Good morning, good evening, good afternoon, everyone. Welcome. I'm Manoj. I'm a Principal Solutions Engineer here at Atlan. I work with a lot of companies to help them with their data, metadata, and governance strategy, and so on. Before we jump in, let me introduce our panelists today. We have a couple of awesome speakers with us today. We have Roy Hassan from Microsoft. He's a Senior Director of Product, focusing on the data and analytics space. He's instrumental in shaping how companies use Fabric, Power BI, at scale. Great to have you, Roy. The second speaker we have today is Chandru Sugnan.
He's a Group Product Manager at Atlan. He focuses on governance, lineage, metadata, and catalogs. His team has been working very closely with Microsoft to build a tight integration between the two products and companies. So thanks for being here, Chandru. So let me set the stage for what we are going to do today. Fabric and Power BI have been accelerating how teams build and scale analytics across the enterprise. Adoption is growing rapidly, but as adoption grows, you know, Microsoft and Atlan started working together on how to bring in visibility, how to bring in lineage, how to bring in governance end-to-end, and how the two products work well together to solve real business problems for you.
So that's what we are going to show today. Microsoft and Atlan work together to bring unified context across your Microsoft stack to your BI and AI applications end of the day. So before we get started into the actual content, a quick housekeeping note. This webinar is being recorded. It will be shared to you after in the next 24-ish hours. So don't worry about like screenshots and everything. All of this is recorded. You'll get a note after this. You can use the chat for comments. We have teams from both Microsoft and Atlan monitoring the chat.
So any questions, go for it. We will get you. But in case we need to do something verbally, longer answers and so on, we will have time towards the end of the call where we can do live Q&A. The three of the speakers we have today, we will kind of talk about that. We do have a couple of polls that are kind of put in different places to try and understand where you're coming from so that we can cater the session accordingly. But also, we can understand what kind of role you're working on, what kind of work your company is doing on, and how we can help accelerate those use cases.
So without further ado, I'll launch the first poll right now. You want to get a quick sense of your AI strategy and where you're heading. So poll is open now. You'll have about a minute until which you can answer that. We'll give a 10, 15 seconds break for you to answer, but it'll be open for another minute.
Chandru, you have access to the slides now? Hey, hello, everyone. Cool. Good morning. Good afternoon, everyone. I think I have the right slide in the text. So let's round on a simple truth. In this AI era, data isn't the bottleneck, but the context is. That's the theme we are going to work on. Every enterprise already has data. A lot of them. I bet you all agree, and scream if you don't agree in the Zoom chat, okay? What we don't have is a reliable way to answer the questions. Like, what does this mean?
Who owns it? What's sensitive? What can I use in order to avoid? And let alone the why questions. We are not even talking about the why questions yet. What itself is, we are stuck, right? And why a number has changed since last month. And here's what I have heard from our customers in several of our executive briefings. The faster the teams move with modern data and AI stack, the more often the what, why questions come up, more pipelines, more semantic models, the reports, more AI agents quitting your data. And yet, the moment of doubt, the pause while you are showing some numbers to your CEO and executives hasn't gone away.
And if anything, and this is because the scale and speed don't automatically bring context and without context, AI is just guessing. So, let's make this concrete. In this slide, you will see, you know, there is a bunch of things happening, right?
So, if you see at the top, let's imagine that you have someone in your team, a human or an AI agent that receives a request. Identify the top 10 new shows to feature on the homepage. Now, it sounds simple, right? But let's see what is actually needed for the question to be answered correctly, who is asking it, and what decisions are they making? Is this marketing asking the question for optimizing the number of views or the editorial optimizing for watch time? It's same question, completely different answers, right? Which is the user context.
And then what's new even mean? Is it released this month or it's like new to the platform in last X number of days? That's the knowledge context. How do we define top 10 even, right? That's the next thing. Is it based on a raw view count or an aggregated time, a watch time? This is the meaning context. And let's come to the data now. Is the data in a table called raw logs or analytics or in stream, which is actually the data context? And how exactly do we even measure viewership?
Total place, unique viewers, more data context. And there's different layers and layers of context. And if AI don't have this context, it gets it wrong. And the very scary part is when AI gets wrong, it doesn't guess or it doesn't tell that, oh, I'm guessing. It just comes out very confidently with a wrong answer. And this is the AI context gap. And this is the most important unsolved problem in the enterprise today. Let's move on.
So, if you see here, how do we solve it? There are a few building blocks. Atlan thinks in this way, these building blocks. Do we need more data and dashboards? I don't think so. It's the missing context layer, which is our theme. A persistent, connected, living layer of meaning that sits across all your data assets and feeds to every other AI system and analysts, the human in the loop, and their decisions. So, this is how we think about it. First, you have your sources, the system of records, the system of, you know, data, layers of semantics, knowledge bases, their system of work like Jira or Slack or Teams or where have you or collaboration happens.
We connect to those systems and bring together into what we call it as an enterprise data graph, which is enriched by lineage, data lineage, SQL query logs, pipeline codes, BI semantics, tags and quality signals, the business glossaries. This is all the foundation. And then we annotate with AI, not manually, which is the old world, right? We annotate with AI-led bootstrapping or stewardship, you can call it as. We use, you know, things like descriptions, link the business terms to your physical assets, surface popular filters and joins, create semantic views, top business questions, your organization is already asking.
All this, almost like, you know, 80 plus percentage of your heavy lifting of stewardship is done, which is in a fast way. And then the next thing that comes is bounded context, which is like, this is where enterprise leaves our time, right? Like we bring in the humans in the loop, your enterprise, humans in the loop. And because AI can bootstrap, it cannot certify and it cannot resolve a conflict between two definitions of active users, right? So we make the human decisions. And finally, in the bounded context, you can see the collaboration environments where, you know, all this resolution happens, conflict resolution happens, labels happens, certifies.
And let's just take this governance flow. What it all holds together, the collaboration and governance flows. And this, let's take this real scenario where your AI blog, AI bot flags a conflict with two different calculations. Let's say the name of the calculation or the metric is customer weekly active metric. One includes lineage page views and other don't. So the AI pings the right person saying like, hey, is this wrong? Is this correct? And the human in the loop, Mark, don't know. He's not the decision maker. So he escalates it. And the decision is made by Sally.
And then the bot, the AI bot picks it up. It gets, it committed and it pushes down to every downstream report, the AI agents and forwards. And your AI answers a question about a win rate. And this is with a medium confidence because it founds two versions. It proactively flags it and it routes to the right person. The owner now, the human in the loop conforms the new calculation and the old one gets deprecated and it just gets persisted everywhere in the loop. This is how the governance actually works because it's embedded and it's active business, active within the business workflows and not just ported on and afterthought.
So let's come to the final picture. So wherever we grounded on, right? Like we grounded that the context is the king. So the unified enterprise data graph with AI bootstrapping context at scale, human certifies in the AI loop, the governance flows, keeping it honest. And then the activation through the APIs, the SQL interfaces, the SDKs and what have you. And through this feedback loops, the layers get smarter every time someone asks the questions.
And this is not a catalog. This is just not a glossary. But this is an intelligence layer that makes every AI system in your enterprise trustworthy. So that brings up to the Microsoft Fabric and Power BI integration or the Microsoft ecosystem integration. And in this case, you know, thanks for Microsoft Fabric doing a phenomenal job in, you know, accelerating the entire data to analytics lifecycle into a unified AI platform. So when Fabric is your engine and Power BI is your decision surface and Copilot is your interface, the value of context layer multiplies dramatically.
Every query, every Copilot runs, every semantic model creation that engineers build, they all draw from the same pool of meaning. And Atlan is that pool of meaning. We are the context that makes business accelerate the data decisions at scale with AI. And if you go back to that question where we asked, like, you know, the pause you had when you go in front of your CEOs. Now with this integration, you can confidently go in front of CEOs because the context is backing the KPIs that you're showing to the CEOs.
I'm going to hand it over to our friend, Roy. Thank you for being here. Please take it away. Thank you so much, Andro. I appreciate it. So this is not me. This is actually Satya Nadella. But the reason I want to call this out is that Microsoft Fabric has been one of the fastest growing data platforms in Microsoft's history. But also being fairly new to two and a half years has evolved tremendously over that period of time. And we're seeing tremendous growth over 60% year over year to 31,000 customers already using the platform.
So that kind of shows you maturity and capabilities are actually meeting the customer needs. So as we go through this, Microsoft Fabric, if you're not too familiar with the platform, you can think of it as a vertically integrated data and AI platform. It provides a set of compute capabilities that you see on the top here, like data factor, analytics, databases, et cetera, built on top of a unified data layer called OneLake. And within that data layer, there's governance built in, as well as co-pilots that's across the entire stack. So I'll kind of talk through each of these layers here with a little bit more detail.
And hopefully that can kind of resonate why Adlan chose Fabric as the platform to bring their capabilities. All right. So as I said before, OneLake is that unified storage layer for all of Fabric. And one of the main reasons why we chose to do this is because our customers have told us traditionally that data silos cause problems. And of course, data lakes and other solutions like that actually help break those data silos. But the problem gets kind of kicked down the road where now that you have a single place to store your data, it's unmanaged, right?
It's difficult to manage, difficult to deal with. You end up with just a data swamp, a lot of stuff in there. So OneLake basically set out to solve that problem by saying, we'll give you a unified storage, but we'll also make sure that it's managed. We'll distinguish between files, Excels, PDFs, PPTs, et cetera, and tables, right? Delta table, Delta Lake tables, iceberg tables. That data is separated and it's managed for you. So all of that is under control of OneLake. So you don't really have to worry about managing your data.
You put it in OneLake and it just becomes available for all of your data consumers. And as we build out this ecosystem, we want to be able to enable users to bring data into OneLake with a zero ETL type of mentality or approach, reducing the complexity of building data pipelines, doing ETL and transformations, dealing with lots of connectors. So we're working with the community to kind of build this out and provide a single unified layer so you can take advantage of it through the Fabric ecosystem, as well as the partners like Atlan who are integrating into Fabric.
So as I said before, you know, the core is OneLake covering that. And then you can see that the arch above that is a lot of the Microsoft services, like the M365, you know, we got Teams, Power BI, Excel. All of them are integrated into OneLake. So whether you have files there, PDFs, Excel, or tables, that's available for you from the Microsoft productivity suite. But as we build that integration with our ecosystem, the arch that you see on the top, those are all partners that we recently onboarded who have integrated into OneLake through our standard APIs to be able to read and write data from OneLake, giving you a lot more capabilities and options to access that data.
And the way that we do it is through open formats, right? If you've heard of Apache Iceberg and Delta Lake, these are open table formats that make the data that you store in OneLake very portable and accessible. So any engine that supports these open formats can simply connect to OneLake and start using it without any custom integration or interoperability. And further with that, as we're expanding more towards Iceberg, Iceberg also allows us to define a set of APIs that let us talk and communicate and sort of federate with external systems, be them AdLan, be them Snowflake and Databricks and the rest of the ecosystem in a standard language.
So that makes the ability for us to connect OneLake into more and more systems very easy and quick and simple. So from a user perspective, it's a simple like click, connect to OneLake and that's it. It just works and it's consistently work across the entire ecosystem. And that's a very, very powerful sort of tenant that we decided for OneLake is that we want to be interoperable and we want to make it easy for users to just bring their tool and start using without having to build a bunch of best practices type of lake designs and configurations and ETL jobs to optimize stuff.
We're just doing it all for you under the hood. So next step is, you know, now that we have the lake, unfortunately, there's a long tail of data sources out there that not that easy to build zero ETL type of replication for them. So data factory is one of our primary tools for bringing this data from external systems into OneLake and into our fabric ecosystem. But it also gives you the ability to do transformations, prepare the data, do data quality. We recently announced something called dbt jobs, which allow you to bring a dbt transformation SQL and actually run it inside of data factory.
So all of the same code you've written and you've executed on your own running on Airflow against Snowflake or whatever, you can now bring that into data factory and you no longer need Airflow for orchestration. You can just, you know, still work with Snowflake. You can work with the fabric warehouse. It just makes it very easy for you to bring data in, transform it, prepare it, and get it ready for your customers or for your users. And then, as I said before, like on the analytics bucket, there's the fabric data warehouse.
There's also Spark. So you can have fully managed Apache Spark with notebooks. But there's also a very powerful fabric data warehouse that lets you run all your queries and your analytics. You can embed this into your applications and your services to our SQL endpoint that is kind of like, you can think of it as like as a headless data warehouse that just runs on top of, run SQL on top of your data in OneLake. So interacting with the data is super, super easy. And then as we kind of move forward, you've probably seen this in the industry where many of the analytics platforms, like Snowflake and Databricks and some of the other ones are realizing that customers want the analytics and the transactional, like OLTP and OLAP, to kind of coexist in the same ecosystem to be well, well integrated.
And some of the approaches others have taken by combining those two together have some benefits, but also have a lot of drawbacks. And so we sort of took the approach of, you know, Fabric is that unifying platform. We can bring both OLTP and OLAP into the same platform, integrate them really well, integrate them with security and governance and writing into OneLake through zero ETL replication, but still let you take advantage of each of these platforms, each of these engines independently. So you get all the best of all the best capabilities of each one, and you don't have to make compromises, but under the covers, they're all very well integrated inside the Fabric platform.
So Fabric database is an area that's growing. We got Postgres, we got NoSQL databases as well through Cosmos DB, and others are coming out as well.
So definitely a growth area for Fabric. Fabric IQ is something new that we announced at Ignite in 2025. This was in November, and this is about taking the context and the semantic models and semantic information that exists across your ecosystem, right, and sort of breaking the silos of that context. So as I said before, OneLake broke the silos of data. IQ is breaking down the silos of context and ontologies and semantic information. As you look across the ecosystem, I think you'll probably see this as well. A lot of the companies today who are doing AI and building agents and co-pilots and things like that have to build their own semantic graph, their own semantic layer that shares additional information and context about the physical data.
But now that's creating these silos. So Fabric is really designed to solve that problem, sort of unify all of that into a single service that has a set of APIs that allow you to publish and consume the context, linking it to data that resides inside of OneLake. And then ultimately that package, right, the context and the data is available to the entire Fabric stack of services, Power BI, data warehouse, data agents, as well as partners like Atlan who are integrating with the Fabric ecosystem. And then Power BI, I think everybody knows about the massive, massive products, been out there for a very long time.
Lots and lots of users who are using it for, you know, for visualizations, for dashboarding. Now there's co-pilot in it so you can actually talk to it and get it to produce a lot of the work that in the past, you know, users would have to click and drag and drop and add icons, widgets to their screen. Now they can just talk at it and just get it done. So all that culminates in the business users actually getting the value that they want. And as Chanju talked about it, and Manoj will show you in a demo, you can, as the data moves from the ingestion of data factory through the storage in OneLake, through some enrichment and other components, and eventually lands in a Power BI dashboard or a Power BI co-pilot, there's a process there that the data takes.
And as Atlan kind of looks at all of that end-to-end, give you the lineage and give you the ability to really understand that the thing that you're looking at, the metric you're looking at at the end of the day, is it accurate? Is it correct? Is it built properly? And that's really the end-to-end message here. So with that said, Manoj, I'm going to hand it over to you. Thank you, Roy. Chandru, if you want to take us home on this one, and then we can jump into the slides from there.
Yeah, I think this is like a summary of like, what's the context layer for Microsoft, about Microsoft we have in Atlan. This includes the discovery of all the systems of records, in this case, whatever is in Fabric, Power BI, and other Microsoft systems. The lineage, where it comes from, where it is going for your root cause and impact analysis, the governance around its access control, the governance workflows, the policy center that Atlan provides out of the box, the data products marketplace, the data quality, governance, and the future-proof iceberg native integration.
So that's the, you know, the circle of life of what we have at Atlan. And Microsoft Fabric, Power BI integration. And without further ado, Manoj, please take it to the, you know, the everything in action. Thank you, Chandru. Thank you, Roy. We asked a polling question earlier in the session. I wanted to share the results with us and then kick off another poll. So a third of us in the last poll told us that we're building internal AI applications using LLM. These are internal apps that are currently being tested. So a third of us gave that answer.
Another third said, like, we are doing a mix of buy and build strategies. That was the second one. And what came a third was, like, a fifth of us said that we're still exploring. We don't have a strategy, but we are testing different mechanics. What was surprising to me was the two areas that pretty much nobody was working on. Maybe one or two of us were working on was buying off-the-shelf AI tools for internal use case or building external phasing AI products. So these were the least scored items in the poll.
The other ones were about a third or a fifth and so on. So it's time for another poll, the last poll, I promise, before we go into the live demo. Ayushi, if you're there, if you can kick off the next poll, I'll start giving the context on the demo and we will go from there. Thank you. So what we're going to see today is a end-to-end view of a typical data pipeline that starts with data sitting in a SQL server in any database on the left-hand side.
As an example system that data passes through, we've picked Snowflake as an example, but Microsoft Fabric is your central unified data platform for the use case that we're going to show through. So the data would sit in one lake, data factory pipelines will be moving the data, the warehouse, the Fabric warehouse will be serving the data to the end users, and then Power BI is our reporting decision layer at the end of the day. Where Atlan would play hand-in-hand with this end-to-end unified data platform is to give you context on all of these.
How do we bring in unified lineage across desperate systems, systems that are across the board? This is a sample representative example, but think of your real pipeline having 10 other systems, 10 other hops before a dashboard shows up, right? How do you create that end-to-end, create and see that end-to-end lineage for one 10,000-foot overview? But also I want to go double-click on things and understand how did the data get classified? Is there sensitive elements being shown? Is this related to GDPR? Which are the certified assets? Is it certified data product that my business users can self-serve and consume?
And how do we activate governance? If you have seen another demo from Atlan, at any point in time, we always talk about activating your metadata. So how do we use metadata to deliver value to your end use cases? And how do we build AI use cases on top of it? Which is what Chandru covered earlier in the session. We're going to pick one sample metric for the first demo, first part of the demo. We're going to pick net revenue retention. We're going to see that from a Power BI world in the first part of the demo.
And how did it come into existence? What are the hops it took? What are the columns it touched? I saw some of the questions on the chat where we talked about column level lineage, how Atlan can help curate that across systems. And then we'll go into looking at two interesting use cases. We call it root cause analysis. Why something broke? And why an impact analysis? If I change something, what are the downstream things that it's going to break? We will look at how Atlan and Fabric are tightly integrated. We look at the integrations, what kind of lift someone has to do.
Kind of a spoiler alert, it's you don't have to do a lot, but we'll see that in the demo. I'll pause there. The last slide for me, we're going to jump into the demo from here. So I would love always to look at it from a top level view, a 10,000 foot overview and start from the business users, the executives out of the box. So the first example of use case you're going to see is a executive who's looking at a portfolio risk dashboard.
He or she is a CXO in the company. They are looking at revenue. They are looking at making critical decisions based on this report. It's an example report. Don't dive too deep into what each metric means or what the dashboard is and so on. But think of this as your CXO in your company, looking at this dashboard, making a very critical business decision. Power BI is amazing at providing you trusted dashboards. But then if I'm the executives, I would ask a question like, OK, this is awesome. The dashboard is perfect.
But can I look at how this data came into existence? What are the hops it took before I see the final number that I'm seeing here? Was there a data quality issue along the way? Or actually, am I at the right dashboard? Like, yeah, I know my team had another version of the dashboard a couple of months ago. Did I accidentally open the wrong bookmark? Like, can I get a view of certified assets at the end of the day? The first value add, the first moment that I wanted to show was Atlan has a built-in browser plugin, native integration, supports Power BI out of the box.
So all the executor has to do in this scenario is to go and click on the browser plugin on the bottom right. It follows where you are. So in this dashboard, we are on this page where it kind of shows up. And it explains kind of a 360-degree view of this entire Power BI dashboard. Tells me what data products it is part of. So there is a product called Finance Master and there is a product called Executive Quarterly Overview, which is having a green tick mark, which is basically a certified data product.
So as a business user, I know that I'm looking at the right dashboard. I know more information on what this dashboard is about. Any business context that one would embed into the dashboard is automatically shown by Atlan. It gets bubbled up. There might be terms here that need a little bit of translation to the business user. So net revenue retention. We had one customer of Atlan who had 12 different definitions of the term net sales in their company. Like 12 different, right? Like the definition is somewhat similar, but not exactly the same.
The formula to calculate a critical data element might vary across teams, but the best way to kind of centralize them is through a business glossary. And Atlan has built-in business glossaries that pulls up the terms right to the dashboard. So if I'm looking at what does net revenue retention mean here, I can simply hover over the term net revenue retention and any formula or any descriptions of that business term is exposed to the end user. There's a lot of information here, but right now I'm just going to go through a couple of them, right?
I want to understand who owns this dashboard. I want to ask that person a question about how the data came into existence. Is this updated by per quarter, per week, per month? So on and so forth. So I can see who the owner is. I can see which department owns it as well. I can see that this is a certified asset. But let's say all this information doesn't answer my question. I feel like for whatever reason, this number may not be adding up to the right number, right? Like I want to see if there is an upstream data issue.
Maybe my source system, my system of record has a problem. And I want to see like, how did this come into existence? I can click on the lineage view or I can click on the native integration we have with teams here. So I click on the teams button here and I can see that a set of our colleagues have asked a similar question. I can see a friend Ryan has asked a question three months ago and he calls out the question that I had in mind. Like he asked, like, it looks suspiciously high.
Like, can I know what the problem is? So I can quickly jump into that team's thread without having to, you know, leave anything. You can look at that from here and so on. But let's say I don't have my question answered. I can simply type in here and I can ask a question to the channel I want and I can attach it back. So tomorrow, if somebody else has the same question, they know this is what happened. This is what the response is. If I take a step back, what we're trying to say here is just like me, who has 100 mobile applications on my mobile, I have fatigue on jumping from one app to another app to another app.
Atlan doesn't want you to face the same thing in your BI, AI work that you build. So all the context is embedded into the place where you're doing your work. In this case, it's Power BI where all, you know, the native governance capabilities and governance metrics and metadata are bubbled up to the end user. Now to the next part of the demo, let's take a look at how this data came into existence. Let's look at lineage. So you have a lineage view here. This is a textual view so you can filter out on the depth that you want.
But if you prefer a graph, I can simply click on the view graph which takes me back into Atlan where you can see a clean end-to-end graphical view of this dashboard. Where did this dashboard come from? What are the sources that it touched? And I can kind of expand this all the way to the left, shift left, and I can understand the entirety of where the data came from. So I can see all the hops it took and I can see all the stages it took all the way to the left where the data was loaded from a database and Salesforce and so on and so forth.
We'll take a slightly expanded example on the screen right now. This is the second part of the demo. So we kind of touched upon the executive overview dashboard earlier. So I'll leave it there. But I'm going to focus on the second part of the workflow for this part. So the data is being sourced from a lake house table that is on Fabric. You can see from the logo. And it's a certified asset. It gets published to the user. You might be asking a question like, so what? What does this help me do?
Pretty picture, awesome graphics, awesome UI. What does it solve for me? So the first use case is when something breaks, you would get to know why it broke. Let's say there is a data pipeline. In this case, there is a Snowflake query that is hydrating data into the lake house. Or if I go all the way upstream, there might be other applications. I might have a legacy SQL tool. I might have dbt, Fibran. There might be different processes that are publishing the data sets. Atlan helps you go to that processing level as well.
So on the sidebar, you can look at what these queries are. And if you want to know what the query is and you want to get it explained, I can simply click explain with AI. So a user doesn't have to read SQL to understand what the transformation is. They can simply get an AI explanation of what that lake house table data eventually contains and so on. Now, the second part of that, you know, the question on like, why does this matter is, let's say there is a column and let's say I make a change upstream that broke something downstream.
Let's make an example. Discount is a column. I want to know why the value of discount seems a little bit off. So I can see when I click on it, it shows me downstream on all the things that can change if I change this column. So we call it impact analysis. So when I click on view impact, it tells me all the downstream applications that are impacted. Let's say I'm an engineer who's making an update. I can see why or what will break if I change this. So I can download a report.
I can talk to my teams. I can talk to my other teams on like, hey, this is going to change. What are the things that we need to do before getting there? Now let's take a step back. We started off at a 10,000 foot overview, an executive overview where we don't go too much into details. We just see the final outcomes. We double-clicked, double-clicked, double-clicked and got into a lineage view. We looked at how teams can adopt Fabric, but to get an end-to-end view and get to a trusted dashboard, a tested value metric, KPI, end of the day, Atlan helps you perform this with embedded context.
Now, the next part of the demo, I'm going to jump into Microsoft Fabric. The same pipeline, the same data that we saw earlier, this is the pipeline that's actually powering it. We are looking at Microsoft Fabric and all this data sits on OneLake, your central data storage layer. It federates across different catalogs and systems and all that data is available in a central place. So the pipeline that we were looking at is simply three different steps in getting data. In my case, I'm getting some data from Salesforce, some data I'm getting in from other lake houses and so on.
We are preparing that, analyzing and training that with the help of notebooks and finally publishing it to a data warehouse and a lake house and finally we are visualizing it with Power BI. If I double click and jump into a linear view, I can show you an example of what we have done in Fabric to get this pipeline ready. So I'll pick an example. Fabric helps you kind of give that end-to-end data pipeline view. So I'm in that data factory data pipeline and when I click on that lake house, this is the final consumed data that you can see.
So I'm going to go into the table view. Data is coming in from Salesforce. So this is account data, customer data, sales data and so on. So Fabric brings that end-to-end data view into a single page. We let it prepare the data, get the data prepared and live store and eventually report on top of the data. The one value add I wanted to show was not just on Power BI but on Fabric itself, whatever page you are on. I can go and click on the Atlan plugin which is again going to show up at the bottom right and it brings you embedded context into your platform.
In this case, Sales Lake House is the page I'm on. Atlan brings you embedded context here and you can enrich your metadata as you go along. For example, we use Atlan AI to expand a description for this table. You can edit your own description but you can ask AI to kind of based on all the information you provided can Atlan predict and give a suggested metadata here and then use as a human in the middle can go edit the metadata descriptions and so on and apply but you can enrich your metadata from there.
I go back to the page, the previous page, the metadata bar works with me to go back to the sales order analytics page. No data or metadata, no data moves between the two systems based on the URL you are on. Atlan is able to pull in the metadata and context for the users here.
Now for the last bit I want to jump into how all of this exists, right? Like I showed you the outcomes, I showed you kind of a executive view, I showed you kind of a technical view on the data pipelines and so on but I want to show the final part where how we got and curated all this information. So for that I'm going to go to the homepage of Atlan. I've logged into an Atlan instance right now. I'm going to go into the connector marketplace. So I click on new workflow.
I see there are 113 connectors here but right now we just want to connect to fabric. So all I have to do is search for fabric. If you have an Atlan instance it may not show up right now but let me know on chat if you need to but it's very straightforward. You can go ahead and say fabric and the connector which is available will kind of show up here. It's a simple three-step process. I can click on setup workflow. I can talk about first as a credential phase I tell it's an integration and what client ID secret I use to connect and what's the tenant I need to connect to and then I can filter on what metadata I want.
Let me jump into a workflow that's already running. I ran it today a couple of 12 hours ago actually and it's basically what I showed right. So users connect to the fabric instance with the service principle secure it's metadata only and I can filter on different workspaces if I want to. In my case I just said everything. I have four workspaces bring me all the workspaces into Atlan. With the workflow page you can monitor the previous runs. I've had 10 runs totally. All the last five runs have been successful. I can look at how many assets have been crawled like 7,000 assets that were crawled last week and since then I didn't create a lot of new stuff but that is here.
Now if you're a user who wants to identify new resources you can jump into the assets page I'm going to clear all the filters right now. So what I see on this page is a sum of all your assets across all the technologies you have in place have 100 quarter million assets across the base. We are only interested in fabric right now so I can simply go and filter by fabric and if I want a specific workspace I can go and filter by that too. I'll leave it default I'll see all the workspaces we still have 5,000 assets.
Me as a user want to identify only certified assets let's go and filter by certified slash verified assets. We now see four assets right like it is almost like you're on Facebook you're uploading a photo it is an in-tudo interface to go click this button click that button click this button go and get to your final state. I could simply go ahead and you know sort things by most popular I can go to a specific asset in this case it's a lake house table and I can look at a 360 view of that asset end-to-end so I can look at all the information I can look at all the columns the descriptions like I said earlier if you want AI to help you enrich some of this I can go ahead and do AI enrichment I can look at classification policies and so on and so forth.
Without further ado I'm going to go back and summarize what we saw today and then bring us home from there. So we saw how a complicated data pipeline with a representative example data coming in from a source system flowing through Snowflake into the Fabric ecosystem data is stored on one lake and eventually Power BI helps us report. We looked at how Atlan can enrich this end-to-end journey how it helps business users technical users governance and BI and AI to get different use cases out of it. Having said that we are going to jump into the final slide here which is a Q&A section I'll keep this QR code open so if you want to go deeper into this if you want to go deeper into a demo more than happy to do that please scan this QR code and book your demo but in the meantime we'll open up for Q&A.
awesome. So we have a couple of teams here who have been helping answer a lot of these questions. So if one of you Ryan or Dan or Ayesha if you have any questions that are left over that are not yet answered if you want to bring it to the audience that would be awesome. If there's one question that we want to bring up because everybody's interested in the answer. Do you want to maybe talk a little bit about I think the question came up a couple times about Atlan and Purview kind of what are the differences how you guys integrating with that?
Would that make sense? Yeah maybe I'll take that question. So I did respond directly as well. By the way Roy the question is to me right? I will yeah. Yeah. Okay so the way I would put it in is like a two-click stop. The first one is you know Atlan have native integration with Fabric Power BI and the overall Microsoft ecosystem you know plug and play connect and point and forget we will extract all the context and drive things forward and similarly there is some integration with Purview Purview connects with Fabric and gets some context.
Now the delta of the context itself I would I would wonder what's the delta of the context that is in Purview regarding the Microsoft ecosystem that's not in Atlan irrespective let's assume there is some delta you know there is underlying APIs of Purview is open and same as with Atlan and there is clean interoperability on the API level so we can bring in any additional context that's missed out and available in Purview and not in Atlan so it's a custom integration we can help support please reach out to us what that context is missing in Atlan that is in Purview and we are happy to help out there.
I wanted to summarize one of the things we're seeing in the chat a lot of excitement around conversing with your data and metadata so if you can speak to how we're approaching that I think that would be great there's good interest from the team okay that's an excellent question thanks for calling that out Ryan I'm going to do the risky thing of like trying to do a new demo live pun intended it is pretty straightforward to do a natural language discussion with your metadata so Atlan has what we call a conversational AI or chat interface so the new home page of Atlan I click on the home page it starts with a conversational AI page if you remember the last part of the demo that I was doing I was doing filters manually this this and that but you could ask the same question to Atlan from a chat interface mine is a power user so I see all these buttons so if the user is just interested in NLP or chat to your metadata you can only see that you can filter that you can create personas to hide away other features but I'm an admin person so I see all of it right or you can go to the sidebar here chat and ask a question and Atlan will help you get that information so I can go pick an example so this is something that I ran in the last week or so I asked a question about custom metadata related to retention it was automatically able to pull up information it was able to pull up something from fabric and another asset as well but in essence the idea is that users can talk to your metadata using the AI interface so you can ask questions that are in natural language so for example another day I was asking a question how does someone calculate customer acquisition cost that goes looks at all the information we have it figured out that there is a business term called CAC and it figured out there is a formula to calculate that and it gives me the answer and it keeps asking you follow-up questions as I go along it will tell me here is the term do you want to know all assets that are attached to it do you know all dashboards that are calling CAC in it so it helps you have a conversation with your metadata and it expands on that metadata and go on from there hopefully I answered your question if not please come off mute and you can add on awesome any other questions that we have unanswered Ryan and Dan or did we catch all of them in chat we have a question in chat about convincing small and medium business clients for another governance tool like Atlan when Fabric is positioned as SaaS with Purview yeah so I I'm I'm responding to the typing the text response I would like to add I would like to start and I will also like Roy to take to have you take a you know opinion on it because we have a system of truth right there who can add the context there so okay so the way I think about is Varun is the Atlan here is the context platform for your enterprise and if I think about the context the data context in your enterprise is is lies in beyond the context that is already available in Microsoft and Fabric ecosystem right there are yes integrations exist between Purview and Fabric but if you think about the way Atlan is positioning is an enterprise context that is within and beyond the Microsoft ecosystem so that's that's if that's making sense to you then Atlan is the place to go and then the other aspect is deeper integration it's about the connectivity and how that richness in that connectivity itself beyond just the context the demo that Manoj was showing about your conversational search is it ready for the business users the consumption layer of it like who the audience is so these are the two side of vectors I would request you to think about and that no means I'm also saying that oh Atlan is the alternative of purview so that's also I just want to make sure that it's clear because purview goes beyond the context world into the security world so I mentioned no I think you covered all of them for the most part I think the way I would think about it is purview is more about the governance and the security representation of your underlying data estate really kind of knowing what compliance is there what's compliant what's not compliant understanding the security posture of your data that's kind of where it shines I mean yes it has a catalog because otherwise how else are you going to demonstrate you know show your data but the strength is really in the compliance and kind of the governance mapping of your estate I think Atlan on the flip side of that is like yes it has similar capabilities but it's really about understanding the accuracy and the trust of your data and just making sure that end to end your data is being delivered to your end customer the best way possible and there's no issues along the way and if there are issues along the way you're able to address that so I think generally there's similar features but the end goal of the platform is a little bit different right that's kind of the way I would think about it does that make sense awesome Roy appreciate the add-on and Chandru thank you I think I'm just combing through the questions to see if we missed anything I think we have answered pretty much all the questions there maybe Roy that's possibly a question to you but can be posted for later on as well so and by the way I did got Michael's email directly and I'm going to set some time with you next week and if needed we can loop in Roy or someone from Microsoft to also depending upon how much deep you want to go there I'll be happy to do it Manoj can you read the question I'm trying to scan through it spread across two three messages I believe so to my earlier demo on natural language queries to the metadata I think the question was more about on Power BI without going into Atlan can we ask natural language questions within Power BI so there's a copilot that you can use right I don't know the extent of what you're trying to accomplish but there's a copilot experience in there that you can say this is the type of data that I want to produce this kind of output results so yeah you can do that thank you thank you Roy I think some of these deeper discussion might be coming in from a specific use case in mind I think it warrants a specific discussion so happy to double click on a conversation on your business requirements and we can take it from there and there is a nudge from Pablo also like MCP MCP MCP we have a powerful MCP interface within Atlan which can ask any questions of your metadata or do any activities on top of it so you could embed it wherever you want including a plugin but let's do a deeper conversation on the specific use case you are looking at with having said that thank you all for your time Roy thank you so much for having spent your time with us on preparing and delivering this workshop thank you Chandru thank you Dan Ryan Pablo awesome work in answering all these questions and Ayushi in the back end to help get all this together thank you all for your time thank Thanks, everybody. Thank you.
