Microsoft Fabric: The Platform That Turns Data into Competitive Advantage

Speaker A: Foreign.

Speaker B: Good afternoon. Welcome to episode 82 of the Leading It Podcast. This podcast is for Australian technology leaders who need to cut through the noise on data, AI, cyber security, cloud infrastructure strategy and leadership. I'm your host, Josh Rubins, CEO, uh, at empyrean it. I'm joined by CIO Tod Laden and we tackle the fast moving IT landscape from both sides of the fence with pragmatic advice you can actually use. In today's episode, we're excited to have with us Microsoft corporate VP of Fabric, Scott Kesselman. And we're going to discuss all things data and AI. So welcome to the Leading IT podcast

Speaker A: and Scott, thank you Josh and Tom, um, for having me excited to be here with you today.

Speaker B: So tell us, tell us the listeners a little bit about your background in technology and how you sort of ended up in this role at Microsoft.

Speaker A: Ah, yeah, so it goes many years back. So I'm not going to ask to do a quiz how old I am, but I started early enough. Started as a software engineer even while I was in high school. That was a, uh, fun thing to do and also a good salary if you're a high school student to work in tech. But more recent experience again throughout my career I've been working with data all along as a software engineer. My current role again leading the fabric team within Microsoft. But before that I was with Google leading their monitoring and observability project products. And before that I was with Microsoft for around 10 years on the Power BI team, kind of building the product from the get go from day one, leading the mobile offering of the product and then the platform and the service. Again, super excited to talk about data and technology also in the combination how we can use this data to make our lives better in every aspect of life. So again, super excited about that, how we can use it for saving lives, healthcare, but others, those other scenarios and other verticals of the industry.

Speaker B: Yeah, excellent. And I mean every conversation we're having with customers is about AI and eventually we start talking about data. So why is data become the bottleneck for AI success?

Speaker A: So again, I'm um, an optimistic M guy. I wouldn't say data is the bottleneck for AI. I would say the data is the field that powers AI. And specifically when we speak about AI, we need to think about like we want the agents to reason about something and that something is the data. And when we speak about the data, uh, the big part around data is like how we bring this data in, how we have the data in the right quality and also an Important aspect, specifically when speaking about agenda experiences is how we're making sure that the data is fresh and also in high granularity. And that's a key part here because if we will use the data which is a day old, a week old and even an hour old and the ability to reason about it and take an action, it's going to be less effective. And the aspect about this around granularity. If you have aggregated high granularity data, that means that we're losing the power of the machines to really reason about all this data, really find the insights behind the data, finding kind of the interesting stuff that we humans struggle with. So we can work with, I don't know, five, ten parameters. We cannot process hundreds and millions of different parameters. And that's where we can use the power of machine. So, so that's where I see data becoming super critical for AI. And recently, specifically we understand that data is not enough. We need context and we need to have a curated context, curated knowledge to really power those agents to work effectively with the data. Because yes, lm, uh, are becoming very powerful. But if the data doesn't have the right context, then you're applying a very powerful tool that cannot guess what you want from the data. So, so that's another evolution in the last, I would say year, a year and a bit.

Speaker B: Yeah. I mean something that we're seeing with customers that we're talking to is, I don't know if it's correct to call it a bifurcation, but it's where you talk to them about what they're doing with AI and it's all about, well, we're using Copilot, we're using Claude, we're doing, it's all about productivity, front end productivity stuff. And then trying to sort of explain to them then, well, you actually need to get a data lake and to really capitalize on your business data and how you're going to actually differentiate yourself in the market moving forward. You need to look at your business data and then sort of opening the conversation up. Well, that's why you need a data lake, sort of. It's a bit of a, I don't know how, uh, you say, but it's a bit of a mind shift change for a lot of customers we're talking to because I think they're doing a great job but all they're doing is the front end stuff. They're not really getting their business data in place. Yeah. What were your thoughts on that?

Speaker A: Yeah, so I can share my experience kind of Working with customers. I meet a lot of customers, hundreds, uh, every year, individual companies. And I see a big trend change in the industry. So specifically we launched Fabric Real Time Intelligence around two years ago. And the beginning of the conversation with the IT people was like, I'm not sure that I need this, like why I need Real time, do I need it? But when I was speaking with the business people, I said yeah, of course, like I don't want to look at the data, which is stale. And also like why I want to look at the data, uh, I want to be notified when I need to take an action. But in the recent, I would say nine months, it's, there's no longer this kind of question even it understand with the power of AI it really changes how people think about what they can do. AI really opened the doors really for customers really to understand. Yes. All their dreams that they were having around data and ability to take action on top of the data and be pushed with actions and notifications that tell them exactly what they need to do and what to pay attention to. That's something is now really available and easy to do with AI. And that's kind of the interesting part. And now I see it even across it that this mindshade happened. Ah, and I agree with you, like working on productivity is great, but that's not where the multipliers are when we're speaking about companies. At the end of the day, companies need to improve their operational efficiency because that's what the business is. Because if you speak about companies, their value all up in those companies is in their ip. Their IP is like how they operate and run the business and they are multiplier on top of their cost of running the business. So if they can improve the operational efficiency, that means that they're improving their bottom line multiplier for their business. And to improve the operational efficiency that means they need to have tools that can reason and understand all their business data, uh, connected to events that are coming in real time or they need to act on when the event is coming in. And then they need to create a context for AI, not only automated in uh, a let's say deterministic static way, but also to have AI that can reason about all this data in a more of a ambiguous and machine learning way with an ability to reason using LLMs, a combination both of machine learning and LLMs and then recommend the right action. And that's where the magic comes in. And that's where we see companies that really maximize and improve their operation efficiency dramatically by A significant amount of order of beautiful.

Speaker C: Yeah, I love that multiplier concept as well. So we talk a lot about maturity of businesses going from personal productivity to then stepping up and trying to work through how do we embed this in our process and knowledge. You talk about machine learning a bit there and LLMs and I see constantly there's so much confusion around AI. They just think it's LLMs. Right. But do you want to just elaborate a little bit further on machine learning and are you seeing that coming back into fashion almost? Or is that.

Speaker B: And also you talking also while you're answering that, can you also for the listeners who don't know what the Real Time Intelligence product is, maybe.

Speaker C: Yeah, right. Yeah, that's right.

Speaker B: Maybe explain that because that is a product, not just an idea.

Speaker A: Yeah. So I'm sure if there are any customers that didn't hear about Fabric Real Time Intelligence. But jokes aside, happy to put, of course everyone knows.

Speaker B: Yeah.

Speaker A: So Fabric Real Time Intelligence was a product which is generally available for uh, almost two years now, a year and a half. Allows customers really to collect streaming data, event based data, reason about this data when event comes, but also take an action on this data. So a good scenario is that you want to run your supply chain. You have your orders, you have your deliveries and an event comes. Now instead of somebody looking at the data manually, now you get an event and then basing an event driven architecture. Now the system recommends you. Hey, the action that I recommend you to take is such and such. And we enhanced it recently with operations agents within Fabric Real Time Intelligence that you can describe in natural language, hey, you are my agent. I want you to monitor my supply chain. In case I see issues in supply chain, I want you to take actions and you have different capabilities, different actions that you can take to meet the business goals of yours. It's really kind of virtual team member that works and help you to run your business. And that's super powerful. There's a but going to the question of what is consists of two things. One is around LLMs and the other and the other one is around machine learning. When speaking about machine learning is the traditional, I don't want to call it legacy, traditional machine learning algorithms that are super powerful, super useful, they are deterministic, we know them for a while and they're very cost effective and they're powerful to use to really find insights behind this data. But where LLMs come into place, if we want to reason about our data, to really not just kind of find the insights because it's easy to find the incidents using deterministic rules or machine learning based rules. But you can use the combination of LLMs and machine learning and applying the LLMs over your insights from machine learning. Now you can reason and understand the important signals, what to do about it and what the right recommendation to take action. So going back to example of operations, agents, partner or fabric real time intelligence. Now you can get machine learning based anomalies or insights behind your data. But the action that you would take, it's not a deterministic action. It depends what is the root cause of this insight that came in. It's not one answer fits all. Now you need to reason, hey, if I have this event, should I cancel the order? Should I reroute the package? Should I replace the item? Should I escalate? So now we need somebody to take reasoning capabilities and that's where the power of LLMs come in into play and the ability to reason effectively. That's where we need to have the context, the understanding of the business to understand what is our business. And again we know this, we know how to describe our business. And with fabric iq, with ontologies we can model it as an ontology to present the business. And now the LMS not just look at the raw data, they have the context, they have understanding. Hey, I have a supplier, I have a customer, I have a package, I have a route. Oh, now I can understand what I can do all about that. And that provides really the context which is super powerful to LLMs to take the right actions and reason about it.

Speaker B: Yeah, that's good. We're going to get into fabric IQ a little bit later on. Definitely. That's really exciting. Okay, so let's get back to this customer. So in Australia, most organizations in Australia are um, what we call mid market. So one uh, hundred to a couple of thousand seats and a lot of them do not have a data lake. A lot of them do not yet have a data lake. So as a Microsoft partner and that has a data team that's focused on implementing fabric, you'd be glad to hear we're out there trying to convince CIOs that they need fabric and they need to have a data lake when they're still sitting in. Well, AI is, I'm using Claude and I'm using ChatGPT. And so how would you start? What architecture would you recommend? Sort of some, someone who's at that point maybe talk through what you think is a good way to go.

Speaker A: Yeah, and again I really love the Australian market. I personally visit Australia Two, three times a year. It's a great country to be with. So you're lucky folks in Australia working with your big companies, but also a bit smaller companies. And the stuff that I love about Australia is like how people are creative, intuitive and they want to really adapt new technologies. And again, I will challenge you a bit, Josh, about whom we should speak about. If you go to the CIOs, it depends on the maturity of the CAO. Do they understand the market and the landscape and what they need to apply? And then the conversation is becoming like, oh, do I need the data lake? Why I need the data lake? And then depends on their maturity and how advanced they are in the understanding of the landscape. But if you go to the business to see to the CEO or the CEO of the company or the CFO of the company, I'll tell them, hey, I can help you and help you improve your operational efficiency. I can use AI to help you augment your team with agents that can improve your operation efficiency and recommend you decisions find you.

Speaker B: That's harder to do. You talk?

Speaker A: I don't know, it depends.

Speaker C: Well, I'll tell you what, can I jump in here? Because now there's this new door that's open and it's called token cost. So the CFO wants to know why the hell AI is costing so many tokens and it costs so much. Are you saying that opens the door to say, well, hang on, let's talk about how we do AI properly and reduce costs. Does that help?

Speaker A: That helps. And also it helps to start the question of like, why do we want to use AI for? Like, AI is great, buzz, great. But what you will use AI for.

Speaker C: Yeah.

Speaker A: And that comes to easy conversation if you want to apply AI. Yes, you can apply AI, uh, on productivity stuff. Great, that's kind of 2023 and we are in 2026. Welcome to reality. And in 2026 you want to create agents that help you augment, uh, your physical team that help you improve your operational efficiency. And to do so, you need to have AI that works on your data. Josh, as you were saying. But you need more than that. Like you need to have a leg where the data is there and you support both types of data. You need to have a data that's coming in batch fashion and ability to consume events that are coming from different systems. It could be from your CDCs, from your CRM, ERP, from your call center, from your devices, IoT devices. So you need to combine both your kind of streaming and event based data together with your Batch data, all of them in one place. And once you have all this data in one place, now you can reason about your business holistically. Now you can decide what are the right actions to take. You can apply analytical stuff to create reports that will be consumed by your CEO, cfo, CEO. But also you want to empower your teams that run the business to have agents that work on this event based data, time series data.

Speaker B: Can I play devil's advocate? You suck.

Speaker A: Go ahead.

Speaker B: But I have Power BI running on my CRM already. Why do I need, why do I need a data lake?

Speaker A: Uh, amazing. So I never heard this question before. Maybe I heard it three years ago. Uh, and I love Power bi. That's a property that I built and feel very proud about it. And it's useful, but it's useful for the scenarios that need to be used. It used for analytical scenarios. It's something that I want to look at that and do analysis. It's not something that I will run and automate my business upon. If I want to get, if I want to get events and understand what the next action should they take. I need to have the right tool. Again, maybe analogy from healthcare, we started offline talking about that before. Like you need to have the right tool, you need mri, but you need also a heartbeat monitor. Like you're not going to use uh, an MRI for something that you need a uh, heartbeat monitor to know, know like okay, somebody is having a heart, uh, condition or heart attack. Like you're not going to use MRI for that. And the same applies here. Like you need to have the right tool for the right job. And in this case, if I want to run operations and I want to run beyond just analytical aspects, I want and not look at reports, I want to drive my business. And the good example there that again speaking of healthcare that I will bring is Apollo Hospital, the largest healthcare provider in Asia. Yes they had Power bi, they have the analytical stack, but they wanted to improve the operation efficiency. Instead of looking at the report every hour or two and understand, okay, what beds they need to clean, what customers they should call to make sure that they bought the medication, they want this all flow to be automated. And also using the capabilities of AI to find insights behind different cohorts of patient what is anomalous in this and using kind of fabric real time intelligence, they were able to improve their operation efficiency dramatically around 40% all up. I think it was even more than that. I think it was around 60%. That means now they can serve larger populations of patients. They can serve 60% more patients and that they couldn't do before. They were looking at the report every few hours. The data was there. But that's not how you run your business. You want to get an automated system that get. Gets the beds clean, that called customers when they need to make sure that they need to buy their medication, etc. So, so that's an example. Like if you want to go from analytical to operational, yes, you need to have analytical for. You need to have the analytics for the analytical side, but you need to have the tool to run your operations. And that's where I'm going.

Speaker B: It's like is. I feel, I mean, again, this will probably sound very 101 to you is what I'm saying is the trying to explain. You're right, we should be talking to the business, not the cio. So I agree, because our customers m are the heads of it, but it's trying to sort of explain the value of having the data from different systems together and getting context of your data together, not in a siloed. Because the issue, the obvious issue is that yes, they've got power bi but it's on one. They're only looking at one application. They're not looking at the data together. Do you know, do you understand what I'm m. What I'm m saying there?

Speaker A: Yeah, so, so that's a, that's a common thing. That again, that's where fabric comes into place. Uh, and very helpful with OneLake that you can bring together all your data across different systems. You can bring from your CRM system from ERP, as we mentioned before, also IoT social, other products in one place. And now when you're taking a decision, you're taking decision not based on one angle of your business. You're looking taking decision holistically and that's how you're making sure that the decision they're making are improving your business all up and they're not siloed to a certain kind of very steep view of your business.

Speaker B: Are you surprised that people are still asking these questions about the value of it surprises me a bit. I would have thought it.

Speaker C: Well, honestly, I've got a lot of customers who keep going back and saying, well, hang on, Data Lake is just a cost. It's another cost that the business don't really understand. That's the challenge I get constantly. It's like, hang on, we've got all this infrastructure now I've got to put another layer in and that's going to add more to my IT budget and the business don't interface with a data lake like we're trying to get out and say well AI will become more cost effective if you've got the data lake in place.

Speaker B: Also you're going to get the value out of AI.

Speaker C: All of a sudden your data knows about all the other data flying around

Speaker A: your business and the critical point here and that's my kind of advice, speak with the business because if you want really to improve the value for your business, improve how the companies works and run, make sure that AI is accurate, is effective, you need to have a data uh, you need to have high ground level data, you need to have fresh data, you need to have data across all your verticals of your business, across all your application, across all your domain, throughout all your company. And that's what makes AI uh powerful. Again speaking of companies like the cost of the technology, the cost for it for those data lakes, no matter what vendor they choose, it's like it's very small compared to the value that they can get if they improve the operational efficiency. And specifically when speaking about products like fabric which is much cheaper than the competition, it makes even kind of a no brainer there. And that's a big part here.

Speaker B: Yeah, we're seeing value at two or three grand Australian a month. That's not a lot. That's not a lot.

Speaker C: It's not, it's not a big deal anymore. But at the same time there's still a fight right for the money side

Speaker A: and fabric again and get example um because we are speaking with technology people, CDO, CIOs. This audience fabric provides you a very generous trial. You get an F64 which is around take 10k a month and you get this trial. Uh, you can try the product, you can try it on your data uh ah on your scenarios like define your business scenarios that you want to achieve. Use the product, you will be able to see the cost, how much it will cost you. Not in the theoretical calculators because those are like doesn't never work. You'll be able to see the exact cost but more than that you will be able to see the value for your business.

Speaker B: And from an architectural point of view how would you recommend is it still medallion or what do you think's the best starting point?

Speaker A: Yeah so my belief in architecture is around lambda architecture. So it's not medallion. I think medallion is great for analytics but we want to go beyond analytics so we need to have analytics plus operations and that's where we need the lambda architecture.

Speaker B: Please I'm not familiar with. Explain that.

Speaker A: Yeah, and maybe we can put a link of whoever, wherever we are putting this. But it was I mentioned before that you have an ability to have batch data in a uh, kind of uh, in a ah, bronze, silver, gold layer. But also you have a pipe where you get the streaming data, event based data and then you can provide not only the analytics aspect of it but also an operational asterisk for it so you can tie into automation and taking actions on top of all this data. So going back to my example from we can go back to the hospital example, uh, is that when an event comes that the patient was discharged, now you can trigger information, hey, I want to make sure that the staff is getting the bed served for the next patient. Or if I'm getting an event that this medication was not, this prescription was not bought in more than 24 hours since it was prescribed. Now I can take and take a call to these customers and there's a bunch of examples. So to move to a state where we have both analytical and operational, that's where lambda architecture is very powerful.

Speaker C: Yeah, that's really important. And we'll post a link to this as well because I think people are struggling with Medallion as being a little bit too simple for the use cases. And that's sort of what you're saying as well is like Medallion's only part of the story.

Speaker B: Well we're seeing that as a good starting point. But yeah, that's because I read a post of yours, you saying that a lot of organizations, it's like they're bogged down in infrastructure when they should just start with data. So do you want to maybe elaborate on that a bit what you meant there?

Speaker A: Yeah, when I'm thinking about kind of companies and uh, going back to the scenarios like if they want to have AI accurate uh, and useful and effective from them beyond just day to day productivity, they need to bring their data together like they need to bring their data together into one lake, bringing all this data into one place. But then the next step is like how we can automate and drive operation efficiency for the companies. And that's where real time intelligence come into place. Ability to connect this data in the lake with the events that coming in the streams of the data coming in and then automating create an action based on this data and uh, the next step after that is really create the context, the, the, the semantic understanding of how my business runs and operates and that's with, with ontologies within fabric iq which important part really to Power those agents to have something that they can run upon so they understand the entities in the business. They understand that those entities are ah, rich entities. They have time series data, they have measures, properties, but they have also ability to have actions. So again an example that we had with supply chains, if we have kind of a truck, we have an entity of a package, a uh, customer, a road, et cetera. If we have those entities, I can apply actions so I can reroute the truck, I can assign driver to the truck. So it's becoming a system where I can start working with and that's where AI come into place, where AI can get this context, this semantic understanding, what we call ontology and fabric iq. And really now say hey, I don't just have the raw data, huh. I don't only have the event that came in, I really understand, oh, this event that came in saying that this package is now getting late to these customers. I can take an action, I understand that in this case I can reroute the package because there is delay in traffic or I can take different actions and that makes really LLM, ah, super powerful and much much more accurate because they understand the context of the business. But again Tom, you mentioned cost and that's an important part. Tokens are not for free. We read a lot in the news about certain companies finding that out. Now I think certain companies get a very nice bills for their models and the stuff that we saw internally and I blogged about it, but also other companies that I'm speak with, they run their own analysis. So that's a common thing in the industry. If you have this context that provided to uh, LLMs. Now the LLMs don't need to guess and process and understand all those.

Speaker C: Right.

Speaker B: This is really important, what they need

Speaker A: to pull out of data, what's relevant for them to pull out. And um, having this context for LLMs, uh, becomes super powerful not only for accuracy but also for cost.

Speaker B: Right.

Speaker C: It means they run cheaper. Right. So same query without a lake is X amount more. With the lake it comes right down and there's a beautiful ROI for just there I reckon.

Speaker B: And I think so I suppose the steps are get your data lake in place using a good partner, just saying getting the right architecture in place and then overlaying fabric iq. So the Symantec ontology layer on top. So do you want to maybe quickly explain some Fabric IQ is a solution that recently went GA from Microsoft.

Speaker A: Yeah. So again fabric or lap and maybe we can put a kind of the architecture diagram or link but Fabric consists of kind of multiple layers. You have first the layer where you have all the data in ah, one lake and on top of that you have multiple engines that provide you from PARC for reporting. We have analytical engines like Data Warehouse, Lake House. We have real time intelligence, again as I mentioned before, that can bring those streams of data events, reason about them and also take an action and automate an action. But we have also fabric IQ and fabric IQ is the context for AI and consists again of three concepts of three layers. OneLake Again that's the raw data. We have the semantic models those are familiar with POWER bi, the POWER BI semantic models that allow you to have this analytical context representation for your data. And the top layer IS ontologies within FabricaQS and ontologies is a rich entity representation, as I was saying before, that allows you to model all your entities in the business. Those entities have both analytical data but operational data like time series data, uh, geospatial data. But they also have ability to have actions on those entities. And those entities are all interconnected to each other by a connection. And this relationship, this connection is a business relationship. Understanding that means that now you can apply different rules and policies on different entities because they're interconnected with the meaning it's not just a data key relationship. And if you provide the LLMs, uh, with fabric IQ with this context for AI now the LLMs are much more, as I was saying, much more accurate, much more cheaper and much more performant to reason what they need to do next.

Speaker B: And then, then the next stage is what we've been doing is just building data agents and operation, the fabric data agents and operations on top of that stack. Is that the right? Is that the right?

Speaker A: And that's what we provide out of the box with fabric data agents that allow us to have this kind of uh, ability to query and have analytical conversation with your data and operations agents. As I mentioned before that those are virtual team members on the team that run 24 by seven. Not only they can reason about data, but they can take an action with human supervision or without and really help you kind of automate and run the business.

Speaker B: We've been an example, we've been working with a very large bakery company in Australia at about 600 bakeries across Australia. And we built the data lake with a data like what is the most popular bread, it's a low GI white by the way, but like that ability for them to do that and which or what's selling best in this area or that area. And I think you're it's that it goes back to the point you said before that value to the business is like that, that's not a uh, CIO value but that's really business value and that, that, that can really change the game customers. And I think yeah, we're not quite at yet the operations agents because we, we want to obviously get there with customers but that's sort of the next step for us working with customers.

Speaker C: It's about trust though, isn't it? So once, once you've got the, once you've got the data and you're proving the analytics then people start to trust you with the operation agents. Right. So they go uh yep, you know what? This data is accurate. We live and breathe by this new data. Let's start doing stuff with it.

Speaker B: Right.

Speaker C: Is that sort of where you're at Josh? You're saying people are saying it's yeah,

Speaker B: once you get the data, get the, that self service. I mean that's a game changer for customers that I have to go to a data guy who has to pull out an Excel spreadsheet and mangle some crappy data together. And so that, that's definitely a game changer. Quick opportunity. Uh, so for customers who are about to do a data lake, I mean fabric is sort of starting to dominate. But as far as looking at things like databricks or snowflake. So is there a situation or a type of customer that fabric is not the answer for?

Speaker A: Again, not because I'm working for Microsoft. Again I believe in fabric and we build and again as I was mentioning, fabric has multiple engines inside. We have power bi we have the analytical agents, we have real time intelligence. And that's kind of the beauty of fabric. It's not one engine fits all. We have multiple engines for different purposes that tell that kind of all tie together as part of one product. And that's kind of the beauty of fabric. And on top of that kind of big value prop of fabric is onelake. So we believe in open source and democratization of data. Uh, the data in OneLake stored in Delta per K. It's an open standard in the industry. It's not our data, it's customers data. Uh, they can choose the site, they can use different workers from different companies to work with this data and they are not vendor locked in. And I think that's the beauty of fabric and more that fabric is much more cost efficient and most cost effective than another there. And again there's like third party benchmark that was done across the board where shows how fabric is much more cost effective there.

Speaker B: So what about if, I mean the customer's very GCP or AWS oriented, Is that a scenario where maybe fabric isn't.

Speaker A: We have again we have a bunch of customers that are, they're on GCP or AWS again I can name a few that are publicly kind of shared their stories. One of them is Veolia which is into energy and clean energy waste management. They're French based customers. They're like fully on gcp. They had nothing from Microsoft and they wanted kinda uh, improve the operation efficiency to have the system where it helps them to drive kind of their plants, the salinators, et cetera. And they did a very thorough evaluation. For a company that's fully on GCP to choose Microsoft, it's a big thing. It's not an easy choose. And they choose fabric, they choose Fabric real time intelligence because it was much, much more cost effective. And the work that we are doing within fabric, we make it really seamless and transparent no matter where your data uh, resides on AWS or gcp to make it very easy to connect to this data and really make sure that you can run your engines on top of this data uh, in one leg seamlessly. So that's kind of huge value prop from fabric. No matter where the data resides we can make it very easy to connect this data and run analytics, run real time intelligence, run agents on top of this data.

Speaker B: Yeah, I think the seamless connectivity is like one of the big selling points but there's almost not many places you can't get the data out of Shell. Do you want to maybe the governance and security piece. So do you want to maybe talk about that and purview and how it sort of works for the audience, the cio, CTO audience.

Speaker A: Again when speaking about data, uh, that's super important about like governance and security. And that's also one of big things that we allow within fabric is the ability to have security from the get go. OneLake within Fabric provides OneLake security. So no matter how you access this data, once you brought it into OneLake you get out of the box OneLake security, RLS and other types of security of this data. And from that point onward you have a full lineage how you, how the data is being accessed, who access the data, all the granularity controls that needed. And now if I'm it like I want to make sure that there is no data exploitation. So there's a bunch of capabilities that, that we invested a lot making sure that you have the right Permissions that you have the right security features and those are capable from the get go. And we provide a rich catalog, one catalog where you can see all this data that you brought in. And this catalog is embedded not only within Microsoft products like Excel and other and other Office tools, but also there is, There is an APIs where you can access all this data throughout third party SaaS products or agentic, uh, products through MCP, et cetera. So you can really all use this data no matter where.

Speaker B: Where do you see fabric going in the next sort of 6, 12, 18? Obviously there's some stuff you can share and some stuff you can't. But is there? What should we expect from the platform over the next little while?

Speaker A: Yeah, there's a ton um, that I cannot share or majority I cannot share. But again the focus is around making sure that we are moving in the direction of making sure it's very easy to work with all this data that you brought in with AI, making sure that you have the right context for you to work with AI. Um, and it's very efficient. And goes bigger to the story of around IQs, not only Fabric IQ, but other IQs within Microsoft Stack Around Work IQ that allows you to bring all your organizational data like all your emails, all your teams, chat. Finderaq allows you to bring all your documents, all your, all your policies within it. And kind of combining them together in a way where you now have agents that not only understand how your business work and operates with Fabricaq but also now you get with Workaq your organizational understanding and with FounderIQ your documents and policies and now those together now you can build agents that really represents how you work all up, how your employees work in the company and that's super powerful to create kind of those agents that are, help you to run the business.

Speaker B: And is there anything, I suppose you said you met with hundreds of customers every year. Is there any sort of things you want to, you can share of things that work well. Things don't work well around building a data lake. Any trends, anything that would I think would be really interesting for us to hear about.

Speaker A: So again the trends that I'm seeing all up in the industry is people are moving more and more into the operational space. I'm seeing it myself speaking as customers, but I'm um, hearing also from analysts uh, around the world that speak also with hundreds of customers. Companies want to be much more efficient. Companies want to make sure that they maximize their business, that they're tapping in, that they can achieve with their business but also using the Right. Technology. And specifically when speaking about AI, that they use all the data that they have, that they build the context that they need. So the combination real time intelligence within fabric, the IQ stack, bringing all this together, that's where I see the success story with companies. And again the product is very easy to use and integrate and we're seeing success from companies. Again, you're in the space for many years and uh, you know how long it takes to build a lake solution? I don't know, three, five, ten years ago. I think that those times are super compressed. And specifically with products like fabric make very.

Speaker B: Yeah, we have something where we call them accelerator that we can stand something up, getting things ready end to end in a few weeks. So yeah, so we're definitely seeing that. Sorry, Tom.

Speaker C: No, I'm just exactly sort of the proof of concept idea is, yeah, it's within weeks now, not months or quarters to be able to get value. The interesting point there is you're moving from analytics to operational examples. So that actually made the uh, the tools are now doing something for the business, something they can now measure and report back on to say we are now saving time, doing more, creating better experience for the customer as opposed to. Here's some insight, Mr. M. Business, go and do what you want with it, which is very hard to measure. But now you'd say it's actually turning into measurable outcomes we can then justify spending money on. Right. So it's a bit of a chicken

Speaker B: and egg thing, but yeah, I see

Speaker C: where you're going with that.

Speaker B: So I think these are the points really. I uh, appreciate you challenging me earlier around Talking to the CEOs and CFOs about operational benefits, but the audience. So we have a CIOs, CTOs and IT managers. That's also, I think a good message for them is they need really is. Yeah, they shouldn't be going and talking to the business about, well, we want to invest in a data lake. That shouldn't be the, that shouldn't be the conversation at all.

Speaker A: And I will maybe the message for the CDO CIOs is the following. Like your business people, they also have access to the Internet. I'm sure if you blocked it or not for them, they can still access the Internet. They read about it, they hear the success stories of others and they will come knocking your door and saying, hey, why you cannot do this? Why I heard this success from this company, like how you can do this for me? And now it's very easy to do so. So like I Promise you they will be knocking your door.

Speaker B: Yeah. So go get in first. Love it. Sure.

Speaker C: Yeah. Lead the conversation. Hey, Josh. Yeah.

Speaker B: And talk about the outcomes, I think. Yeah, that's. That was. Sometimes you say you just need to hear it told straight. But saga as far as where can people find you if they want to see your work follow you, what's the best place?

Speaker A: It's on LinkedIn. I'm sure you will put, uh, a link there. It will be published there as well. So, yeah, available for folks kind of to reach out and happy to connect, happy, uh, help customers and more than happy to get the feedback. Again, we're not perfect. We're improving, we are growing and feedback from customers. What works, what doesn't, how we can make it better. Super important for me and my team. So, again, appreciate any feedback there.

Speaker B: Good to hear and thank you so much for your time. Really appreciate you joining the podcast and giving us your valuable time and insights.

Speaker A: Of course.

Speaker B: Thank you very much.

Speaker A: Thank you.

Speaker B: And to, uh, our listeners out there, thank you also for your time. If you want to get in touch with it, you can hit him up on LinkedIn. We'll put a, we'll put a link in the show notes and also obviously you can contact Tom or myself on L if you want to talk about your data and AI projects. For everyone out there, good luck with your data and AI program. Thanks very much.

Speaker C: Thanks. Thank you. Thank you. It's great.

Speaker A: Really appreciate it.