Techcombank's Multi-Million Dollar Transformation Leveraging Cloud and Databricks

Summary

• Digital Transformation: This video showcases Techcombank's multi-million dollar transformation journey using cloud and Databricks to modernize data infrastructure.
• Customer Success Story: Learn how a major enterprise bank leveraged the Databricks Data and AI platform to achieve scale, agility, and competitive advantage.
• Real-World Impact: Discover the measurable business outcomes and lessons learned from a large-scale digital transformation project.

Techcombank's Multi-Million Dollar Transformation Leveraging Cloud and Databricks

Talk by Santhosh Mahendiran, Chief Data and Analytics Officer, Techcombank (TCB)
The migration to the Databricks Data and AI platform has enabled Techcombank to more efficiently unify data from over 50 systems, improve governance, streamline daily operational analytics pipelines and use advanced analytics tools and AI to create more meaningful and personalized experiences for customers. With Databricks, Techcombank has also introduced key solutions that are reshaping its digital banking services: AI-driven lead management system: Techcombank's internally developed AI program called 'Lead Allocation Curated Engine' (LACE) optimizes lead management and provides relationship managers with enriched insights for smarter lead allocation to drive business growth. AI-powered program for digital banking inclusion of small businesses: An AI-powered GeoSense assists frontline workers with analytics-driven insights about which small businesses and merchants to engage in the bank's digital ecosystem. And more examples, which will be presented.
Here’s more to explore:
Economist Impact: Unlocking enterprise AI: https://www.databricks.com/resources/webinar/unlocking-enterprise-ai
The State of Data + AI: https://www.databricks.com/resources/ebook/state-of-data-ai
See all the product announcements from Data + AI Summit: https://www.databricks.com/events/dataaisummit-2025-announcements

FAQs

What was Techcombank's transformation goal?

This video showcases how Techcombank embarked on a multi-million dollar digital transformation to modernize its data infrastructure using cloud and Databricks.

What benefits did Techcombank realize?

According to this video, Techcombank achieved improved scale, agility, and competitive advantage by leveraging the Databricks Data and AI platform.

How long was the transformation?

This video provides insights into Techcombank's transformation journey and the timelines involved in implementing a large-scale digital initiative.

What lessons can others learn from this story?

This video shares real-world challenges and best practices from Techcombank's experience implementing Databricks for enterprise banking operations.

Full transcript

Okay, let's start I think. Yeah, thank you. So, first of all, thank you for the limited people who have come. I think you're first of all, you deserve congratulations and you're all pretty brave, I should say, because this is their end when we start at 7:30 in the morning and they put me in the last room at the last end of the day. So hopefully hopefully you guys get something useful out of this session. Um. I'm the chief data analytics officer at Techcom Bank.
So for those who have heard of Techcom Bank great but otherwise let me give you a small introduction before we go to u I mean this is the agenda we'll talk about it but um the primary focus of my presentation if if there is something you can take away at the end of the day two things I would say the first is how data bricks helped us to build a data platform. Okay, so we are in a datab bricks conference so I better say the good things first but I'll also talk about some of actually why we did it.
We started off on AWS native and then moved to data bricks as an accelerator and it has proven to be beneficial to us. Second and probably the most important is how did we create value because many in this is a technical conference everyone talks about sparkl which is rightfully should be spoken but most importantly nobody talks about is the platform generating value is it actually useful and that's what I'll talk about and I'll talk about some examples so let's quickly dive straight in right I mean techcom bank for those who have not heard is it's Vietnam is Vietnam's largest bank private bank it we I mean we make we made a profit of $1.
1 billion US dollars, not dong. So that was last year. So we got 300 plus branches, almost 300 branches, 12,000 employees, and we have a 15. 5% ROE, which is pretty strong for any any listed bank even by American or world standard. So a top bank in Vietnam and our aspiration is not just to be the largest private bank, but the largest bank in Vietnam and put Vietnam on the ASEAN map. That's our objective and that's one of our stated objectives we are working towards.
So again some some introduction about the bank from an awards perspective we've been rated the best bank by every conceivable press whether it's world uh I mean global finance euro Asia name the bank eurom money finance Asia I think every every award under the sun we have won and it's pretty pretty uh and I I feel pretty proud standing here putting all this in front of you but yeah it's the hard work of all the 12,000 employees but yeah we call it the grand slam bank.
In fact, I am proud to share that in 2025, we have again won it again all the three awards. Only bank not only in Vietnam but in Asia to win all the three awards again. So we are the top bank by every measure and some of these awards are not bought because most awards can be bought but these are not bought. These are evaluated with strict criteria. So that's that's that's who we are at tech compact. Now coming to the crux of the presentation the question is how did we transform or how are we transforming right and then as I told you how did data bricks help us or did they even help us and then what are some examples I think I'm happy I will share some examples like I said you will go away with those examples you'll also understand how we created value and I think value creation is probably the most important aspect of my presentation because in this uh technical presentation While while we can talk technology like I said is there money to be made after all this implementation but and I think how did we also measure impact that's also something that I'll share but fundamentally I think what we need to start is with the why the how and the what is the easy to answer but the why is the most important and why did we start that transformation journey if you look at if you look at the why you should first understand our bank structure we are pretty good at our affluent segment affluent are a pretty rich guys out there and we are pretty good at the large cap.
Large cap are the biggest conglomerates in Vietnam. So as a bank we are pretty good in financing these but if you want to go down the value chain whether you want to do it to the mass or the mass affluent or the smallermemes or the smaller businesses you just can't do it by hiring more people. Not that you cannot do it but you did it. You need to replicate what you did at the at the top level which is to get some really really smart people and that cannot be scaled by hiring more and more and for that if you look at the left we said powered by DDT which is our formula which is data digital and talent.
So we we early on our strate strategy was very simple. If you have to really win we need to we need to unlock these three. And if you look at if you look at the and I'll go back here. I think the key is to understand that massme mass affluent and your upper theme all this can be enabled using this power because if you take a big conglomerate or a rich customer you need to have a RM or one RM to service four or five customers.
But if they're mass customers, if you have a good product which is pretty easy to use, then they would use it well without the need for too much. So technology can help you. That is the key, right? It's just not about using data, digital and tech, but it's of understanding what people will use. So our three transformation pillars which we which we proudly call I know this everybody's gladly copying which which is also okay is it's no secret sauce, right?
But we said we will use data, digital and of course talent are three pillars to differentiate ourselves. But the key is that circle in the middle. I think it's very easy to say we can get data, we'll we'll do digital, we'll we'll transform our business etc. But creating that value proposition, creating that customer differentiation in the middle is the most important key. And that for me is the secret sauce. And you will see some of what we have created as secret source because any company that needs to succeed can buy the best.
I mean data bricks is a great product. You can buy data bricks. You can buy AWS or GCP. I mean you can also buy talent to an extent. You can't buy too much talent but you can buy talent. But you cannot if you don't get that to formula to work you will not get CVP you will get customer dissatisfaction. So yeah so we kind of that's that's the key thing in the middle is the secret sauce and every company that's the secret sauce.
So let's start off with how our data foundation right. So the first thing from our data foundation we clearly said we want to have a journey where we have our data epics or capabilities mapped out. I just listed down some of the capabilities. There were 36 capabilities in our data platform journey that we mapped out right from building injection patterns to having feature stores to having user workbench to having a machine learning workbench to having lineage.
I mean name it the usual stuff. There's no rocket science here. Anyone who's done a data lake or or any kind of data implementation will understand this is key. But like I told you I when we when I started this journey in 2021 the story that I want to share with you is we said we'll go I mean we had to choose between AWS, GCP and Azure. You'll not go wrong with any one of these. I'm not promoting any of these but we went with AWS primarily because they had more capabilities and features built.
But as we started doing development one of the biggest bottlenecks that we faced was everything required hand coding glue jobs. You all are pretty familiar AWS request lots of glue jobs and that's when when we started our conversation with data bricks we understood that the whole injection pattern the way the orchestrator works the way that you actually can set up your pipelines the way you can do your patterns the way you can set up your lineage your metadata unity catalog that was pretty huge it was just not an accelerator it's like either you want a PC or you want a so data bricks was like a to us right so the advantage that data bricks offered was pretty immense.
So we were able to scale up and in fact we were one of the few banks to launch a successful data platform in record time. So Techcom Bank is um I mean I mean I'm I'm extremely I'm I'm a great fan of how we are able to do the acceleration and data bricks truly helped us actually build that acceleration. Now if you look at our platform again in the morning I think uh uh during the keynote also they spoke about the medallion architecture.
We also followed the same medallion no rocket science here typical bronze, silver, gold where we had different stages for our data as you can see and then we clearly said how do we actually operationalize and then how do we give access to different users across different schemas and then how do you actually upload new data? How do you make sure that it all goes through the same pipeline and then how are the patterns established.
So we we kind of build that and build that into the process. So if you look at our key I think uh domain specific workspaces was very important because one of the biggest challenge was we had lot of users waiting to use the platform and we could not wait till the platform was built before you could onboard. So we had to do stuff in parallel and that's where again data bricks melene architecture helped because if you had said everything had to be golden data and everything had to be fully structured and everything linked up we would never see the end of end of the day.
So I think light of the day. So I think the biggest advantage we had was we could the melanin architecture helped us to democratize much faster of course with the right guard rails but we are able to really uh start creating value. Now this is a key slide right and I think why because this talks about how we actually layered our data platform I I call it the four tier lasagna where we have data foundations which is at the bottom zulio.
So Zulia in Vietnamese for those who know Vietnamese Zulio means data Zulia right so so we changed that to the English it's a play on the word Zulio so while we said it is the king of the jungle our data is king of the jungle but Zulio also means data so that was why we gave the name our data platform is called Zulio so it's a play on the Vietnamese name Vidya is our knowledge repository where our unity catalog and Calibra come together and vidya stands for knowledge for those who know Sanskrit and uh so it's so so I worked in Vietnam so I used a combination of Vietnamese and the Indian Indian nomature to bring so vidya is where our unity now actually after matai's presentation this morning I don't know if even we'll need kalibra in the future because unity catalog can kind of expand beyond just so so probably but right now we use unity and kibbra that's our knowledge repository the third is Our product layer we call it taps which tech bank app store very simple and there we create all our data products and the fourth is our intelligence layer.
I I think this morning also in the keynote everyone was talking about hey we will abstract the intelligence from the systems and then start building this intelligence layer. We've already started doing that and that's where we started actually building the engagement and the intelligence even we have started hollowing out our core systems even our core banking our uh cards and others are getting hollowed out and all the intelligence is built in the AI layer.
So we are not waiting to be an AI bank in the future we are already transforming to be an AI bank in the current circumstances. So if you look at all this, this is where our road map and how and we have layered our data architecture, right? We are working the full value chain right from the platform to the knowledge repository to creating data and AI products to actually building the intelligence and also making the intelligence available across the entire ecosystem.
So this is the this is how we are actually uh building the full uh data value chain. Now one of the things that I told you right it's enabled by TDDC and the unity catalog is our knowledge repository if you look at it it's pretty comprehensive so we are probably I don't know how many I mean you're all working for different organizations we have our data glossery right from source systems to BI reports today Matai was saying even unstructured even yesterday Matai was saying and was saying right and how they created this unstructured reference information we have actually right now done it a bit manually but if unity and uh is going to provide that out of the box that's going to be a big feature for us so this is right now so those new features are coming in the current data bricks release which we are going to take advantage of but the key message I wanted to share with all of you is right from the moment we source a data to the data being put into a report we have entire lineage so I can tell you if I source a data and if one of my team members is making a change.
I have a dashboard that traes the lineability. It's not just a dashboard that tells you also this is going to be the impact to this level of uh model or this report of dashboards or this report or this BI change that is very powerful. I I'm not sure how many organizations have that. I mean I worked in city bank and standard charted earlier. We tried doing that. we could get to halfway or we could do it in between some processes but doing it right from sourcing till consumption was extremely challenging.
So we are again something proud to say that so we have again lots of active users we have 87% of our data is cataloged. So meaning to say out of we source close to 2,000 to 3,000 files 87% of elements of that is cataloged. We have the definition for the rest we are working on it. We have um who actually owns that who owns definition how they maintain it and it was all and again it's all entirely put in as you can see the number of users etc is just going up now our focus is trying to get ecosystem data it's just not techcom bank's data and when I mean ecosystem the beauty is we are connected to our retail friends our insurance friends and our telco friends so the bank is part of this greater ecosystem so now we are started collecting that data and started curating it and as you know for those who are all in the AI world labeling data getting clean data is the biggest challenge and we are fixing it already.
So this is again extremely powerful and this is something that we are pretty proud of. As I told you all this why are we doing we are not just doing because we need to build an AI bank and many people say building the AI bank of the future for me AI bank is present we are already using if if we don't have this today we would be at a we we would be at a disadvantage and we need to make that advantage count primarily because as people spoke about AI agents virtual assistants services we we already have virtual assistants we have services data APIs we are experimenting on agents We don't have too many agents in production but we do have agents and all this is coming to build our whole bank as a AIdriven bank.
So the whole notion of whether a bank and if a bank can be an AI driven bank we are a legacy bank. We do have our whole system on T24 teos but still we are like I said we are hollowing that out. We are able to build an AI bank. So I'm sure each one of you can but again this is just like I told you a structure of how we have done it. um you with with all the de all the developments I'm sure you will also start as you can see all the some of these need to be vectorized I'm sure you all would be thinking about it but I this is again a blueprint for how we are building our bank as an AI bank now I spoke a lot about our digital um our data platform I want to talk a little bit about our digital footprint right our mobile is probably one of the best if not the best I we consistently ly receive as you know we have won not only just awards but feedback from clients.
Uh one one of the key things at an ASEAN level or even world level is we do 60 interactions with our customers on a monthly basis. For those who do digital knowing that even if you do 60 interactions it's I mean that's pretty big number. So almost 97% of our of our transactions happen through our digital channel. So our retail customers anything and everything happens through mobile. So our mobile is an extremely extremely important channel for us to actually um kind of engage not only with the customers but also to build the next level of u u the next level of communication and also to ensure that the customer can actually benefit from all the offers that we push through it.
And we'll talk about some of these next best action and offers shortly. But again from a mobile perspective probably one of one of one one one of the great success stories in our bank. Now um so what you have till now seen is the foundations of the bank. I told you how we built our data platform. I told you how we built a medallion architecture using data bricks. I gave you a glimmer of uh sliver of uh data on how we have built our mobile.
Now we'll get into what we have actually what what what's the intelligence right? how we have uh as you saw the four level layers the this is the intelligence layer where we started building some of the products so the first one that I want to talk about is a product called lace and then I'll talk about what tenno is but lace stands for lead allocation and curation engine right for those who work in banks you would all understand you assigning leads to frontline or assigning leads to people is one of the biggest jobs of of the banks right I mean there's a huge army of people doing it what we did was and typically you would say hey why can't CRM do it the and for people who worked in CRM you'll know that there's a lead allocate of course CRM can do but what CRM does not have is all the data what CRM needs is a set of data for allocating the leads and it can only look at one set of information it doesn't have all the information and that's why we what we said is our CRM will be just a front end and today even was saying that right by 2030 there won't be any specialized CRM tools and the reason being is the intelligence will be in the intelligence layer.
As I told you, if you go back to my slide uh here, the intelligence is going to be in the topmost layer and that's the brain because once you have all this hollowed out and once you start building it and once I know how my models work, I don't need a CRM system, right? I mean, why would you why would you need apart from the excellent workflows that they provide? I know some of you are from Salesforce and this thing, but you saw this morning also.
But the reality is you don't need you don't need a CRM system. Similar with digital marketing all these specialized systems will disappear. That's the reality unfortunately and that's what we already have seen. I mean said 2030 for me even our current CRM for those who know how we have implemented at techcom is actually a good front end. The intelligence is all built by us. The intelligence is built by our models. My team's models actually powers the intelligence.
So that's what happens in lace. We have O2O journeys which is O2O stands for offline to online or online to offline depending on how you look at where the leads have been passed off. And here is our equivalent to what I would say I mean for those who know Salesforce Einstein we built our own intelligent assistant which basically is a is is our first agent for the bank which basically tracks and gives explanation. So lace is our one of one of our intelligent products but basically which basically does the allocation and curation the numbers are here you can see I'm not going to talk through the metrics but the key that lace does is it looks at our current data it looks at external data it looks at behavior data it looks at how the RM has responded to the customer it looks at telco score of that particular customer It looks at um social data and then says this lead should be assigned to this RM for this particular purpose.
The level of intelligence that goes in is extremely extremely mind-boggling. I can tell you that much. So what happens now for this is in Vietnamese. So those who are Vietnamese here can obviously read it. But I mean the whole idea is if you look at the rightmost right the lace allocation we actually do it for the right RM we actually allocate the lead. This is our actual production screen and then we have something called as AI score right we generate this AI score based on our models and leads and then we say we give an explanation.
This explanation takes into consideration why this particular lead was assigned to this particular RM and that is entirely driven by R models. So we turn all these features that we collect we collect close to 10,000 features actually now 11,000 features. Yeah, in my slide you'll later see it's 9,000 plus but that was the old thing. Right now as of yesterday we have 11,000 plus customer features. So I can describe a customer in 11,000 different ways.
So we take those we build those turn it to an AI score and then as I told you depending on how the RM interacted even understanding the RM's behavior in understanding the RM's potential we have we in fact it's just not about customer how well the RM spoke to the customer last time we collect feedback and then even based on that he or she might get a new lead or the different customer every time so that's the level of intelligence that we have built this is live in production we are seeing a great uptick but also I want to share with you we have a cultural challenge also because till now RMs were very used to there was a person who was assigning leads and they're very happy to go to talk to just set of customers now they're suddenly seeing oh the computer is telling me to talk to this customer the computer is telling me this is the reason now they're saying hey how why is it that I should now follow the computer but the reality is till now that's that's the point I'm trying to make till now they were using their gut now the intelligence is saying you talk this way they're seeing the conversions go up so it is a cultural change.
It's the data clearly shows the conversions as I told you as they start using this our conversions are just going up tremendously but again our resistance is primarily coming from people saying that hey I've been doing this for many years now you're suddenly coming and saying there's a external digital brain that tells me how to talk to a person and the beauty about it is we also give the script what needs to be spoken so as a result one of the biggest challenges we earlier had was we had to get RMS we had to train RMs Now with uh within a few days we can get someone we can have this whole intelligence supplied by our brain and that person can talk to the customer seamlessly.
It's it's such an it's such an experience. Of course it truth be told we are not there fully seamlessly but that's where we are getting to. As you can see here as I told you is the agent again Vietnamese but essentially it summarizes on the left is in Vietnamese on the right it's in English but it summarizes entire uh customers insights on one page right this gives you every explanation of what the customer needs RM needs and also explains why this particular lead was assigned any insight additional insight about the customer which the RM did not know would also be here for example I'll give you a simple example customer shopped at say Vinmart and bought a bottle of champagne yesterday for $100 might sound trivial to you but today the RM it actually then say hey sir you had a party yesterday and then we would link it up to know that there was a party for a celebration and that helps in the conversation so every piece of information we model we get every information and then we push it real time to our RMS so here know is where we get our customer insights and we get those actually um it helps in a planning and preparation state now this is uh a video I want to show I'm going to play a video this video tells so we saw how the leads are assigned, how our hoi agent works.
This video tells how we actually hyperpersonalize, how we actually make the offer really relevant to each customer. So, let me just play the video and then you will understand the power of hyperpersonalization. In today's dynamic banking environment, it is crucial to provide customers with personalized and timely services. And our hyperpersonalization project driven by customer brain and customer centric decision engine CCDE is a key step in our AI lead digital transformation.
It focuses on tailoring unique engagement plans for each of our 15 million customers to align with their individual preferences and needs with over 9,000 internal data features like demographics, transactions, geolocations combined with external insights like ecosystem penetration, family ties, and community scores. Our comprehensive customer profile leverages powerful AI models such as next best offer, next best action, segmentation, and churn to gain a deeper understanding of our 15 million customers.
Furthermore, to enhance engagement and avoid overwhelming customers with impersonal product focused communication, the decision engine has been developed to enable an event triggered customer engagement system across omni channel martekch CRM VAC. So what have we achieved through this approach? Maximized TOI while balancing product targets by shifting from product campaigns to a customer ccentric approach. integrating existing AI models, optimized engagement preferences with personalized content and offers based on individual customer interests and needs.
Consistent communication across all touch points using a dynamic contact strategy. Continuous improvement by leveraging feedback such as click-through rates and transaction data to enhance quality and refined strategies. An example with customer men, a 20-year-old banking employee in Hanoi who regularly uses his credit card and enjoys a morning coffee routine. With the help of NBA/NBO and AI model, CCDE recognizes significant transactions on M's card and sends him a personalized installment offer via push notification at 700 p. m.
This ensures that he receives relevant and valuable financial solutions tailored to his needs and preferences. Additionally, CCDE also sends a customized inapp message, offering a 20,000 grab food discount for Starbucks purchases when using Techcom Bank card at 7 a. m. in the morning, perfectly timed to enhance his experience. Before CCDE, campaigns are generic, targeting an entire group of 100,000 customers at a fixed times across various channels.
After CCDE, he introduces offers based on customers lifestyle and habits, delivering through personalized channels and at a personalized optimal time to boost engagement. For men, we offer Grab Food discount for Starbucks purchases via message and app. However, for other people like Hi, she may also get Grab discount but grab traveling discount to go to their preferred restaurant and the message will be delivered via push notification at 6:00 p. m.
Friday night when Hyde engages the most. Overall, CCDE is a comprehensive solution that harnesses advanced analytics and AI techniques to provide hyperpersonalization of customer insights through personalized channels and personalized time. If you the key here is if you look right, it's just not about the offer. It's about the time, the channel and the preference. See the biggest problem everyone has is we only have grabb voucher.
One of the practical challenges I'll tell you and I'm sure you all face in organizations when we go to market people say oh we have only grabb vouchers to give but in grab we have food vouchers we have buy vouchers and within that also we have people who prefer grab um vouchers at a particular time and people who prefer at a different time understanding those nuances and making sure that you deliver is still personalization and I think that's that's the key that we have been able to achieve so I have two more slides and then I'll stop but probably my most important two slides which will be my takeaway for you guys for having sat through a presentation at so late.
So what I'm going to share with you is how did we create value? As you can see right I spoke what I told you in the beginning of my presentation was we created not just a set of tools but we actually created value. How did we create value? How did we measure value? These are the key things. First I I'm I'm the chief data analytics officer. I have a P&L target. So every year I have to make money for the bank. Last year I made 914 billion which is close to 40 million US dollars and this is incremental revenue.
So you can imagine we target you saw how not only new leads I have to give you have to measure it very very important. So creating this is our first example. As you can see products like lace and henoi we have been able to give intelligent leads and based on those leads we are able to increase our conversion and that's yet another way of how we are able to drive our uh or measure value and the impact that we are creating in our credit underwriting we have more than 20% genie uplift for those who know credit underwriting even a 1% genie uplift is great we have a 20% genie uplift which means we are able to actually predict the probability of default and we are able to underwrite more customers customers.
We are able to offer pre-qualified leads. Basically, we are able to say these customers are qualified for a loan even before the customer knows about their spread worthiness. Our campaigns, you have already seen that there's a 100 more than 200%. I this is something that I'm proud of. When I joined the bank five years back, our chairman used the word this is a bank with excel excellence. Everywhere there was excel. Today I'm proud to say even our management meetings the XCO and Alco which is the top decision-m point of the body where the 16 members meet everything is run online everything is done online using live data so it's literally it's literally uh we we have done away with Excel right and it's literally using all the intelligence we built this is something again very very important that you should know we launched an entire new product using data and that's shop cash shop credit in Vietnam as many of you know many of them don't have access to formal credit.
So based on just the transaction data and as I told you we have the ecosystem data which is the tel code their behavior the shopping patterns we told customers hey you you are eligible for this loan for people who run their businesses out of their homes. So we called this shop cash shop credit. We just launched it three months back. In 3 months 50 billion is $2 million US. Not big but in three months we have been able to underwrite customers who never had loans before based on just data that we have and that and using external data that's something very big and these customers don't have a credit history at all.
You could all said what's the big deal but they didn't have credit history. We have all obviously 20x it's actually 25x. We have grown the number of campaigns as I told you it's 11,000 customer features. We are able to underwrite customers more and more importantly uh at least 85% of our staff have undergone training in data and AI some form or the other. We are transforming the workforce. We don't want to leave anyone behind.
Even our front line now when we hire we are hiring AI personnel. So sharing a true story. So I met one of our brand staff and she knew that I was the AI officer for the bank. So when she came she said sir I passed the AA test but I had to take it three times I had to retake it now I've joined an AI course and this is a teller in our branch in Hanoi. So you can imagine this young lady has so interested. So she says next I want to join your division.
That is the interest that people have. So I think transformation is truly happening and truly changing. So this is my last slide but I want to leave you with some um uh learnings to share and then I'm happy to take some questions as look as much as I talk about success stories it has not been a bed of roses we've gone through our own challenges right we one thing we were clear is even whether it was our commitment to data bricks or AWS or whatever we said if we are going with them we do the full journey and that way it is not like we are testing waters our commitment was very ironclad so we said if you're doing it let's do it so Please don't start and stop.
I think having a stable environment I le said is good. Talent aware is very important because as you build up these just knowing spark doesn't make you an AI engineer right or just knowing. So that's so you must understand who is good what the skills are but everyone has a talent which we should harness obviously start small but finish big. The key thing about commercialization is u what should you refactor what should you shift that's something that we have said and the second point is probably the most important the willingness to change operating model um today demon spoke about AI having a seat at the table he said AI is such an important thing five years back I think Sicily and I were sharing some notes we had a vision to keep AI a clear differentiation we didn't keep AI as part of technology we said data and AI should have its own path so we are willing to even change our operating model how we run our businesses We are even changing how loans are being dispersed.
We continuously monitor and then like I told you my focus of my presentation and u my my whole uh my whole uh life at techcom you can say till we are all going to be there is our focus on business value because we are all technologists I'm still pretty hands-on I can quote but I think the reality is unless we all deliver business value whether you work for a bank or whether you work for a farmer or whether you work for retail you could even work for insurance if you don't deliver business value if you just say I've implemented data bricks.
I've implemented AWS or GCP or whatever and then you say it's a great technical implementation. It is it's a good toy. I can tell you that much it doesn't generate value. You need to focus on business value. I think generating value migration alone is not enough. And like I said continuous upskilling continuous upsklling. This is an area where we are probably learning every day. What was good today whatever I spoke today like I told you I did my presentation on unity catalog and kalibra.
I when I came here I realized that unity catalog can do much more. So now I'll go back and reinvent our whatever we have done. But you don't need to reinvent you can extend. So in short my friends first of all thank you for sitting through the presentation. I hope you're able to take up some value but most importantly like I said keep measuring business value always. Thank you very much.

Learn more about the Databricks Data and AI platform.