thank you for spending your, uh, second half of your lunch hour with us here.
Uh, I'm the senior vice president of platform engineering at Health Catalyst. Um, I'm going to spend a few minutes today talking to you about our mission, our flagship product, our partnership with Data Bricks, and an example or two of how we're using Delta Sharing to make life a little easier. Um, easier both internally. selfishly uh but more importantly for our clients. So, Health Catalyst is a healthcare outcomes improvement company founded in 2008 and our mission is to be the catalyst for massive measurable datainformed healthcare improvement.
Uh a little more about us, we serve hundreds of health systems and integrated delivery networks around the world, but our true focus is on the patients like me and you. Uh there are a lot of ways to make money in the healthcare sector. Some of them may be more noble than others. Uh but we found that if you really focus on the patient and enable providers to give them the right treatment at the right time, everyone wins.
So the patient gets out of the hospital sooner and is less likely to need to come back for the same issue or complicating factors and hospitals get to utilize their resources to effectively treat more of the population. So win-win. Um so how do we do that? So, we built a healthcare specific data and analytics platform called Ignite. Um, it's an ecosystem that combines best-in-class technologies, healthcare specific data models, self-service tooling, and industry expertise.
Uh, we use that to break down the cost, time, and access barriers that exist to improving healthcare. Um, and we do that by getting the right data to the right people at the right time. Um so concepts here probably 100% of you are familiar with uh healthcare has data all over the place in all kinds of formats in all kinds of systems that aren't necessarily good at uh talking to each other. So a typical health system or client of ours will have at least uh their electronic medical record, their claims and billing systems, their patient experience software, finance and cost accounting software, a variety of just other ancillary components as well.
We pull all that data into our Ignite Lakehouse uh conform it into those healthcare specific data models. Um we use a medallion architecture uh that's governed by Unity catalog and then on top of all of that we put a suite of um self-service analytics tools. Um now I want to quickly touch on some of our history just for context um because probably again some of you have have gone through something similar here. Uh so the early iterations of our platform were built on SQL server.
Uh the healthcare industry overall tends to be 10 to 15 years behind on using cutting edge tech. Uh so when we chose SQL Server back in like 2009, we actually felt relatively forward-looking. Uh but in the late 2010s, kind of before the pandemic, it was time to evolve. uh our client base and their workloads had grown dramatically and had gotten a lot more complex and we knew we needed some kind of cloudnative solution to continue to scale and to scale both like technically literally as well as just from a company growth perspective.
Um so an example there like our we used uh SQL server on Azure VMs those VMs and like SQL server licensing costs were starting to be just overwhelming. we we just pass them straight through to the client, but the VMs needed to be beefy enough to support their most intense workloads, which were usually kind of overnight pulling the data in and then integrating it into dozens or even hundreds of um different data marts and data products.
Um and then so that was great, that worked, but then all that memory and CPU pretty much was sitting idle for the rest of the day, which is obviously a waste. So, enter data bricks. Uh, as we looked to upgrade our underlying technology, we assessed several options and it finally came down to either data bricks or snowflake. Uh, probably obviously since you're listening to me talk today and not last week, uh, we chose to partner with data bricks and I want to talk about a few of our reasons for that.
So, uh, first of all, again, we were very sensitive to cost. uh healthcare margins from a hospital and system perspective are slim and investing in analytics can actually be a luxury for them. So it was really important for us to find the best overall price to performance ratio we thought we'd be able to squeeze out of a platform. Uh secondly in talking to our clients many of the most mature ones already had data bricks talent and expertise and sometimes even existing relationships and uh projects with data bricks.
Next, we were really impressed by their road map and the rate at which they were deploying new features, especially around AI. Um, and honestly, it's still hard for us to keep up. Um, I'll tell you, this is my my third data and AI summit, and for the third year in a row, I have this funny combination of feeling like really excited about the future and also feeling like I'm the dumbest person here. Um, if you caught the JP Morgan guys uh keynote chat yesterday, he said one thing I I really liked, which was just that this is going to be fun and this is going to be hard.
And I think he nailed that. Um, last point here, uh, we really resonated with the history of data bricks just coming from academia and kind of the open source community. Uh, we just felt like there was some good inherent uh, mission alignment there. Uh so we are part of the built-on data bricks partner program. Uh this has been really successful for us because of the extra access we have to technical and go to market resources within data bricks.
They're actually some of them relatively embedded within our teams and they help us keep moving in the right direction from like a feature just like forward-looking feature perspective because the last thing we want to do and we've we've stumbled over this a couple times but um we don't want to go and build something and then find out that something similar is going to be data bricks native in a couple sprints. Um so yeah, shout out to for that.
Um, so I I work with a lot of extremely smart people at Health Catalyst, but a couple years ago, none of us knew anything about data bricks. Uh, so without the built-on program and some help there, um, I just can't imagine we'd be as far along the journey as we are today. So big honest recommendation. Uh, they didn't even ask me to say this, but uh, consider talking to your account reps and, uh, see if that kind of relationship makes sense for your organization and what you're trying to build.
Um, because really it it it's been great for us. Okay, so very related to that, uh, sorry this is a little sales pitchy. I go kind of quick through this one, but basically the story we tell and part of our value prop is that anyone can go out and get data bricks, start ingesting data, start integrating it and uh and go from there. Especially, you know, since the yesterday's keynote, we know that there's a free version now.
Um, but we, you know, in our experience, combining data bricks with health catalyst enables our healthcare clients to get value faster than they could by either building it themselves or building it with some kind of in-n-out consultant type relationships and hopefully have a lower total cost of ownership. So, why is that? Um, it's because healthcare is hard and it has unique challenges. Um, and even a state-of-the-art data and AI platform doesn't solve some of the weirdest and most important challenges out of the box.
So, circling back to my overall point here, uh, I want to highlight a couple examples of those today and show you how we can use delta sharing to save time, complexity, and money. Okay, so this is a busy slide. Don't worry about trying to read it. Uh, the point is just that there are a ton of standards in healthcare. Um these lists represent a lot of the reference terminology and ontology data sets that hospitals have to use across their systems.
Uh for example, things like how a patient's gender or ethnicity is coded or how individual diagnoses or procedures are recorded and communicated across systems. Those things all need to be standardized. uh like when hospitals are submitting their claims to insurance companies or their quality measures to CMS or other entities uh they need to have some consistency there and remember it's common for a healthcare system to have dozens and sometimes even hundreds of disparate systems that they're working within just to do their business and again those systems are not necessarily incentivized to be interoperable with each other.
Uh so to alleviate this problem for our clients or to help um we created a catalog of the most common and useful healthcare terminology data sets each of our clients now has access to that catalog as soon as they have access to their Ignite tenant with basically zero manual effort thanks to Delta Sharing. Uh so that's some of them over there that are circled. Um okay so that's step one for us. Bring in the data add in reference data to enable users to understand it better.
But let's take the next step. So just like standards, healthcare also has a lot of common data models. Uh if you squint, you might be able to see a couple of them on the sort of bottom left, but things like HL7 or FIRE or OMOP, USCDI, etc. Um none of these data models are necessarily ideal for analytics or designed for it. So we took in our humble opinion um the best aspects from each of them and created a suite of data products that we call expert data collections which are designed to give analytic users a single comprehensive source of truth for all of their data.
Um so jumping into that one of the challenges we face when combining that data is around master data management and identity resolution. Okay, so on this slide here, uh, kind of top left area, you can see an example of a patient named. Uh, I know it's hard to imagine, but across the four systems that data lives in, some of her data isn't perfectly accurate or pristine. Uh, it looks like in one system they spelled her name wrong.
Um, in another one maybe her home address and email changed over time. Also, her last name changed at some point. Maybe she got married. Um, if you were an analyst and your job was to create a longitudinal view of endtoend experience as she moved through your health system, which is kind of the heart of a lot of healthcare analytics. Uh, you might be tasked with trying to figure out how to put those pieces together. Um, that is difficult.
Um, it's subjective and it's not something you could necessarily do at scale without some help. So to solve this, we can use delta sharing to feed demographic information into a centralized probabilistic matching algorithm that returns us a mastered or golden record. Uh which is that piece in the middle there. Uh so this golden record can be crosswalked back through all four of those systems and give us that longitudinal view of experience enabling her providers to again improve her carefully.
Uh now the main use case here is for patient identity resolution. Uh we also do some other domains. Um we look at providers. They're actually a lot easier because providers sort of have a built-in golden record ID called an NPI that comes from CMS. Uh but we also look at things like locations and departments um and even custom stuff when our clients have different cool use cases. So, let's see. Um, it's mentioned on the slide on the right over here, but I want to call this out and kind of throw back to something I said earlier.
Um, this functionality is provided to us by a partner called Civica. Um, I talked earlier about how part of the DNA of Ignite is combining best-in-class technologies into a cohesive package. Uh, we are not trying to reinvent the wheel, right? Um so we we don't want to be working in a space where the technology is already commoditized. Um we are very much in the business of focusing on what we think we're best at which is applying healthcare specific expertise to create those better outcomes for real people.
Okay, one of my last slides here and yes the video is working. Um this is one of our self-service analytics tools called population analyzer. It allows non-technical people to drill down into data from our expert data collections and is one example of Ignite attempting to get the right data to the right people at the right time ultimately to deliver on our mission. Uh we have a couple other areas we're excited to look into delta sharing to improve uh our own operating efficiency while also saving our clients storage costs by moving closer to that zero copy architecture.
Uh, and if you're interested in hearing about those, happy to chat. But, uh, I think that's about it for my time. Awesome. So, I think with that or so, we have Yeah, make sure you complete the survey, please. I've been asked 10 times to make sure I say that. Um, and yeah, I want to thank our customers and guests for speaking. Uh, for the discussion and, you crushed it. Amazing job. Um, so I don't know how much time we have at all for questions for two minutes.

Learn more about the Databricks Data and AI platform.