View Recording: Ship Enterprise Apps From Fabric: A First Look at Rayfin

0:0:10.374 –> 0:0:14.854 Suneer Mehmood Hey everyone, welcome for this webinar. 0:0:16.14 –> 0:0:17.174 Suneer Mehmood where we talk about. 0:0:18.281 –> 0:0:18.721 Suneer Mehmood Bing. 0:0:19.801 –> 0:0:22.81 Suneer Mehmood from vibe to production. 0:0:24.561 –> 0:0:25.281 Suneer Mehmood So. 0:0:26.801 –> 0:0:28.1 Suneer Mehmood We will give it. 0:0:29.41 –> 0:0:31.281 Suneer Mehmood A minute or two for people to just… 0:0:33.361 –> 0:0:35.41 Suneer Mehmood Join in. 0:0:36.961 –> 0:0:55.521 Suneer Mehmood So for those of you who are new to the concept of Rayfin, it’s an open source engine which was recently announced under Microsoft Fabric Apps. And I would say that’s a missing last mile between your data and 0:0:56.641 –> 0:0:57.681 Suneer Mehmood New operations. 0:0:58.561 –> 0:1:2.561 Suneer Mehmood So people are trickling in, so I let them. 0:1:3.841 –> 0:1:8.961 Suneer Mehmood Trickle N, and if I were, you know, we were all physically in a… 0:1:9.921 –> 0:1:14.401 Suneer Mehmood Room, I would have asked for you to maybe raise your hands. 0:1:16.401 –> 0:1:31.121 Suneer Mehmood Like for those who have used who have used AI tool to build something like a maybe a script or an app or a macro in the last six months. Okay, I see hands up. Thank you. 0:1:34.481 –> 0:1:44.81 Suneer Mehmood I see hands up. Yeah, I can, you know, keep your hands up, please. You don’t have to, you don’t have to lower them. Thank you. Because there is a follow-up question. 0:1:49.841 –> 0:1:54.481 Suneer Mehmood All right, so I see at least four hands up. 0:1:58.321 –> 0:2:3.921 Suneer Mehmood All right, so keep your hand up now if you made it. 0:2:5.41 –> 0:2:8.1 Suneer Mehmood into your daily use at work, whatever you build. 0:2:14.641 –> 0:2:21.841 Suneer Mehmood Okay, that’s great. Keep it up now if you have productionized, deployed it to production. 0:2:28.1 –> 0:2:29.601 Suneer Mehmood Well, I see at least. 0:2:30.881 –> 0:2:32.401 Suneer Mehmood Some hands going down. 0:2:33.521 –> 0:2:35.241 Suneer Mehmood At least one hand went down. 0:2:37.481 –> 0:2:45.921 Suneer Mehmood So to be realistic about this, right, there is a gap between what AI helps us build and what actually ships. 0:2:47.121 –> 0:2:47.681 Suneer Mehmood Um… 0:2:48.801 –> 0:3:3.201 Suneer Mehmood You know, that’s going to be the entire subject for the next 45 minutes. At Build 2026, Rayfin was announced. And I think it’s the most practical answer to that gap I’ve seen. 0:3:4.401 –> 0:3:19.841 Suneer Mehmood One promise before we start, whether you write code or have never touched a terminal, you leave knowing exactly what this is and why it matters and what to do about it. Thank you for raising your hands and participating in that survey. 0:3:21.521 –> 0:3:25.761 Suneer Mehmood So before we jump into the content, 0:3:27.921 –> 0:3:40.401 Suneer Mehmood Well, if you’d like to follow us, stay connected with our webinars, please do so. There’s a QR code that you would be able to scan and then follow us on LinkedIn. 0:3:44.961 –> 0:3:57.681 Suneer Mehmood All of our webinars and and summits, technical deep dives, and all are announced in LinkedIn, so please do give us a follow if you would love to stay tuned about it. 0:3:59.761 –> 0:4:17.441 Suneer Mehmood And for those of you who are new to CONSULTANCY, we are a Microsoft solution partner and ServiceNow consulting firm. And we help organization get the most out of their technology investments. So, yeah, feel free to scan them, you know, or if you are in LinkedIn already, search us. 0:4:18.81 –> 0:4:20.721 Suneer Mehmood and give us a follow if we’d like to stay tuned. 0:4:22.321 –> 0:4:25.201 Suneer Mehmood All right, that being said, moving on. 0:4:28.561 –> 0:4:30.881 Suneer Mehmood All right, here’s a roadmap for today. 0:4:33.41 –> 0:4:33.361 Suneer Mehmood All right. 0:4:34.481 –> 0:4:49.121 Suneer Mehmood Five stops and the same journey structure as our AI Summit session. For those of you who have attended that, we’ll start with the problem, concept, mechanics, proof, and action. 0:4:50.961 –> 0:4:52.241 Suneer Mehmood OneNote on that. 0:4:54.961 –> 0:4:58.241 Suneer Mehmood I’m going to say that, you know, loudly so that nobody’s surprised. 0:5:0.401 –> 0:5:7.681 Suneer Mehmood Stops one and two are for everyone. Stop 3 can get a little bit technical for about like roughly say 10 minutes. 0:5:8.641 –> 0:5:27.641 Suneer Mehmood If that’s not your world, watch for the plain English strip at the bottom of every slide. The demo is where everyone converges. Because watching an app appear out of 1 command, it wouldn’t need any translation, I hope. Planning to do some live demos. 0:5:27.761 –> 0:5:28.241 Suneer Mehmood So… 0:5:29.681 –> 0:5:35.281 Suneer Mehmood All right, these are two numbers from the build, and they tell us a story. 0:5:36.561 –> 0:5:44.481 Suneer Mehmood Agent decoding and AI agent writing substantial part of the software grew 76% in a year. 0:5:46.1 –> 0:5:49.921 Suneer Mehmood But 4 out of five AI generated apps never make it to production. 0:5:51.201 –> 0:6:1.841 Suneer Mehmood Not because the code is bad, because there is no database provisioned, no sign in wired up, no security review, no answer for the auditor. 0:6:4.81 –> 0:6:14.321 Suneer Mehmood For, you know, not so technical folks in this meeting, the analogy is like, imagine your team could design and build a car in a weekend. 0:6:15.201 –> 0:6:21.361 Suneer Mehmood But registering it, like insuring it, or you know, getting the license plates takes about nine months. 0:6:22.881 –> 0:6:27.121 Suneer Mehmood That’s Enterprise software right now. Now hold on to that feeling. 0:6:28.241 –> 0:6:35.201 Suneer Mehmood Everything that follows is about to about collapsing that that period, like that nine months, right? So… 0:6:40.81 –> 0:6:59.441 Suneer Mehmood So this slide is lifted almost verbatim from the build session because it’s the most honest framing of the problem I’ve seen from Microsoft. Left side is what a developer or an AI agent acting as one has to assemble before an app is real. 0:7:1.121 –> 0:7:2.81 Suneer Mehmood On the right side. 0:7:3.161 –> 0:7:7.681 Suneer Mehmood What is your IT? What your IT team is accountable for, no matter who build the app? 0:7:9.121 –> 0:7:12.641 Suneer Mehmood Nothing on the right is optional and nothing on the right is fun. 0:7:14.161 –> 0:7:25.41 Suneer Mehmood Every shadow IT horror story your organization has lived through happened because someone did the left column and they kind of overlooked or skipped. 0:7:27.681 –> 0:7:30.561 Suneer Mehmood the right entirely or at least a part of it. 0:7:31.761 –> 0:7:34.81 Suneer Mehmood So, the pitch for the day is… 0:7:35.361 –> 0:7:39.121 Suneer Mehmood What if the right column came for free or you know? 0:7:41.601 –> 0:7:43.281 Suneer Mehmood Yeah, so that’s a pitch for the day. 0:7:47.761 –> 0:7:51.441 Suneer Mehmood A quick recap for anyone who missed our AI Summit session. 0:7:52.961 –> 0:7:59.841 Suneer Mehmood Let me give you a 30-second context. We used a demo retail chain as our example company. 0:8:1.361 –> 0:8:19.601 Suneer Mehmood with daily sales, product stores, etc. And what we call as freezer telemetry, like IoT temperature sensors inside the stored freezers, streaming readings all day so you can spot a freezer drifting into the warm territory before the ice cream melts. 0:8:20.641 –> 0:8:35.681 Suneer Mehmood We took the raw data, cleaned it through the bronze, silver, gold layers you see on the screen, added an ontology. So AI speaks our business language and finished with a data agent. Anyone could question in plain English. 0:8:38.1 –> 0:8:45.201 Suneer Mehmood It landed well, and the most common follow-up question was, like, people asking, like… 0:8:46.321 –> 0:9:6.521 Suneer Mehmood The agent told me which freezers are failing. Now who builds the app that schedules the repair, right? So that’s the last mile. Analytics and agents produce insight, right? Operations run on apps. So until now, those apps lived outside of the governor state we just built. 0:9:6.641 –> 0:9:10.481 Suneer Mehmood built. So keeping that retail data in mind. 0:9:12.561 –> 0:9:26.401 Suneer Mehmood Building apps on that governed model is exactly where this, you know, leads us, right? So we want agents to do actions based on findings, right? So that makes the picture complete. 0:9:30.81 –> 0:9:30.561 Suneer Mehmood So. 0:9:31.401 –> 0:9:32.241 Suneer Mehmood What is refin? 0:9:35.481 –> 0:9:54.201 Suneer Mehmood It’s an open source toolkit, genuinely open, like MIT license, it’s on GitHub today, that lets a developer or an AI agent describe an app’s backend in code, and then deploys that backend inside Microsoft Fabric with one command. I mean, something as… 0:9:54.481 –> 0:9:58.961 Suneer Mehmood Easy as one command. I’m not saying always one command, but like something as easy as one command. 0:10:0.241 –> 0:10:0.721 Suneer Mehmood Um… 0:10:1.921 –> 0:10:8.1 Suneer Mehmood In the Fabric product list, it surfaces as a new item type called Fabric Apps, and that’s currently in preview. 0:10:9.681 –> 0:10:10.241 Suneer Mehmood Um… 0:10:11.841 –> 0:10:24.241 Suneer Mehmood The analogy that I stuck when I explained this to the slightly non-technical people, Fabric is a state. Ifran is like the building permit plus the utilities hook up. 0:10:26.241 –> 0:10:34.481 Suneer Mehmood You don’t negotiate with five utility companies every time you build a house on a governed land. So that’s the whole idea. 0:10:38.721 –> 0:10:40.161 Suneer Mehmood All right, let’s look at this. 0:10:43.521 –> 0:11:0.801 Suneer Mehmood This is a, you know, Microsoft architecture picture, redraw a little bit. The top half is the build loop, an AI agent like GitHub Copilot, Claudio, Replits agent, or a human developer who works with Rayfin SDK. 0:11:1.881 –> 0:11:6.561 Suneer Mehmood which authors the apps back end as ordinary code in your repository. 0:11:7.681 –> 0:11:25.601 Suneer Mehmood The bottom half is what deploying produces, not a random cloud VM, but managed services inside fabric. So it could be a database, a one link storage, real-time capabilities, serverless functions, authentications, et cetera, et cetera. 0:11:26.521 –> 0:11:39.601 Suneer Mehmood So the two halves are the two audiences in this room, like builders, live in the top box. Everyone accountable for governance lives in the bottom one. So Raven is like the arrow between them. 0:11:44.881 –> 0:12:1.601 Suneer Mehmood So this is a preview of the command that you’re going to hopefully see when I show this all live. On the left is a slightly abridged transcript of the deploy command. You will see the real, you know, when we do the live demo, hopefully. 0:12:3.201 –> 0:12:16.481 Suneer Mehmood So you can see it’s pretty much a single command, right? It creates an app as an item in a fabric workspace, generates a database from the code, wires of the auto sign in, deploys the front end, and enhance your URL. 0:12:19.41 –> 0:12:38.441 Suneer Mehmood Now, if we connect this to the opening stat, four and five AI built apps die before production. And what kills them is almost never the code. It’s the plumbing, right? Standing up a database, wiring a sign-in, getting hosting approved, passing a security review. So in a normal enterprise project, 0:12:39.571 –> 0:12:44.451 Suneer Mehmood That plumbing is weeks to months of tickets, meetings, and reviews. 0:12:45.281 –> 0:12:53.481 Suneer Mehmood This one command, well, although in a small scale, did all of it right in about a minute inside the platform you already governed. 0:12:54.561 –> 0:13:8.241 Suneer Mehmood So that’s a headline here, right? So for engineers, yes, the API layer is GraphQL over Data API Builder. And the SQL database comes with a read-only analytics endpoint. So we’ll see that in the demo. 0:13:13.761 –> 0:13:18.81 Suneer Mehmood So this workspace picture is a mock-up. 0:13:19.161 –> 0:13:27.521 Suneer Mehmood Of the example delivery app, right? It’s not deployed, you know, so… 0:13:28.481 –> 0:13:35.41 Suneer Mehmood This in this particular slide, you can see that your this is what pretty much your. 0:13:36.681 –> 0:13:37.681 Suneer Mehmood CISO R. 0:13:38.841 –> 0:13:40.961 Suneer Mehmood Chief Information Security Officer would want to see. 0:13:42.921 –> 0:13:50.481 Suneer Mehmood When A refin app deploys, three items will appear in the workspace. The app itself. 0:13:52.81 –> 0:13:55.521 Suneer Mehmood The database, in this case, it’s a SQL database. 0:13:56.641 –> 0:14:0.481 Suneer Mehmood and SQL Analytics endpoint sitting right next to the reports. 0:14:1.681 –> 0:14:3.201 Suneer Mehmood and semantic models they relate to. 0:14:4.961 –> 0:14:21.441 Suneer Mehmood So it’s like a, think of it like a mock-up. We can see that, you know, live in the demo when I show you. And here you can see that everything is co-located within the fabric workspace. And why does it really matter? Why does co-location matter? 0:14:22.721 –> 0:14:33.521 Suneer Mehmood Because everything you already do to govern that workspace, like roles, permissions, orching, capacity monitoring, 0:14:35.521 –> 0:14:35.921 Suneer Mehmood Like… 0:14:36.761 –> 0:14:41.281 Suneer Mehmood Fabric will cover that, right? Like it is covered within the fabric ecosystem. 0:14:42.561 –> 0:14:52.161 Suneer Mehmood Now let’s contrast that to the usual story. The app could live in some SaaS tool or a personal crowd subscription and IT finds about it during an incident. 0:14:53.241 –> 0:14:57.361 Suneer Mehmood Here, there’s no finding out. The app is born governed, right? 0:15:2.641 –> 0:15:5.441 Suneer Mehmood So let’s look at what an app’s data model looks like. 0:15:6.321 –> 0:15:11.41 Suneer Mehmood It’s a one, it’s a single TypeScript class, and that class is a database table. 0:15:12.241 –> 0:15:14.641 Suneer Mehmood Each field becomes a typed column. 0:15:15.721 –> 0:15:19.521 Suneer Mehmood The many decorator declares a relationship between tables. 0:15:22.881 –> 0:15:29.761 Suneer Mehmood And the part I care most about as an architect that is that role block that you can see over there. 0:15:31.121 –> 0:15:34.561 Suneer Mehmood That’s A role-level security sitting right next to the data definition. 0:15:35.881 –> 0:15:38.961 Suneer Mehmood What does that mean? It means a user only gets… 0:15:40.641 –> 0:15:58.81 Suneer Mehmood the roles that they can see, right? It is enforced by the platform on every request, not by hoping that the front end will filter it correctly, right? It’s a security you can code review rather than a security you can document, right? So 0:16:2.721 –> 0:16:11.761 Suneer Mehmood So that’s pretty much what I wanted to cover. So the security is built in while the app is being built in Fabric. So. 0:16:13.601 –> 0:16:15.761 Suneer Mehmood Governance comes as a part of this. 0:16:18.801 –> 0:16:21.761 Suneer Mehmood All right, so let’s look at what ships inside every template. 0:16:22.881 –> 0:16:27.441 Suneer Mehmood This is a tell that Rafin was designed agent first, not like agent compatible. 0:16:28.881 –> 0:16:34.721 Suneer Mehmood So agents.markdown is documentation written for the AI that will extend the app. 0:16:35.761 –> 0:16:40.321 Suneer Mehmood MCP servers are live tools that let AI inspect the project. 0:16:41.601 –> 0:16:53.201 Suneer Mehmood The scaffold assumes your next developer is an agent and prepares onboarding docs for it, the way we prepare them for a new hire, right? You’ll see both files on screen during the demo. 0:16:56.401 –> 0:17:17.801 Suneer Mehmood So there is a replicate code at the bottom, right? And that owns its place for one reason. It shows a strategy, right? So CEO of a company whose entire business is in AI app building, saying like, agents write the code anywhere, but the data, identity, and governance stay in the customer’s own fabric tenant. So creativity anywhere, custody with you. 0:17:19.41 –> 0:17:24.881 Suneer Mehmood For business leaders, that division of labor is a pattern to demand from every AI tool you adopt. 0:17:34.481 –> 0:17:39.921 Suneer Mehmood All right, so there is also a feature to build an app on your semantic model. 0:17:42.1 –> 0:17:45.281 Suneer Mehmood Most of you, I hope, are familiar with semantic model. 0:17:47.521 –> 0:17:48.801 Suneer Mehmood So it is. 0:17:50.321 –> 0:18:6.241 Suneer Mehmood Think of it like a layer that talks between your report and your backend data storage, where you establish your relationships between your different entities or tables and apply your business rules for calculations. 0:18:8.561 –> 0:18:20.241 Suneer Mehmood Rollups and whatnot and conversion of your technical language to business language, right? So that’s semantic model. RBI users know that. 0:18:23.361 –> 0:18:43.281 Suneer Mehmood So this was built by, you know, the part built by a Power BI team, right? The workflow is like, paste a share link to a semantic model and tell the agent what app do you want. And it builds an app that queries a model live using DAX queries, right? It’s a way of querying a semantic model. 0:18:46.321 –> 0:19:6.201 Suneer Mehmood So the two properties that this enterprise, this makes this enterprise safe rather than anything else, like, you know, the metrics cannot fork. The app reads the same measure definitions as every report. So revenue could mean the same thing everywhere, right? And permissions cannot be bypassed. 0:19:6.721 –> 0:19:9.361 Suneer Mehmood Pro-level security follows the sign-in user, right? So… 0:19:11.841 –> 0:19:16.881 Suneer Mehmood So beyond the templates, apps can also mount data you already govern, like. 0:19:17.961 –> 0:19:23.921 Suneer Mehmood warehouse tables, analytic endpoints, through a connectors mechanism with fabrics permission staying in front. 0:19:24.761 –> 0:19:32.321 Suneer Mehmood Right. So I’ll be providing a handout after this. So details will be there in that handout too. 0:19:34.81 –> 0:19:42.81 Suneer Mehmood For today’s demo, in the interest of time, we’ll be using a todos app to keep it. 0:19:43.721 –> 0:19:48.81 Suneer Mehmood you know, to keep it all within the timeframe that we have. 0:19:52.241 –> 0:19:56.961 Suneer Mehmood But you know, in the handout, you can pretty much see the details to build an app based on the semantic model. 0:20:0.641 –> 0:20:2.961 Suneer Mehmood All right, so let’s go to a live demo. 0:20:5.801 –> 0:20:7.441 Suneer Mehmood Let me keep a track of time. 0:20:8.721 –> 0:20:16.321 Suneer Mehmood If there are questions, please throw them in the chat. I might not be able to see. 0:20:17.601 –> 0:20:19.761 Suneer Mehmood Raised hands, so. 0:20:21.41 –> 0:20:35.681 Suneer Mehmood Amy, who is with us, will help me with those questions. She’ll read out, and then I’ll try to answer that. But let me set up my screen so that I can show you a demo here. 0:20:36.841 –> 0:20:38.81 Suneer Mehmood Bear with me a second. 0:21:6.961 –> 0:21:8.321 Suneer Mehmood Show my screen again. 0:21:17.81 –> 0:21:21.921 Suneer Mehmood Let me know if you can see my screen. Hopefully, you can see my screen again. 0:21:23.641 –> 0:21:24.961 Suneer Mehmood Um, so… 0:21:26.161 –> 0:21:39.281 Suneer Mehmood I have, I was like, I in fact was testing all this, so I have a backup copy if in case I have, we have some glitches while I’m showing this demo. So I’m going to try to do it all fresh for us. 0:21:40.481 –> 0:21:41.521 Suneer Mehmood So, wish me luck. 0:21:43.841 –> 0:22:3.481 Suneer Mehmood So for those of you who are new to Fabric, Fabric, there is a concept called workspaces. Workspaces is like a way where you can organize all your objects, storage objects, your different apps, semantic layers, and stuff, and you can assign a workspace to a specific capacity in Fabric so that. 0:22:3.601 –> 0:22:9.761 Suneer Mehmood you have control over your compute needs. So I’m just going to name this Rayfin. 0:22:10.921 –> 0:22:16.961 Suneer Mehmood You can see that I’m going to use one of our trial capacities. 0:22:18.561 –> 0:22:32.721 Suneer Mehmood Well, not trial capacity, but like, you know, it’s named as trial capacity. So you can see this workspace is pretty much fresh, nothing in here. Had there been anything like it would have shown those items over here. 0:22:33.561 –> 0:22:35.441 Suneer Mehmood I’m going to try to use this. 0:22:39.601 –> 0:22:46.161 Suneer Mehmood This particular workspace, and I’m gonna try to create an app. 0:22:47.361 –> 0:22:48.161 Suneer Mehmood Let me. 0:22:53.201 –> 0:22:53.601 Suneer Mehmood Hey. 0:23:2.321 –> 0:23:6.961 Suneer Mehmood All right, hopefully the font is big enough. 0:23:7.721 –> 0:23:8.641 Suneer Mehmood For us to see. 0:23:14.201 –> 0:23:18.241 Suneer Mehmood So you can see that the MPM create Microsoft refin at latest. 0:23:20.281 –> 0:23:21.841 Suneer Mehmood That’s the single command. 0:23:23.921 –> 0:23:27.281 Suneer Mehmood Once I have installed the APM, sorry, NPM, excuse me. 0:23:29.441 –> 0:23:42.561 Suneer Mehmood So it asks you, hey, how would you like to start? You can use a template, or if you have an external Git template for an app, you can use it, or you can configure manually by starting from scratch. 0:23:44.241 –> 0:23:46.241 Suneer Mehmood So I’m going to just use a template. 0:23:49.401 –> 0:23:50.81 Suneer Mehmood For now. 0:23:52.801 –> 0:23:55.201 Suneer Mehmood I’m going to use the third option, a to-do app. 0:23:57.121 –> 0:23:57.441 Suneer Mehmood Um… 0:23:58.521 –> 0:24:3.121 Suneer Mehmood just to enter our to-do list, right? So let’s see what that does. 0:24:6.641 –> 0:24:7.361 Suneer Mehmood All right, I’m gonna… 0:24:9.441 –> 0:24:11.521 Suneer Mehmood Name it to dos. 0:24:19.441 –> 0:24:20.721 Suneer Mehmood All right, this is live. 0:24:21.801 –> 0:24:25.921 Suneer Mehmood Show or like demo, so you’ll have to bear with me while the app builds. 0:24:31.361 –> 0:24:37.521 Suneer Mehmood Disclaimer that I’m not a full stack software engineer. I’ve vertically, I mean, I’ve been. 0:24:38.881 –> 0:24:47.601 Suneer Mehmood Specializing in the world of data and AI, and it’s so awesome to see these capabilities being brought into. 0:24:48.841 –> 0:24:52.641 Suneer Mehmood Uh, platforms like Fabric, so that, like… 0:24:53.681 –> 0:25:1.201 Suneer Mehmood Even people who have not built a full stack softwares can build applications now. So that’s awesome. 0:25:8.721 –> 0:25:18.481 Suneer Mehmood So now it’s synchronizing with the latest Rayfin scaffolding. It says the project created successfully. I’m going to change directory to that. 0:25:20.721 –> 0:25:21.361 Suneer Mehmood I’m here. 0:25:23.761 –> 0:25:24.321 Suneer Mehmood Gonna. 0:25:26.801 –> 0:25:27.841 Suneer Mehmood Run dev. 0:25:31.921 –> 0:25:38.641 Suneer Mehmood So now it asks us like in a hey, what is that fabric workspace that you want this app to be deployed in? 0:25:40.1 –> 0:25:45.361 Suneer Mehmood Now, I’m going to give the new Rayfin workspace that I’ve just created. 0:25:47.201 –> 0:25:49.201 Suneer Mehmood So that the app gets deployed there. 0:25:59.601 –> 0:26:7.41 Suneer Mehmood You can see that it’s coming coming through, right? And Apple’s built real time. You can see that it also wired up. 0:26:7.921 –> 0:26:15.921 Suneer Mehmood a backend database. I did not have to configure it separately. I did not have to worry about like the wiring and all. So did all of that. 0:26:16.881 –> 0:26:29.521 Suneer Mehmood in the back end. And one beautiful thing about fabric is that whenever fabric creates a storage, like your one link storage or whatnot, 0:26:32.321 –> 0:26:35.41 Suneer Mehmood It creates a SQL analytics at the back end as well. 0:26:37.41 –> 0:26:41.921 Suneer Mehmood I need to see if it is reusing one of the ports that I already have created. Let’s see. 0:26:44.881 –> 0:26:56.561 Suneer Mehmood Okay, so Enterprise sign in is built in. I’m using that. It’s a local host. I’m just showing that in my local machine or the app is. 0:27:0.81 –> 0:27:1.281 Suneer Mehmood So, intrasign in is… 0:27:2.481 –> 0:27:3.441 Suneer Mehmood The way that you… 0:27:6.81 –> 0:27:6.961 Suneer Mehmood Sign into the A. 0:27:8.721 –> 0:27:11.681 Suneer Mehmood Okay, cool. The app is there. It is built. 0:27:13.721 –> 0:27:28.961 Suneer Mehmood Right, so you can pretty much see it’s a to-dos app where you can pretty much add items, you can check them as done, right? There is an action item to open it in Fabric. You can also open that in Fabric, which is awesome. 0:27:32.841 –> 0:27:49.521 Suneer Mehmood So this app is not available yet because it’s a brand new app. I have some old apps that I’ve already opened. I’m not going to confuse you with that. So this will count, you know, I mean, in a while, like give it a couple of minutes and it’ll start showing up there. It’ll be exactly like how it is being shown over here. So. 0:27:51.681 –> 0:28:2.401 Suneer Mehmood All right, so we can actually edit the app using GitHub Copilot chat, but I want to show you the scaffolding, the backend before we go there. 0:28:4.681 –> 0:28:5.761 Suneer Mehmood I’m just gonna say… 0:28:6.721 –> 0:28:9.681 Suneer Mehmood Check here. I’m going to publish this app. 0:28:14.481 –> 0:28:15.361 Suneer Mehmood Just say. 0:28:16.801 –> 0:28:17.681 Suneer Mehmood And BX. 0:28:19.361 –> 0:28:20.81 Suneer Mehmood Open up. 0:28:21.361 –> 0:28:22.921 Suneer Mehmood So that it publishes the front end. 0:28:28.441 –> 0:28:41.521 Suneer Mehmood Once the app is published, what you get is a shareable URL, right? So what I’m showing this app is in my, as you can see, it’s in localhost with a port of 5174. 0:28:44.81 –> 0:28:52.1 Suneer Mehmood Once the app is published, it says your app is live at this particular URL, right? So you copy that URL if you want to share that with someone. 0:28:53.361 –> 0:28:56.721 Suneer Mehmood Gonna try to see if I can see that app through the shareable URL. 0:28:59.281 –> 0:29:18.761 Suneer Mehmood See how fast and how easy it is, right? So that’s my point here. With all the governance within the Fabrics ecosystem, right? Think about your object level security, authentications, and rule level securities that you might have built in the back end. 0:29:18.841 –> 0:29:39.641 Suneer Mehmood right? So it’s the exact same app that I was showing you in my local machine over here with the same status. So you can see that like all the first three to-do list has been checked as done. And I’m going to check because I just published the refund, you know, the front end to a shareable URL as well. So I’m going to just. 0:29:42.321 –> 0:29:50.641 Suneer Mehmood Check this. Now, before we go to anything else, I want to show you how the backend looks like, right? 0:29:55.801 –> 0:30:6.961 Suneer Mehmood Going here, go to Fabric. I was telling you, like, you know, this is the app, right? This is the Fabric workspace, and you can see that the app is there. 0:30:9.521 –> 0:30:11.761 Suneer Mehmood Ann, you can also see that in the… 0:30:13.361 –> 0:30:23.361 Suneer Mehmood like the services that is available within the fabric list, right? When you add a new item, if you go ahead and add a new item, this will be shown as an application. 0:30:26.481 –> 0:30:30.881 Suneer Mehmood And you can see that the backend database is configured. Let’s go ahead and look at the data. 0:30:32.321 –> 0:30:33.201 Suneer Mehmood Now, open it. 0:30:38.401 –> 0:30:45.361 Suneer Mehmood Right, it created a backend database called todos going to DVO schema, so it created two. 0:30:45.681 –> 0:30:45.841 Suneer Mehmood Um… 0:30:48.201 –> 0:30:50.961 Suneer Mehmood Two tables, right? Like, and so it came as a part of the template. 0:30:56.801 –> 0:31:4.1 Suneer Mehmood You can see the four to-do items that I had, right, in the app. 0:31:5.201 –> 0:31:6.161 Suneer Mehmood So… 0:31:9.281 –> 0:31:12.481 Suneer Mehmood I had checked this somehow the local copy had not refreshed. 0:31:18.561 –> 0:31:19.761 Suneer Mehmood I refresh it, oops. 0:31:22.1 –> 0:31:32.721 Suneer Mehmood That’s a demo failure. I don’t know what happened there, but I’m gonna close it, but we can refer to this external app over here, and… 0:31:34.321 –> 0:31:35.121 Suneer Mehmood If I… 0:31:36.561 –> 0:31:39.681 Suneer Mehmood Refresh this and open it back up. 0:31:41.921 –> 0:31:50.481 Suneer Mehmood Says this app isn’t available somehow; it’s a little bit slower than, you know, you know, it was when I tried it. I’m gonna add an item like… 0:31:51.841 –> 0:31:53.361 Suneer Mehmood Best line item. 0:31:55.161 –> 0:31:58.321 Suneer Mehmood Let’s see. Okay, it got added. 0:31:59.761 –> 0:32:0.961 Suneer Mehmood Now, let’s go. 0:32:2.81 –> 0:32:13.41 Suneer Mehmood Look at the back end, right? So this is the back end. It had all those four items. I’m going to run this again so that the data gets refreshed. Well, you can see I’m going to order by. 0:32:14.641 –> 0:32:15.841 Suneer Mehmood Bye bye, created that. 0:32:19.161 –> 0:32:19.681 Suneer Mehmood That. 0:32:24.121 –> 0:32:31.281 Suneer Mehmood You can see the last item that I just added, that slide item, and it’s not completed, right? Because I just had not. 0:32:34.841 –> 0:32:41.601 Suneer Mehmood Where is it? Here, OK. I had not checked. That’s right. I’m gonna check it, take it again. 0:32:43.121 –> 0:33:2.81 Suneer Mehmood Going back, gonna close this window so that we are not confused. Run it again. Now that false should turn to true, right? So think of this like your transactional database, right? You have your application running and your data getting registered in the backend. So this is pretty much your transaction database. 0:33:4.241 –> 0:33:21.281 Suneer Mehmood Going back to these are my trials so that I was getting ready for our demo over here. Now, I’m going to open up a SQL Analytics endpoint as well, and some of you might ask, hey, why do you 0:33:21.921 –> 0:33:23.121 Suneer Mehmood I’m gonna close these. 0:33:25.521 –> 0:33:46.41 Suneer Mehmood Why do you need a SQL analytic endpoint? So think of it like it’s like your read-only schema, right, for your analytics, so that you can actually have separation of concern between your transaction database and your analytical needs, right? And what I would like to call out is that 0:33:47.761 –> 0:34:4.161 Suneer Mehmood The sync between the backend transaction database and the SQL Analytics endpoint is also managed by Fabric. So there’s no, you could see that I just built it all in one command, and there’s no ETL pipeline that refreshes the data between the backend and the SQL Analytics endpoint. So 0:34:6.201 –> 0:34:25.201 Suneer Mehmood There you go, like you know, so in a regular analytic scenario, you would want to do that. Like you would want to build an ETL pipeline for your analytics or warehousing needs. Again, doing a create that or a buy. You can see that all items are there. The last item that I just created is also there. 0:34:25.841 –> 0:34:28.81 Suneer Mehmood in the analytics endpoint. 0:34:29.481 –> 0:34:42.721 Suneer Mehmood Now, the good part is, in a SQL analytics endpoint, you can also create a semantic model if you wish, right? So, for those of you who have done Power BI, you can see that, you know, you know, like what a semantic model is. 0:34:45.841 –> 0:34:51.521 Suneer Mehmood Enterprise semantic model name, so gonna say refund. 0:34:53.121 –> 0:34:54.321 Suneer Mehmood Toodles. 0:34:57.521 –> 0:34:58.801 Suneer Mehmood I think all. 0:35:1.201 –> 0:35:2.321 Suneer Mehmood All right, confirming. 0:35:6.401 –> 0:35:9.921 Suneer Mehmood So this will help you build analytics, your Power BI dashboards on top of it. 0:35:30.641 –> 0:35:31.441 Suneer Mehmood Come on, Fabric. 0:35:32.881 –> 0:35:44.561 Suneer Mehmood All right, we can see that both the tables that I selected are available right now. I’m just going to say, hey, user ID is user ID. Just going to join them. 0:35:46.881 –> 0:35:50.1 Suneer Mehmood And cardinality is many to one. 0:35:51.641 –> 0:35:53.281 Suneer Mehmood Assume referential integrity. 0:35:54.881 –> 0:35:55.841 Suneer Mehmood Save it for now. 0:35:59.681 –> 0:36:1.41 Suneer Mehmood That, that, right? 0:36:2.161 –> 0:36:3.201 Suneer Mehmood Let’s build a new report. 0:36:4.841 –> 0:36:6.1 Suneer Mehmood But we are power bearable. 0:36:10.361 –> 0:36:13.121 Suneer Mehmood You can see my Tulus table. 0:36:14.401 –> 0:36:16.481 Suneer Mehmood Remember, it all came from… 0:36:17.681 –> 0:36:18.561 Suneer Mehmood Our app. 0:36:20.561 –> 0:36:23.441 Suneer Mehmood close this. Is that my old? 0:36:24.881 –> 0:36:25.761 Suneer Mehmood Forms. 0:36:27.921 –> 0:36:29.521 Suneer Mehmood Doll came from. 0:36:30.721 –> 0:36:32.561 Suneer Mehmood Our tools app, OK. 0:36:33.441 –> 0:36:34.961 Suneer Mehmood So, going back. 0:36:36.881 –> 0:36:37.761 Suneer Mehmood To. 0:36:41.441 –> 0:36:44.641 Suneer Mehmood Screen’s open, think I should close. 0:36:46.1 –> 0:36:52.1 Suneer Mehmood I see, I see a new report, what I did. Ohh, it’s opening another tab. OK, my bad. 0:36:55.121 –> 0:37:4.641 Suneer Mehmood So I’m going to use this table to-dos and I’m going to just use a tabular visualization for now. I’m going to just minimize the screen so that we have more real estate. 0:37:7.521 –> 0:37:8.321 Suneer Mehmood And then… 0:37:10.81 –> 0:37:11.761 Suneer Mehmood Ohh, I just want. 0:37:12.961 –> 0:37:13.801 Suneer Mehmood Aydingul. 0:37:15.681 –> 0:37:16.721 Suneer Mehmood We do that. 0:37:18.161 –> 0:37:19.121 Suneer Mehmood And the status. 0:37:26.161 –> 0:37:27.521 Suneer Mehmood Like this, right? 0:37:30.401 –> 0:37:32.161 Suneer Mehmood Not gonna worry about any. 0:37:33.561 –> 0:37:35.121 Suneer Mehmood Beautifications for now. 0:37:38.881 –> 0:37:41.361 Suneer Mehmood values so that we can see them. 0:37:43.641 –> 0:37:44.241 Suneer Mehmood Hopefully. 0:37:45.761 –> 0:37:46.161 Suneer Mehmood Yeah. 0:37:50.241 –> 0:37:52.1 Suneer Mehmood Not gonna worry about any, any. 0:37:54.721 –> 0:38:8.641 Suneer Mehmood Any painting effects, any beautification, just making sure that this is readable in the shared screen. So you can pretty much see if I were to… 0:38:10.321 –> 0:38:14.161 Suneer Mehmood Sort this and the last line item that I added, right? 0:38:15.921 –> 0:38:17.681 Suneer Mehmood Let’s go ahead, back to the app. 0:38:22.441 –> 0:38:23.201 Suneer Mehmood Another item. 0:38:26.241 –> 0:38:29.201 Suneer Mehmood Just adding it. All right, let’s come back. 0:38:30.681 –> 0:38:35.841 Suneer Mehmood What is this? Is this an analytical endpoint? Let’s first check the transaction database first. 0:38:38.161 –> 0:38:57.41 Suneer Mehmood Okay, another item is added and it’s not checked. You can see that. There will be a little bit of delay between SQL analytic endpoint and the transaction database. I want to call that out and as I expected, you know, because there is fabric works things that analytic endpoint based on a schedule. So that’s an. 0:38:57.761 –> 0:39:16.721 Suneer Mehmood schedule with it. So once that is refreshed, all we need to do is like refresh the semantic layer and pretty much the data will be available. So any changes that you’re making to your transactional database gets into the SQL analytics endpoint and. 0:39:16.761 –> 0:39:37.241 Suneer Mehmood Can get refreshed by the semantic layer, and you have analytics right there, right? And, as I was saying, like, and you can flip this, like, and you can use semantic layer as your data input to your application as well, so that, like, you know, if there are curated data sets in your warehouse or something and you, you’re considering that. 0:39:37.361 –> 0:39:45.841 Suneer Mehmood as a single source of truth for your analytics that can pretty much be used as, you know, the source for your backend. 0:39:48.241 –> 0:39:49.601 Suneer Mehmood No, I’m gonna… 0:39:51.121 –> 0:39:52.241 Suneer Mehmood Try one more. 0:39:54.481 –> 0:39:55.441 Suneer Mehmood I have this. 0:39:56.401 –> 0:39:58.1 Suneer Mehmood VS code over here. 0:40:2.881 –> 0:40:3.521 Suneer Mehmood Ann. 0:40:5.201 –> 0:40:5.681 Suneer Mehmood Well. 0:40:14.641 –> 0:40:15.121 Suneer Mehmood This one. 0:40:19.921 –> 0:40:27.41 Suneer Mehmood So I, this is the structure of these files. You saw that I just created that todos app, right? 0:40:28.441 –> 0:40:41.201 Suneer Mehmood This is a scaffolding. This is a structure of everything that Rafin created in the back end. So it’s a full-fledged app, right? Now that it is there, you can actually create it. 0:40:43.321 –> 0:40:43.521 Suneer Mehmood Um… 0:40:44.721 –> 0:40:47.121 Suneer Mehmood you can actually ask enhancements. 0:40:48.721 –> 0:40:55.761 Suneer Mehmood using Copilot, agent mode, identic mode, right? So let me see, I had a prompt here. 0:40:57.681 –> 0:41:4.161 Suneer Mehmood I’m going to try, okay? These are all trials. So wish me luck. 0:41:5.601 –> 0:41:15.121 Suneer Mehmood So the prompt that I’m adding to Copilot is like, add an optional priority field like low, medium, high to the todos. 0:41:16.161 –> 0:41:35.521 Suneer Mehmood Make it optional on the entity so existing rows remain valid. We don’t want to make it mandatory because if it is mandatory, God knows what happens to the ones that, you know, because that will become invalid because they wouldn’t have any data unless we see it with some default data. And I’m asking you to show it as a colored chip. 0:41:35.681 –> 0:41:54.881 Suneer Mehmood in the list and add a filter for the high priority items, right? Update every place that validates or creates to do so adding a task still works. So I’m asking to be like cognizant of the entire code repository here. And I’m particularly asking not to change the author role policies. 0:41:55.441 –> 0:41:57.281 Suneer Mehmood This is again, like this is. 0:42:0.81 –> 0:42:2.881 Suneer Mehmood Having access to a prompt and and and a good. 0:42:4.801 –> 0:42:25.81 Suneer Mehmood model does not mean that it will always turn out to be right. You have to make sure that the prompt is explicit enough and a good software engineer or an architect will be able to, it’s always valuable so that you know they can ask, I mean they can prompt it properly and think about any loopholes and try to. 0:42:26.481 –> 0:42:44.161 Suneer Mehmood explicitly ask that in the prompt. It’s asking GPD, I mean, it’s using GPD 5.4 mini right now. My earlier trial was with Opus 5.0, Clouds Opus 5.0. Let’s see. 0:42:44.241 –> 0:42:45.441 Suneer Mehmood if one does it better. 0:42:50.1 –> 0:42:57.201 Suneer Mehmood So yeah, it’s reading through. It’s reading through the entire file and trying to make those changes. 0:42:59.201 –> 0:43:2.721 Suneer Mehmood We’ll have to give it some time because last time I did it. 0:43:3.761 –> 0:43:7.841 Suneer Mehmood It took at least a good couple of minutes or more. 0:43:9.681 –> 0:43:11.41 Suneer Mehmood So, yeah. 0:43:17.521 –> 0:43:21.681 Suneer Mehmood Meanwhile, let’s go back and keep our app ready over here. 0:43:24.481 –> 0:43:24.921 Suneer Mehmood The the. 0:43:30.881 –> 0:43:33.921 Suneer Mehmood app and I’m going to keep this. 0:43:35.41 –> 0:43:38.321 Suneer Mehmood Terminal also ready over here. I don’t need this. 0:43:40.641 –> 0:43:43.121 Suneer Mehmood slightly decreasing the form over here. 0:43:50.321 –> 0:43:53.41 Suneer Mehmood So, let let let it do the cooking, I would say. 0:43:54.681 –> 0:44:0.321 Suneer Mehmood because it’s a model making all of those changes. So the point I’m trying to make is… 0:44:1.201 –> 0:44:1.601 Suneer Mehmood Look. 0:44:3.761 –> 0:44:6.321 Suneer Mehmood It was one command and it built an app. 0:44:7.201 –> 0:44:22.401 Suneer Mehmood From A pre-existing template, it could have very well built it from any other template that you already have, and it wired up all the connections from the app to the back end under one single governance within Fabric. 0:44:23.201 –> 0:44:42.161 Suneer Mehmood I just built a semantic layer, established the connections for Power BI, and I built, you know, a bare-bone Power BI dashboard, and every time the data refreshes in the app, exactly like how an ETL would. 0:44:43.521 –> 0:45:3.201 Suneer Mehmood be able to see it, you know, it sees the data, right? So this is now, I was talking about SQL Analytics endpoint taking some time to actually show this. All right, I’m going to allow this on the Copilot side. So you can see that now the SQL Analytics endpoint is also refreshed with the additional items that I added. 0:45:4.81 –> 0:45:5.41 Suneer Mehmood Throw away. 0:45:7.681 –> 0:45:8.321 Suneer Mehmood Say hello. 0:45:9.761 –> 0:45:10.161 Suneer Mehmood Um… 0:45:12.521 –> 0:45:16.81 Suneer Mehmood Let’s, in the meanwhile, see our… 0:45:21.441 –> 0:45:23.361 Suneer Mehmood Or, we can refresh this. 0:45:26.681 –> 0:45:31.761 Suneer Mehmood I’m going to go here, back, there’s a refin semantic model, refresh that. 0:45:33.961 –> 0:45:40.241 Suneer Mehmood Now the semantic model is getting refreshed. Hopefully my Power BI shows that data. 0:45:43.201 –> 0:45:44.481 Suneer Mehmood There’s a test line item. 0:45:45.921 –> 0:45:48.161 Suneer Mehmood Yeah, it’s refreshing. Yes, there you go. 0:45:50.481 –> 0:46:11.521 Suneer Mehmood It just added this line item, but that is complete is gone false. So think of it like, you know, think of it right, like it’s a transactional data, your application wrote data to your back end without an ETL, your analytics endpoint got refreshed by Fabric itself. You didn’t have to run any pipeline, you did not have to ingest anything. 0:46:13.761 –> 0:46:23.761 Suneer Mehmood you know, your once your semantic layer is refreshed, which you can like, you know, trigger using periodic schedule or something like that. 0:46:25.361 –> 0:46:33.521 Suneer Mehmood Power BI refreshes right here. There is analytics on top of the data as well. Now, Copilot agentic mode says it’s all done. 0:46:34.801 –> 0:46:34.841 Suneer Mehmood Uh… 0:46:36.801 –> 0:46:37.761 Suneer Mehmood Which is good. 0:46:40.321 –> 0:46:42.961 Suneer Mehmood Let’s try this, right? So… 0:46:47.121 –> 0:46:48.721 Suneer Mehmood I will have to. 0:46:50.921 –> 0:46:51.201 Suneer Mehmood On. 0:46:52.481 –> 0:46:52.801 Suneer Mehmood And. 0:46:59.641 –> 0:47:2.321 Suneer Mehmood I’m going to deploy that changes. Let’s see. 0:47:18.561 –> 0:47:25.761 Suneer Mehmood All right, so let’s go ahead and look, go check our app. Gonna do a hard refresh. 0:47:27.201 –> 0:47:28.81 Suneer Mehmood On the screen. 0:47:31.681 –> 0:47:32.561 Suneer Mehmood Let it load. 0:47:35.361 –> 0:47:37.681 Suneer Mehmood Did you see that a new priority? 0:47:39.601 –> 0:47:41.121 Suneer Mehmood Label is added right now. 0:47:42.561 –> 0:47:49.361 Suneer Mehmood This is what I we prompted, right? So, hey, add a new priority label, right? Let’s see. 0:47:50.881 –> 0:47:53.121 Suneer Mehmood I’m gonna add you. 0:47:53.961 –> 0:47:59.201 Suneer Mehmood priority label. So I just did an application change using Agent KI now. 0:48:0.81 –> 0:48:2.321 Suneer Mehmood I’m going to set the priority to low. 0:48:3.481 –> 0:48:4.401 Suneer Mehmood and try to add it. 0:48:6.81 –> 0:48:11.121 Suneer Mehmood There you go. In the prompt, if you remember the prompt. 0:48:12.321 –> 0:48:33.441 Suneer Mehmood I had like, add an optional priority field, low, medium, high, make it optional, update every place, and then show it as a colored chip in the list, right? So that’s what it is doing, low is green. So again, large language model, right? Kind of displaying that, although it was not explicit in my instructions. 0:48:33.681 –> 0:48:37.921 Suneer Mehmood Right, testing medium line item. 0:48:39.521 –> 0:48:42.1 Suneer Mehmood Here, I’m going to set the priority to medium. 0:48:43.121 –> 0:48:44.721 Suneer Mehmood It has a yellow color chip. 0:48:45.881 –> 0:48:46.401 Suneer Mehmood Testing. 0:48:48.41 –> 0:48:48.801 Suneer Mehmood Pi pri. 0:48:50.881 –> 0:48:51.601 Suneer Mehmood Line item. 0:48:53.201 –> 0:49:13.841 Suneer Mehmood Setting it as a high, adding it. There you go. Green, yellow, and red, all added. I just changed my app using agentic AI using Copilot. Again, to summarize, the app was built with one command. It did all those scaffolding, all the 0:49:14.161 –> 0:49:33.401 Suneer Mehmood wiring all the back end provision, all the analytics endpoint in the fabric ecosystem where it is well governed, I could go ahead and build a semantic layer and pop BI on top of it. And then I came and thought like, hey, let me add something to the app and that is done, right? So. 0:49:34.401 –> 0:49:53.841 Suneer Mehmood Just refreshing, not my analytics endpoint here, the transaction table, because that’s the first point of entry. You can pretty much see that. Let me expand this a little bit. All the line items that I just added, new priority label, medium line item, and high priority line items, where I have not checked anything to be done, is all added. 0:49:54.1 –> 0:49:57.201 Suneer Mehmood with their respective time stamps, right? Could be. 0:49:59.401 –> 0:50:1.121 Suneer Mehmood you know, universal time or whatnot. 0:50:2.401 –> 0:50:20.881 Suneer Mehmood Analytics endpoint, as I said, could take some internal refresh time. Oh, it’s refreshed. That’s awesome. So we timed it in such a way that, you know, it got refreshed quick as well. Now, if the semantic layer gets refreshed, we’ll be able to see that data. 0:50:24.961 –> 0:50:28.161 Suneer Mehmood I’m not going to do that, but you get the idea, right? 0:50:29.841 –> 0:50:30.241 Suneer Mehmood Well… 0:50:31.481 –> 0:50:33.1 Suneer Mehmood Where is my report? 0:50:35.281 –> 0:50:35.921 Suneer Mehmood Ohh. 0:50:37.361 –> 0:50:39.281 Suneer Mehmood But the data will be, well, let’s do it. 0:50:55.281 –> 0:50:57.521 Suneer Mehmood Let it get refreshed. 0:50:59.761 –> 0:51:0.561 Suneer Mehmood Ann. 0:51:2.401 –> 0:51:5.761 Suneer Mehmood It’s done now, bring it back. 0:51:8.1 –> 0:51:9.361 Suneer Mehmood Do I need to refresh it again? 0:51:13.521 –> 0:51:16.641 Suneer Mehmood Yeah, just going to save file, save. 0:51:26.641 –> 0:51:28.721 Suneer Mehmood Just saved my report. 0:51:51.521 –> 0:51:52.161 Suneer Mehmood Come on. 0:51:59.441 –> 0:52:1.841 Suneer Mehmood Well, the expectation is that the report schedule refreshed. 0:52:12.401 –> 0:52:32.161 Suneer Mehmood All right, in the interest of time, I’m not going to worry about it. But you get the idea. You saw the dashboard getting refreshed earlier. So I’m going to switch back to my slide to cover any points, because a large part of what I wanted to cover in this demonstration was like, hey, let’s see the Ifran app working. 0:52:33.41 –> 0:52:50.401 Suneer Mehmood Let’s see the scaffolding, what we do and stuff. And this demonstration was just limited to building a Tulu’s app. So, but you get the idea, right? Like, you know, we can get creative about this. You know, we can just go ahead and build pretty much anything. 0:52:51.921 –> 0:52:59.961 Suneer Mehmood That we want right, um, with these features, so yeah, like you know, you saw the live demo, not gonna uh. 0:53:3.281 –> 0:53:3.361 Suneer Mehmood Uh… 0:53:6.81 –> 0:53:24.321 Suneer Mehmood You know, you saw that, like, you know, this is like, hey, let’s time. Like, you get an idea how long it took to actually build the app and how long it took to build a backend scaffolding, how long it took to update the data and such. I’m not going to read through all of that. It was just, you all saw it live, happy that it all worked out. 0:53:26.481 –> 0:53:48.1 Suneer Mehmood And you saw, you know, what the database is, you know, the backends, the analytics loop and such. One thing I would call out is like the SQL database can be altered in the backend. So, and I would caution us, never do that because you know, the source of truth is always the code. 0:53:49.281 –> 0:54:8.481 Suneer Mehmood And if you want to do any structures, always do that through the code base and not, you know, changing the back end. Because if you change the back end, it’ll be like much, much like, you know, you change the back end of an application without letting the application’s code base know about it. So the code could break. 0:54:8.561 –> 0:54:29.361 Suneer Mehmood Or the code will probably be unaware of that changes, so I would recommend not doing any indirect database changes. Always make the code changes to the application. I wanted to pick a live feature, but I think we covered the concept there, like by adding. 0:54:30.401 –> 0:54:48.641 Suneer Mehmood a button through Copilot. It would have been interesting if we had some time, but I’m gonna skip through that for now. And yeah, so you get the idea, right? We are asking the Copilot agent, we saw the code changing, and then we need to… 0:54:48.721 –> 0:55:9.121 Suneer Mehmood deploy the changed code by running NPX, Rafin up, we can pretty much see those files getting changed and stuff in real time and whatever we have asked the agent AI to build, hopefully, always test it out, always test it out, right? Sometimes it’s all, again, at the end of the day, it’s large language models, non-deterministic models. 0:55:9.601 –> 0:55:30.721 Suneer Mehmood So they could make mistakes, make hallucinations, they could overlook certain things. So always test your app out. Again, this entire feature is in preview, but it’s a very promising feature, right? Like the greatest part is that the template itself ships, you know, the agents dot. 0:55:31.681 –> 0:55:36.321 Suneer Mehmood Markdown file and the MCP servers for other agents to agentic. 0:55:38.321 –> 0:55:39.201 Suneer Mehmood Tools to work. 0:55:40.721 –> 0:55:45.841 Suneer Mehmood So if we zoom in all the way out, just keeping a track of time here. 0:55:47.641 –> 0:56:8.401 Suneer Mehmood This, this one slide version of the last two, you know, last two talks, for those of you who have attended our AI summit, this is the pipeline we all know, right? Sources through the data gets sourced through the bronze, silver, gold medallion structure to the semantic model and ontology, where meaning lifts from that same governed core. 0:56:8.721 –> 0:56:29.361 Suneer Mehmood two branches grow, right? One is the agent’s answer, which was covered as a part of our AI summit, and then the Application Act, right? That’s what you pretty much watch today. We are building and deploying an app within the fabric ecosystem, either the app, you know, writing the data to our relational database. 0:56:29.841 –> 0:56:39.121 Suneer Mehmood or using an analytics one lake data source as your applications backend, right? And… 0:56:41.601 –> 0:56:56.161 Suneer Mehmood So the arrow at the bottom part, you know, I find it genuinely elegant, right? The apps write their data back into OneLake, where it’s instantly analyzable and agent readable. The insights produces actions, right? Actions produce data. 0:56:58.561 –> 0:57:13.521 Suneer Mehmood Data produces insight, so that covers the loop, right? So, it’s not a slogan you watched it happen, right? So, yeah, there is, this is a honest read, right? I all, you know, I’m very enthusiastic about this. 0:57:15.921 –> 0:57:34.881 Suneer Mehmood Some honest read about what you know about all this, right? It’s a preview, and I don’t think Microsoft has announced a GA date yet, so the capabilities and pricing can change under you. The region list is short, so let’s check your fabric capacity region before promising anyone anything. 0:57:35.681 –> 0:57:45.121 Suneer Mehmood Sign in is Entra only, so nothing customer facing yet. And the CLI cannot log in as a service principle, which makes real CICD automation clumsy for now. 0:57:46.241 –> 0:57:47.121 Suneer Mehmood So. 0:57:49.201 –> 0:58:2.1 Suneer Mehmood So the schema is code owned, as I already covered that part. You know, the apps table structure is generated from the code, so please don’t do any back-end changes directly. So… 0:58:4.401 –> 0:58:14.321 Suneer Mehmood And these apps are excellent at what most internal apps actually are, like forms, trackers, consoles, approvals, and whatnot, right? Read rows, write rows. 0:58:16.561 –> 0:58:35.401 Suneer Mehmood I wouldn’t recommend this for heavy transaction processing yet. You know, think about an auto system that must update every, I mean, inventory, payment, shipping as one all or nothing operation, right? Like thousands of times a minute. This app is could be could not be the right tool. 0:58:35.601 –> 0:58:48.481 Suneer Mehmood Right, and at least it’s not there yet. Pick a single app shaped problems, right? So now this disqualifies a pilot, all of it disqualifies like something like. 0:58:49.681 –> 0:59:3.921 Suneer Mehmood rebuilding an auto system, right? Let’s go ahead and rebuild with this. That’s not what you should think of. Think about small applications that you can build. And again, in the pilot mode. 0:59:5.921 –> 0:59:26.721 Suneer Mehmood So if you lead a business unit, you have a backlog of small apps you never staffed, like trackers, request forms, and consoles, because none of them justified A three-month project. The cost of those just, you know, just fell to days, right, and not months, right? And because governance comes built in, saying yes to them. 0:59:27.41 –> 0:59:47.681 Suneer Mehmood It’s not risky anymore. And if you’re a data or a BI person, you are also a winner in this part, right? Like the semantic models that you’ve been curating over time, it’s not just a few of the reports, right? They become the foundation people build applications on. And apps built this way cannot fork your metrics definitions, right? So single source of truth concept. 0:59:49.601 –> 1:0:7.681 Suneer Mehmood So if you are an IT or security, people will build apps with AI, whether you sanction it or not, right? So this is the version where it happens inside your perimeter, your identity system, your permissions, your audit trial, your security, and such. 1:0:10.561 –> 1:0:11.281 Suneer Mehmood You know. 1:0:14.401 –> 1:0:18.641 Suneer Mehmood Yeah, like, you know, think of it like it’s in preview. It’s like, so do not promise your customers anything. 1:0:21.201 –> 1:0:22.161 Suneer Mehmood So. 1:0:23.841 –> 1:0:43.1 Suneer Mehmood That being said, yeah, there are a few five things to remember, right? Like, it removed the bottleneck, right? Writing software, stop being the expensive part, you know, governing it. It’s built within Fabric now, right? Riffen is the last mile, which actually… 1:0:43.601 –> 1:1:2.961 Suneer Mehmood Makes do operations right on top of the data. Governance is built in with the app. Data is state is the mode. Everything you watch today stood on the semantic model and ontology. Well, not ontology, we didn’t, you know, that was part of the AI Summit, in fact. So, governance. 1:1:3.201 –> 1:1:25.41 Suneer Mehmood comes built in as a part of this fabric ecosystem is the biggest mode that I can call. It’s in preview. You can start the pilot this month and there’s a one contained workload named owner. Production can wait. You know, you can, but learning does not have to. That’s all I wanted to share. This is the last slide. So. 1:1:26.561 –> 1:1:44.81 Suneer Mehmood You’ll get a survey, so please feel free to fill out that survey so that, you know, we can help improve. I mean, that’ll help us improve our, you know, this marketing demonstrations on our side. So, yeah, like, you know. 1:1:45.361 –> 1:2:4.401 Suneer Mehmood We, I mean, I have prepared a handout. I’ll be sending that out, and and these are certain URLs, like if you want to know more about refund, you can go to aks.ms. There is a community on GitHub, so there is also an announcement channel, you know, which… 1:2:5.561 –> 1:2:25.681 Suneer Mehmood you know, was has been there from June 2026. So yeah, that’s all I wanted to share. I hope the live demo was helpful and this presentation was helpful. Yeah, I know that we ran a couple of minutes over. Thank you so much for watching. Thank you so much. So if you get time, please. 1:2:26.401 –> 1:2:27.361 Suneer Mehmood Share your feedback.