Skip to main content

Revolutionizing Banking Data, Analytics and AI: Building an Enterprise Data Hub With Databricks

Summary

  • Enterprise Data Hub strategy: This video showcases how First Horizon Bank built a unified data foundation addressing legacy system complexity from decades of organic growth and M&A.
  • Regulatory readiness: Learn how a regional bank modernizes data infrastructure to meet escalating regulatory obligations and reporting requirements as it scaled to $85B in assets.
  • GenAI-powered insights: Discover how the new data hub unlocks advanced analytics and generative AI capabilities, enabling data-driven decision-making across the enterprise.

Revolutionizing Banking Data, Analytics and AI: Building an Enterprise Data Hub With Databricks

Watch: Revolutionizing Banking Data, Analytics and AI: Building an Enterprise Data Hub With Databricks
Talk by Jamie Cosgrove, RVP - Financial Services, Databricks ; Mohan Sankararaman, Chief Information Officer, First Horizon Bank ; Shailender Sidhu, Principal, Deloitte
Explore the transformative journey of a regional bank as it modernizes its enterprise data infrastructure amidst the challenges of legacy systems and past mergers and acquisitions. The bank is creating an Enterprise Data Hub using Deloitte's industry experience and the Databricks Data and AI platform to drive growth, efficiency and Large Financial Institution readiness needs. this video will showcase how the new data hub will be a one-stop-shop for LOB and enterprise needs, while unlocking the advanced analytics and GenAI possibilities. Discover how this initiative is going to empower the ambitions of a regional bank to realize their “big bank muscle, small bank hustle.”
Here's more to explore:
The best data warehouse is a lakehouse: https://www.databricks.com/product/databricks-sql
How to migrate legacy data warehouses: https://www.databricks.com/resources/ebook/migrate-your-legacy-data-warehouse-databricks
See all the product announcements from Data + AI Summit: https://www.databricks.com/events/dataaisummit-2025-announcements

Chapters

FAQs

Why did First Horizon Bank need to modernize its data infrastructure?

As a $85 billion regional bank with 160+ years of history, First Horizon faced increasing regulatory obligations that hinged on sound data capabilities. The video explains that crossing certain asset thresholds (like $100M) triggers significantly more regulatory reporting requirements, making a modern data foundation essential.

What challenges does a bank face when built through M&A?

This video illustrates how First Horizon's growth through both organic expansion and M&A created fragmented data across legacy systems. The Enterprise Data Hub unifies this data, providing a single source of truth for operations, regulatory compliance, and business intelligence.

How does the data hub enable generative AI for banking?

By consolidating disparate data sources into a unified lakehouse, the data hub makes data available for GenAI applications. This enables new capabilities like fraud detection, customer insights, and operational optimization that weren't feasible with siloed legacy systems.

What is the role of a data marketplace in this platform?

The video mentions a data marketplace component, which allows business units to discover and access curated data products. This promotes data democratization, reduces duplication, and accelerates time-to-insight across the enterprise on the Databricks Data and AI platform.

Full transcript

0:00 Introductions and agenda

All right, good morning, good afternoon everyone. I'm the chief information officer for First Horizon Bank. We have two other partners with us. I'll let them introduce and then we'll get rolling. Yep, I I'm from Deote. Um lead lot of financial services, AI and data transformations. So happy to be on. And unfortunately I'm the other uh uh with Data Brick. So apologies if you're expecting somebody else. Uh I lead uh part of the financial services teams focused on our clients that enterprise financial services clients particularly banking and insurance that are early on in their journey with data bricks.
So kind of helping customers on board with data bricks. All right sh if you can move to the next slide. The agenda here is three parts. I'm going to set the stage, tell you more about who we are, uh then cover what are we trying to do today, where are we today, and then also talk about where are we headed. Right? So, it's a three-part agenda I'll try to uh get through. If we can move to the next slide, please. Uh one more.
First Horizon Bank is a $85 billion regional bank. We're headquartered in Memphis, Tennessee. Our roots go back to 1864. We have been around for more than 160 years. uh our legacy has been on uh being resilient through all this time, being innovative and also being committed to the clients and communities we serve and that's how we grew. Uh the growth obviously over the last 160 plus years has been organic has been through M&A and you can see we started as a very small bank in a very small town uh in Memphis in the 1800s and today we're about 85 billion in asset size.

1:06 First Horizon Bank vision

If you look at the stack ranking of banks in asset size, I think we're somewhere in the lower 30s based on where the numbers show up. So that's a little bit of context for you on where we are. We can move to the next slide. So for people that know banking, uh there are some asset numbers that are very material. 100 million mark is a special number. When a bank crosses that in asset size, our regulatory obligations increase significantly.
Right? There's a lot of reporting we have to do. There are a lot of things like that and you look under the hood for you know what do I need to report on all these data shows up everywhere right so data is a big deal and this is just one reason why we're doubling down on data there are many more that I'm going to talk through but regulatory obligations all hinge on how sound our data capability is right and we anchor everything on five uh principles you know when we do our data journey there are five things I zero in on first one is scale and speed.
Second one is quality. Third one is innovation, right? And of course, you have to factor in AI and other new capabilities that are available. Fourth one is intelligent automation, right? And the last one is datadriven insights. Data matters. Data uh is everything. So that's why we're doubling down on data. If you can move to the next slide, please. So it's not easy, right? The journey is very hard. There are a lot of things to solve for siloed data, fragmented data and I can go on and on and on but what's more important is if you look forward we as a company we as a bank we want to materially change uh the customer journey the interaction with our customers the experience with our customers we want to change that we want to become very efficient we want to automate as much as we can right that's where the intelligent automation automation at scale comes in right these are some examples of where data plays a big role If you look at the common element with all these examples I gave you, it's all about data.
So getting the data foundation right was a big deal was very important. So a year ago the first thing I had to solve for is what platform should I do all this on? That's where and data bricks come in for us. We partnered with them. That's the platform we picked. All the work you're going to hear uh is being done on data bricks. The next one is who can I lean on partner with to build all this and uh get to the end state right that's where deote came in for us right we partnered with deoid so it's a three party solution here the bank data bricks and deoid coming together to deliver on this agenda that I'm going to walk you through with that I'm going to invite if you want to chime in here.
Yeah, I I would just say that I think I think the the journey that on is very representative of what we see in our space. So I my teams handle uh banks with assets from sort of 10 billion on up to the largest banks in the world. Um and I you see a common thread which is sort of this data foundation I think that gets created and banks aren't waiting any longer like the pace of innovation is moving so fast that as we think about it we're not just building a data toolkit anymore for the data teams like what we're thinking about is with partners like deote you know what what can what business value can we deliver to the organization in short term so that we can continue to innovate and I think you'll you'll I'll talk about that in a couple slides I guess is really important and I I think as you look at the slide the issues that banks and other organizations in FSI uh face they're across the data value chain whether it's around architecture around governance around how do you manage data so the question was how do you realize the vision of the future and then there were few guiding principles that drove the decision of building uh the enterprise data hub for uh for first Horizon bank u the first was how do you establish that authorized provisioning point or authorized data store for data.

5:33 Enterprise data hub strategy

Uh it has to be a govern platform which people can trust in terms of this is the ecosystem I go for my data needs. The second principle is governance. This has to be driven by very strong data management and data governance principles. Um the choice uh taken is uh towards a data product and how do you ensure that those producer and consumer data products are managed and governed the right way. People are clear about the accountability structures.
U they ensure that uh the data completeness and quality is assured in this platform. Um third is now uh how do we accelerate this journey and for that use the nextg technologies. uh we'll do a demo after u the presentation of how we are realizing the true uh platform um delivery and build and uh using genai and advanced uh uh analytics capabilities across the value chain whether it's about data modeling whether it is around data uh quality or data governance building those business glosseries uh using those advanced analytics and advanced and next technologies to drive this innovation further and finally um from adoption perspective how do you remove any barriers uh for uh the customers internal and external to access the data so creating those data marketplace and uh I think uh in the conference it was announced data bricks marketplace so we're working very closely in creating that marketplace where the adoption and access to data will come in for the different consumers in the bank um but yeah those are the kind of four uh guiding principles four pillars which established how this vision of the future that Moan just shared and the bank is establishing um will be realized.
Okay. Uh I I feel guilty taking credit for the work that Moan and his team have done kind of talking about this architecture. But I but but at the core of what you see here and what what uh the team at First Horizon has built is so if you look at um you know a number of systems of record sort of on the left are being brought into data bricks. So the entire um EDH is built on the lakehouse platform with data bricks. Um so you see these data points on the left.
Um so data bricks uses the medallion architecture for ingestion of data, curation of data, creating sort of a single source of truth. uh the role that uni catalog plays in that middle section is to provide the metadata metadata layer for things like security governance which is really critical when you think about hitting the hundred billion dollar asset and the amount of regulatory uh influx that creates for the organization.
So Unity catalog plays a really critical role in that and then you sort of uh and you're building sort of data products there on the top of that stack, right? So it's you're using data bricks to provide uh that those those data products and then uh to the right you see um the deote data as a service playing a a fundamental role which is Genai embedded there into the tools for data mapping and things like that and we'll talk about it a little bit when you see the demo.
Um, I think we'll pass it back to you to kind of talk about, you know, kind of what what the organization built. So, clearly you can see uh building a one-stop shop is our goal. That's where we're headed. Enterprise data hub is the term we're using. We're bringing on sources. Uh, I want a one-stop shop for my bankers, from for my analysts, for anyone in the bank. Today, in the absence of this, people have to solve this themselves.
That's a waste of time. Everybody has to go source the data. Everybody has to go curate it, apply all the quality gaps, fix them. And I want that to go away completely. A one-stop shop where data is ready for you and you can start focusing on what you do very well, which is the analytics or whatever you do for the bank, right? So that's where we're headed and EDH is going to help us. We started this journey uh about 8 months ago.
We're all making very good progress and it's the medallion architecture uh that we're implementing. So our goal is our end state is to have a data marketplace for all my internal consumers so that they can uh get access to the data very easily. Next slide please. Right. So you can see where we are. uh I've made the choices platform partners we've started sourcing uh and we're trying to run this like a data factory right there's 25 sources in phase one we want to bring in so that's running very smooth in an very iterative fashion so we'll continue to uh finish that journey the next thing is uh several of us touched on this concept data marketplace you probably heard this in several breakout sessions and whatnot that's where uh our end goal is right source all the data, curate it, get everything ready, get it in one place, apply all the access, control, security, whatever you want, and let the bankers and other analysts inside the bank focus on going to a single marketplace, getting access to the data they need.

10:08 Progress and modernization

So, we've been busy getting the bronze and the uh silver layer stood up to use the medallion architecture terms and now we're actually doubling down on getting our golden layer stood up so that uh we can start showing value to our bankers and to the bank. All right, anything you want to add? I think uh the next five minutes we'll uh give you a sneak peek of uh what this uh Genai assets and uh solution is and then we'll open up for Q&A.
Sorry. So to introduce myself. Um again I work with and data bricks one of the leaders in deote. So surprise package cameo we decided to show make it real for you guys rather than talking through slides because making it real you know I think you can correlate better you can imagine how you can apply it you know in in your life. So can I have a show of hands? How many of you are working on genai in in any shape or form?
Okay. How many of you working on agentic AI in any shape or form? Yeah. So that's where you know this is heading. So when we start this journey with first horizon with moan and team a year back you know it we clearly made a mandate that we're not going to do this you know in a traditional manual way. The only way to speed to market is how do we use genai end to end you know starting from producer of data to consumer of data across the value chain you know uh let me just show it to you that it becomes easy for you to visualize so what you're seeing on the screen here when we started this journey a year back starting from left of data preparation analysis going all the way to the data marketplace where your consumer is going to come and you know take the data we sat together and said let's think of use of genai end to end how can we accelerate how can we start using Right.

12:23 Generative AI demo

So, unfortunately, we don't have time to click on every single tile that you see there. Um, I'll just show you a sneak peek of couple of such cool agents that what we're using on the ground to accelerate this whole journey and then you know if time permits or if there's other opportunities where we can do a deep dive right so what you see here is a identified data value chain end to end to build this modern data platform on data bricks.
Now just to go and sneak peek I'll take one example or a couple of examples. What you see is the what we call as an intelligent business glossery under the hood what it is using is it uses genai it reads through all the context documents of fast horizon. So think of like you know 2 years back or four years back before geni you know came came to this world if you for a for a highly regulated bank if you need to generate glossery across thousands of attributes you know some of you might have lived through that pain how difficult or painful it is to go write glosseries generate data quality rules write data quality rules for it and then make it all curated so what we have done here is like you know use you know genai for example I'll just use delta lake you know as my platform here I'm just choosing the silver layer or the confirmed layer, you know, whichever term that we want to use.
I just taken a one example here. Let's just say the loan loans table because, you know, we're talking about banking. And what you see here on the right side. Give me a second. Let me even go directly here and show you how to how does it look for you. Give me one second. Yeah, you see all those options here. generate business glossy, view business glossy, discover critical data elements, edit business, you know, critical data elements.
All of them is driven by geni at the back end. So you as an organization now you're not spending like weeks and hours trying to generate glossy. There's an geni agent that is running behind it is contextualizing all your documents and generating glossy on the fly. Right? That's just one example. If I go here is another example. Let's just say this, you know, the smart map that you see here. one of you know there's a lot of pain in the organization when you're trying to map your system of record to your target data model trying to write social do mappings and all of that stuff what we have done with data bricks is like you know use use genai under the hood can I look at the schematics of the documents can I look at the schematics of my glossy that I've generated can I automate this mapping to some extent as as much as possible that's that's other thing that we have done and the if you I can keep going on on this one if I just take one more example here the data quality engine that you see here You see this rule authoring and rule execution here.
If I go to this rule authoring for a second, what this one does is very similar concept rather than you trying to understand the banking domain, banking specifics, whether it's deposits data, loans data and etc. trying to understand what kind of quality control that you need to put in place whether it is risk and regulatory purposes, whether it's analytical purposes. You can rather than choosing all of that stuff, you can then provide all the context documents.
The engine is going to come up with data quality rules the stewards can review and then generate code for it automatically because in the era of genai I don't think we are we no longer going to sit and write code manually end to end right that error is gone. you want to add anything more? Yeah, I think the change that is happening is you don't spend two years doing data transformations anymore and then after 2 years realize well maybe something was not done right.
I think the what we're doing with and First Horizon Bank and partnering with and data bricks is how do you use these nextg technologies and what we just kind of give a brief sneak peek into uh use that to uh remove the pain that was traditionally in data analytics modernization programs. So that's kind of the nutshell what we're driving towards and um and we very feel very confident and uh thankful to the partnership we have with the bank and with data bricks in this journey.

16:37 Data marketplace overview

Yeah, I'll just close with one thing. I I know we're not doing justice to all the stuff that we have done just in the interest of time but what we're also now working with first and team is Moan just mentioned that you know we just finished building the silver layer and the attention is now to create this data marketplace build this like you know what we call as a data products I mean in in data bricks marketplace and how many of you have started experimenting with data bricks genie yeah so there's already so that's exactly what we're doing so what we're doing is like bring up this data marketplace where you and come and consume your data.
On top of it, we're building, you know, what we call as a talk to my data concept using Genie on top of marketplace. Now, rather than like building like hundreds and hundreds of dashboards where, you know, the business team has to come through it, put in a request, create a dashboard. So, what we're doing, we trying to take away all of that stuff. How do we you know give this power back to business? Create the centralized data marketplace, put Genie on top of it as talk to my data and they can come and pretty much have their own agent running which can create dashboards for them, create visualizations for them and that way you know it doesn't become the bottleneck and you have given the power of data back to business.
They don't have to know SQL. They don't have to work with like you know IT engineers to build dashboards for them. You know it's it's in their hands. They have their own personal agent. Right? That's the vision we are going next. Come on. All right, that is our story, guys. That's the journey we're on. Um, hopefully that resonates with most of you in the room here. It's not easy, but I think we're making very good progress.

18:03 Future goals and recap

So, uh, just to recap, we we have the silver layer, we have the bronze layers, uh, being stood up, right? Like I said earlier, we're now doubling down on getting our golden layers stood up. You saw what he showed and uh I think that's where the value for us is going to uh come out right as we make more data sets available uh purpose-built data sets we have our internal analysts bankers whoever if they can come uh consume this data so that they can put this to good use that's where we're going to materially alter.
So think about some of the domains where I'm focused on is the customer experience the customer journey that's one fraud u bank operations automation that would be another area uh of course technology productivity you know that uh is bu baked in uh the next area that we like to really make a material difference is the use of agentic agents u and really trying to transform wherever we have manual processes in place you know if I can change that'll be a big boost.
So that's the journey we're on. Uh I think we're making very good progress and u I think we have time for questions. So if there are any questions we'll take questions now. All right. There are no more questions. Thank you everyone. Thank you so much. Thank you.

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.