Microsoft Build Fabric – Ep. 322

Mike joins from an early morning at Microsoft Build and shares what he’s seeing on the floor—especially how Copilot is showing up across experiences and what’s new/important for Fabric and Power BI practitioners.

News & Announcements

• Microsoft Fabric May 2024 Update | Microsoft Fabric Blog | Microsoft Fabric — Welcome to the May 2024 update. Here are a few, select highlights of the many we have for Fabric. You can now ask Copilot questions about data in your model, Model Explorer and authoring calculation groups in Power…
• PowerBI.tips Podcast — Subscribe and listen to the Explicit Measures podcast episodes and related content.
• Power BI Theme Generator — Create and download Power BI report themes using the PowerBI.tips theme generator.

Main Discussion

This episode is a Build conference recap with a Fabric/Power BI lens.

Key points from the conversation:

• Build conference impressions: what’s different about Build (developer-heavy) and the kinds of questions people are asking.
• Copilot everywhere: the team talks about AI being woven into more product surfaces and what that means for day-to-day workflows.
• DAX Query View: why a more code-centric modeling experience is useful, and how it changes the Copilot/demo story for modelers.
• Fabric updates worth watching: the discussion calls out items like task flows and real-time analytics as meaningful “workflow” changes.
• How to keep up: a practical approach to tracking announcements without chasing every feature.

Looking Forward

Pick one Build/Fabric announcement that directly impacts your current work (e.g., task flows or a modeling improvement), test it in a small sandbox, then decide whether it belongs in your production standards.

Episode Transcript

0:02 good morning and welcome back to the explicit measures podcast with Tommy Seth and Mike good morning everyone from a very early morning Seattle how’s everybody doing man not as good as good as you I’m up we’re moving it’s going to be a good day I’ve got a I’ve got two Booth sessions that I’m doing today I’m helping out at the Microsoft fabric Booth so I’ll be answering questions and doing things with people and just talking about architectures and answering questions around what’s new inside powerbi when things are coming out so that will be fun I’m looking

0:33 out so that will be fun I’m looking forward to it I get really energized at conferences shocker I like people so yeah it’s been it’s been super fun it’s the the awesome major difference between I extroverts and introverts yeah people people energize me I get all hyped up are you excited are you excited are you more excited at the end of a day by interacting with so many people or are you just exhausted if talked to that many people yeah I I’m a selective extrovert

1:04 people yeah I I’m a selective extrovert then because it depends who I’m talking to or about what right if I’m talking about if it’s if it’s things that I’m passionate about yeah all day long no problem I’m there I’m talking about it how how our between our previous podcast where you were out there and there and hended what I haven’t seen any shorts Mike I thought I thought we were GNA have like little candid candid mic yeah they don’t turn out very good when you’re holding your own

1:34 you’re holding your own phone so just giving you grief dude I never no it’s fine I’ve actually seen so build is build is interesting I’ve never been to build before it’s my first time at build it’s been a really good conference the there’s a lot of really like there’s a lot of production to it there’s like a whole stage they got this huge like when when we’ve been to fabric conferences yeah there’s there’s like a stage and there’s like a camera or two hidden around but like everything feels a lot more compact there’s there’s a lot of like booths right next to the main stage a lot of vendors are supporting and a lot of

2:05 vendors are supporting and a lot of developers are here so I’m talking to a lot of different companies now you lot of different companies now talking to Miro is here one of the know talking to Miro is here one of the companies that that is at the conference which I absolutely love Meo I use Mir all the time for whiteboarding and putting things down and working with clients I the class that I run the training class that I run the data in governance class for parbi we use miror all the time in all of our presentations and there was a ton of things that they were showing me like I didn’t even know existed so I was like this is great I came by their Booth they taught me these new things people have been very

2:36 new things people have been very friendly and a lot of a lot of the people have been talking to so I I’m I’m Manning a booth area so I’m actually at the there’s like a a did and AI Channel they have like Cosmos DB SQL Server MySQL they have realtime analytics now so there’s like a a whole data platform area and I’m been representing the fabric and the powerbi side of that so a lot of people have been coming up in questions and I’ve actually had a number of really good conversations around embedded a lot of

3:06 conversations around embedded a lot of customers are developer conference right maybe maybe they’re exploring some things I don’t know but so the questions for me feel feel like they’ve been split 50/50 between how do I do in bettered for my organization what does that look like and how can I do it securely and then the other half of the questions is like okay everything’s co-pilot show me some co-pilot inside powerbi like what what is co-pilot look like inside the context of desktop so I’ve been doing a lot of demos around co-pilot one of the

3:37 demos around co-pilot one of the things that I don’t think anyone’s really seen it’s not been announced at build but it’s been around for a while is the Dax query view I show people this there is that people are very much so if you think about co-pilot inside the context of parbi you have like co-pilot for the report author hey how do I make pages are insightful things make me a visual something more towards the visual there’s like a co-pilot experience it’s more like report

4:07 experience it’s more like report authoring experienced yeah and then you you flip over to Dax Cory viw and now you have a more code Centric a builder a modeler experience of co-pilot and there’s and the demo I have is actually pretty good I just pull up one of my star models that I have and it has a little prompt for co-pilot suggest measures and it suggests three measures that you might want in your model it writes out all the dacks for you and it it follows the same naming Convention as my other measures in my model nice

4:39 as my other measures in my model nice which is really cool so I have like a lot of my measures are some of the column name right some of sales some of cost of goods sum of units sold and then it was going in and said well you might want averages on things and then it would do like average price per product and so it was doing some other divisions and adding not super complex tax but tax but some simpler ones and then I was doing a demo around going to Performance analyzer and then you can drop a query directly down to Dax query

5:09 drop a query directly down to Dax query view so I have a visual it’s running and then I immediately take the query pull it down in the Dax query View and you explain the query oh my word it gave me like three paragraphs of explanation of like what is this Dax doing and it literally it caught it said it looks like this Dex is from an autogenerated tool it looks this is being it’s passive aggressive tag tagged myself in this like I wrote this for you I wrote this for you yes yeah but it was really cool so it did a

5:40 yeah but it was really cool so it did a good job of that and then what it did is it gave you like a summary of things and then it went in how you have when you you write a fact quer review or sorry when you get a performance letter it has like DSO core it has like DSO filtered DSO one dso2 like so all the filter contacts that it builds for the Dax it builds in desktop well it goes through each section DSO one DSO 4 dso2 like with all the numbers that it has and it describes why is this STX here here’s here’s what this Dax is

6:10 STX here here’s here’s what this Dax is doing and then it went through and it found all the functions and gave you this function does this summarize columns is this function here’s what it’s doing keep values here’s what this function is and what it’s doing in the formula it was pretty impressive it’s it’s definitely like it’s not it’s it’s not going to make you a Marco or Alberto but it definitely provides a lot of technical support where you need it like this stuff doesn’t work help me figure out this formula say what a great learning

6:41 this formula say what a great learning tool I yes exactly well and it brings a lot of there’s other references you could do like Define this function or what does this function mean or like you can pick out additional things that that you can talk to the co-pilot but that’s I think really the value is it there’s a whole bunch of searchable internet or or information about these like you’re just trying to get to An Answer quickly of like what’s going on here something’s not quite right and it’s giving you a lot of good context around that so anyways that’s what I’ve been doing and then sadly I haven’t

7:12 been doing and then sadly I haven’t been able to make a lot of sessions because I’ve been so excited about all the training and helping and and talking to people so I’ve been doing a lot of networking but I haven’t been intending a lot of and there’s actually to be frankly honest there’s not a ton of fabric sessions it’s it’s a lot more around like General like AI huge this year it feels like build is tons of AI it’s large language models it’s doing rag with large language models it’s how to deploy models like it it there’s really a big

7:42 models like it it there’s really a big sense around co-pilot is in everything and every experience is going to be easier with AI and and I think the bigger announcement here is say like Ai and PCs right so they’re doing this whole like we’re going to bring direct AI things and run them on your computer without having to go back out to the cloud so anyways yeah pretty interesting stuff I I did watch the main Kino they talked about that and that’s when I yelled hey I need a new computer right

8:14 computer right well think about it the most one of the more expensive things that Microsoft has to run in the cloud is the AI it’s it’s a it’s a graphics processor unit it’s an expensive piece of Hardware and it and it is is so if you have a ton of people hitting AI all the time the models themselves they’re not they’re not building net new models I me they’re building them but they’re like deployed or whatever right

8:44 they’re like deployed or whatever right but they need some Hardware to run this stuff on and so you do the training you build the model and the model just gets served into a computer and the graphics processor just choose and all your questions get done with this I I feel like there’s going to be needing to be a lot of work around making that execution of the model as efficient as possible and it’s going to get so efficient that and this is why I think PCS are going to be coming a very big thing here is they’re going to want to push those down and have the the AI basically living on

9:15 and have the the AI basically living on your machine and being able to serve the information directly from there this is going to be I think the new movement I guess because now you don’t have all that cost of AI sitting in a server in Microsoft space you’ve already purchased the computer you’re just going to use it locally to keep turnning through it yeah and be cheaper for them to run it they talked about Mo’s law but now they’re calling this the scaling law where they we’re like almost not restarting with PC and computation but to your

9:46 with PC and computation but to your point who thought graphics cards were what really runs AI so I think they’re realizing if we try to scale this on a cloud we would bankrupt everyone so we’re like let’s get everyone with the competition anyways I know that’s a little off topic that gets me EXC excited but I wanted to be trying to get a new a new computer after after you just bought one like what two months ago right but weren’t you already trying to do this

10:09 weren’t you already trying to do this Tommy right like load up AI under the computer but it’s slower but the like I said the fact that they’re building this intertwined already I already upgraded my graphics card because of this but the fact that you can intertwine this like they’re building the PC specifically to run llms or local models or or do a lot of the things that Lally slow down AI like the pieces are meant for that which is is phenomenal yeah I feel like it will maybe go both places right it’ll it’ll get more efficient and efficient enough that I can run my local PC and then

10:39 that I can run my local PC and then maybe it’ll get so efficient that they then pull it back in the cloud and they just give it to you as a service again like it’ll probably just like yes juggle back and forth between these different things to compete there so like when we a white paper about powerbi here’s what you need for your power how to run powerbi in your computer now like that’s the cloud speaking of so we can we can go through so the so so just to be clear in build there is a a session from Amir and Arun going through all the the book of

11:09 through all the the book of announcements updates and things current new things that are coming out features coming out for just fabric so I believe if you are not currently attending build you can sign up for free online you can watch the keynote online for free so the the topic I guess for today is going to be going through more of the announcement and there’s a blog post that’s also in the in the the description of the video where you can actually go check out and see the actual announcement I’ll put that in the chat window as well

11:39 I’ll put that in the chat window as well but there’s there’s an announcement of the May 21st landscape of fabric at this point so that’s what the main topics we’re going to go through there is touching on some of the new features there anything that stood out to you guys on from the the main blog what what things come out or stood out to you or even things that you’re already seeing inside Fabric in your tenant that are already standing up to you well I think the two biggest ones and we’ll see which direction you want

12:09 and we’ll see which direction you want to go is without you doing anything we now have task flows which are probably the biggest Game Changer in the last year I I’m gonna I I would go out and say and then the real-time analytics that’s now on your left hand pain two mics saying takes up some real estate some precious real estate those to me are the two ones if you were living under a rock and just opened up powerbi you would you would notice so task flows was announced at fabric conference so it’s not it’s not a

12:40 fabric conference so it’s not it’s not a new thing but we haven’t been able to play with it really and so literally like a week or so before build task flows gets launched and so we have these things and we actually I don’t think we’ve actually talked about task flows on the podcast yet so maybe we should just start with quickly touching on some task flow pieces new features out do you guys like it do you think the functionality is right what do you think about the task flow maybe we Define what it is a task flow is this UI it’s

13:10 is a task flow is this UI it’s basically allowing you to have blocks of information that attach to each other through basically the powerbi workspace environment so think of it like I’m I have pipelines I have potentially lake houses I potentially have semantic models and reports you can now group those things into what you would almost use as a diagram on a whiteboard right if you’re explaining to someone hey I’m going to build this thing in in Fabric and I’m going to draw out the Whiteboard diagram this is what you would use to

13:41 diagram this is what you would use to you draw this right yeah ingestion over here store data here serve data there and that’s like the the mentality there and so I think what this is doing is giving you a lot more flexibility to at a high level you’re going to have a lot of artifacts that was think I was quickly at one of my and Tommy’s observations of doing learn our learn fabric series on YouTube within two or three episodes we had gobs and gobs and gobs of of assets and resources

14:12 gobs and gobs of of assets and resources and things all over the place so what we should do Tommy I’m G to I’m going to tag you on this one then we need to go back to our learn fabric workspace and with all the things in there we should then reorganize them into task flows so we have it already organized out for this new like we it in organiz let’s learn fabric we’ll call it let’s clean fabric let’s clean fabric let’s organize fabric yeah let’s organize fabric well no Mike to your point this is lineage View 2. 0 this is I need to visualize all

14:45 View 2. 0 this is I need to visualize all these artifacts I have in a workspace which a lot of you already know can get very cluttered and actually organize it both by lineage like this starts first and this is dependent on this tag it and really to your point organize it in a visual mapping view which not just makes sense is is The Logical way to look at all these artifacts I I would say it’s not just it’s not a better view it’s probably should be the default view on how we’re trying to do all these different artifacts and engines and runs and if you haven’t tried it out yet it

15:17 and if you haven’t tried it out yet it should be the preferred way to look at all of your artifacts if you’re doing any runtimes in fabric one of the coolest things I like about it is is the ability to click on a card in the task flow and it immediately filters everything down below to only the items I care about so that by itself is just an incredibly helpful feature and then also again with this is again people people won’t do it but they need to be like we need to we need to build muscle memory to do this but each card can have its own

15:48 but each card can have its own description you can actually describe why does this thing exist what are we doing with it because I have found myself when I build Medallion architectures inside a workspace I’m trying to add numbers in the front of of the different things to sequence them together right I may have one pipeline that runs a lot of other pipelines well I name them a certain way in order to get the order of the items in the list to show up correctly so previously you might do like initial

16:18 previously you might do like initial load pipeline all these other things you don’t technically need to do that you’re you’re wasting valuable name space to try and Orient the things in the in the pipeline correctly so you can at least see the list of objects correctly I think this simplifies a lot of that as well that’s interesting it doesn’t all need to be in fabric though right and you like the the nice thing is obviously in each one of those tasks the

16:48 in each one of those tasks the objects that are within the workspace Etc are all part of grouped by that but you can still like if you have an external Source right that’s not part of it you can still name it and still use this as the documentation or the area where you have those other things coming into this flow prior to it it being self-contained within fabric right and I guess the value there would be if you’re not directly just building straight in fabric right away or it’s a migrated

17:18 fabric right away or it’s a migrated effort or something that you want to get kick the tires on and just see like this this to me speaks to that flexibility that that’s in there already yeah I I also you bring up a good point there Seth and I think something else that’s very relevant with this as well is there is this workloads or not workloads it’s what’s it called it’s pre-built task flows I guess templat templates templates okay templates what I’m going for there are templates of

17:49 I’m going for there are templates of this that make this even easier for you to start with something so like hey I’ve heard this thing called Medallion architecture let’s start there and literally builds the task flow with no objects in it and then describes here’s the objects you’d want to fill each of these cards with and you can delete them you can move them around you can like rearrange them but the neat part for me was there’s there’s here are examples of other architectures that are already existing start here I think that’s also very helpful because people are overwhelmed on how to build

18:19 people are overwhelmed on how to build things so when I saw the template that posed the question immediately of course I think of you guys now all the time probably more than I need to but I saw the templates and POS a a question I want to pose to you guys that’s in an amazing ramp up time for someone who’s maybe not familiar with Medallion architecture someone who’s not a data engineer maybe they’re an analytics maybe they’re a business user maybe they’ve heard about this and they can automatically start with a medallion architecture or some very very

18:50 medallion architecture or some very very common patterns around data engineering is this really going to ramp up the learning curve for nonata engineers no Mike Mike shaking his head like like n bobblehead I I I agree and I disagree right like I think I think what’s helpful about it is I don’t I don’t think it’s like a learning tool but I think it is a helper tool from the standpoint that it’s like you can take these templates and you can

19:21 you can take these templates and you can see what the default setting or default steps in that process should be right so for brand new folks to a particular area or you’re exploring a little like a few of these pre-selected outputs there’s value in like oh hey go select this thing here’s all the steps you’re going to need to do and then it’s just a matter of like plugging in the the parts of that so I I I yes it’s kind the parts of that so I I I yes it’s helpful but I wouldn’t say it’s like of helpful but I wouldn’t say it’s like this massive learning tool okay I yeah I

19:51 this massive learning tool okay I yeah I think there’s more of a like there’s some methodology there’s some more education there’s some like you do you do need to read up on Medan architecture to to like figure out what that looks like so I I do I would want to Echo you probably should do some more research yes hey here’s some training here’s some documentation hey here’s the learn. Microsoft that talks about what is

20:17 Microsoft that talks about what is Medallion architecture right there should be some pairing between this so here’s the here’s the go fast button and and then here’s links to training and things that enhance that even further I like that because I think it’s I think the biggest hurdle a lot of people have especially with a lot of the news products is access to what’s right and I think yeah this is a good first step however to your point this is not like hey build me a star schema I’ll get some coffee and it’s done there’s obviously the the learning but I think just the

20:48 the the learning but I think just the access access is is pretty neat that they’re doing that I I really like that little little nice touch that they’re doing on the on the task flows D we have a ton of other things to cover I’m I’m looking through through all the blogs and notes like Tas was not even a build announcement no mean they touched about it on the keyot Note but it’s not it’s you’re right it’s not an actual like new thing that came from build yeah let’s let’s go on a little bit further with some other items here so I’m going to pick

21:18 other items here so I’m going to pick on one here just really quick this is probably not in any particular order I’m just picking up things that I was thinking about here the new addition of Iceberg as a format for being able to be used inside fabric which is really interesting this seems like a very direct play towards snowflake snowflake is out snowflake uses the iceberg format to store their data so now fabric both absorbs data bricks and snowflake data

21:49 absorbs data bricks and snowflake data sets because the the delta or the format of their files is either Iceberg or Delta Delta formats so that’s really interesting I I agree I would play on that too I think that they also have Now shortcuts for on premise data sources so it’s like this whole game plan that they’re having on like no we when we said we’re going to be this Central Storage or Central area for all your data we mean it so yeah oh I would be REM Tommy that’s a good one too mirroring is a

22:21 that’s a good one too mirroring is a very hot topic mirroring is like really so I’m I was like okay mirroring fine like I I don’t do a lot of mirroring I’m not sure if I want it I don’t know but man people are really interested I had a lot of questions around really what is mirroring where can I get it how do I turn it on so and and there’s a really big push from Microsoft they’re giving you mirroring for free like the compute used to do the mirroring they’re just going to do

22:51 mirroring they’re just going to do they’re going to pay for it the storage that’s going to be used for miroring there are limits there’s thresholds but you’re getting free storage to collect that information and mirroring this is very strategic and I think to me wherever that data lives I mirroring so remember how I said maybe this in the podcast before look Microsoft make it easy there’s common workloads I know I need to do over and over and over again yeah one of them is

23:21 over and over again yeah one of them is an activity like mirroring I just want that production data in the Raw tables as they live in my lake house just make it happen it happen mirroring is that feature to some degree so I’ve got I’ve got to take a so for me one who’s been micro batching and loading in large chunks of data I really need to seriously take a look at mirroring and making sure I understand like where do I put it and aply understand why what’s the value here to me as the business and what data

23:51 me as the business and what data sources can of connect to this could solve a lot of simple ETL engineering patterns that I just don’t care about managing I I just want the current record showing up in my tables period Well I think that’s the big shift too I think there’s this whole big strategy if you were to look at Microsoft big whiteboard of fabric on their game plan this is which I would love to see it’s pretty big whiteboard I have a pretty big one but I’m assuming

24:21 have a pretty big one but I’m assuming they have a lot of marker and marks on it but I think a big point of this Mike is with mirroring with what they’re doing with iceberg what they’re doing with shortcuts with on Prem too when we said we thought on Prem was dead it’s no no no Fabric’s not just all the products it’s not just we centralize all your products which by itself is amazing it’s the fact that no you have one place like one Lake I think Amir said this one lake is the heartbeat of fabric and I that you’re seeing as

24:51 that you’re seeing as like the underlying wave of all these other features that they’re releasing everything else is cool but would you just take one by itself even if it didn’t have all the other bells and whistles I whistles this is this is a pattern that I mean this is this is a pattern that I think is think is occurring I don’t know I in the last five six years I’ve been using data bricks to do a data engineering like it’s this idea of separation of storage and compute it’s it’s using big data Technologies to do asset

25:21 Technologies to do asset transactions like th this feels like we’ve we’ve seriously moved away from the days of I need to land all my data into a SQL engine SQL computer SQL Server right then to render it to powerbi it makes no sense I had a such Seth and I argued about this a lot when it first came out I I I was like Seth there’s no why do I need to keep a SQL server on for the entire month and pay for it to run when all I really need to do is load data for like

25:52 really need to do is load data for like 30 minutes to an hour every morning and it stays cached in my analysis Services models so like to me this there’s like a whole story here that I was like I don’t need all this extra weight of all the systems I just I just want to store it and then I want to read it so just make that easy for me and I was very annoyed that there was no Auto easy to use system to do this and this now feels like fabric this feels like one Lake it’s this easy to use store things down pay for what I use and not

26:23 pay for what I use and not actually have to consume all this computer the time just to get data into reports it does change the questions that we ask in terms of you if I’m dealing with a few projects right now or a few projects I had in the past it’s like okay how am I going to push this to a SQL database like what are the different steps now it’s like well how do I just push it the fabric I have shortcuts I have mirroring I can get this done in a week to to be fair like Microsoft is making it much easier like all the tools that Microsoft’s building data flows python notebooks or spark

26:53 flows python notebooks or spark notebooks the pipelines the fact that every single compute engine is talking to the lake and reading and writing data to the lake that’s that’s huge because that was that was what was missing before because you could do it but you did need a bit of knowledge and code and figuring things out like it wasn’t as smooth and easy as it is right now it it’s literally a click of a button pick the thing here’s the destination name the table boom done I’m I’m off to the races so the fact that they’ve really

27:23 races so the fact that they’ve really simplified that experience I think makes this a lot more accessible to a lot more the organization yeah and I think your last point is what what really hammers at home right like fabric is is making things easier so that we can focus on the most important thing that we’re trying to do which is get data to a place that we can analyze and use it right make it meaningful and it’s removing a lot of those barriers of just the structure of things in the infrastructure right that we have to build to move the data to a position

27:54 build to move the data to a position where we can start to analyze and use it so I I love that there this is all very I think strategic of course it in order I hope so right and I’m gonna give you I’m gonna give you Captain Obvious I’m gonna give you the big game I’m GNA give you the game the big game plan here the big game plan here is you can’t do AI on any data until the data is one place so so like the whole yeah this this is this is

28:25 the whole yeah this this is this is really feeling like this is the we one lake is that’s what his the heart right it’s the central place it has to be the single plot spot where you put all the information because from there then we can do all the other things co-pilot doesn’t get good at doing data things unless you have all the tables there described with metadata it it the AI won’t know what your columns mean but your AI will be able to better understand what columns you have when you actually add descriptions to them

28:57 you actually add descriptions to them and you describe what information is in this table and how is it valuable to your business if you put that into your models and into table descriptions that enhances this whole experience and so this brings me to like my next Point co-pilot was just announced that it’s now generally available so copilot INF fabric is now generally available that was one of the the announcements and it is now on by default in your eligible tenant so if you if your tenant is eligible meaning you have a a P1 or f64 or higher those

29:31 you have a a P1 or f64 or higher those capacities will have co-pilot turned on by by default I don’t really love it when things switch to turn on by defaults there is a Blog article about co-pilot being on by default and you can go change your admin settings to turn it off if you want to or have certain people use it or not but be aware your co-pilots will start appearing in your fabric enable capacities I’m pretty sure you’re going to get a lot more messaging around trying to get that to work work in your in your tenants yeah so we we these These are so far

30:03 so we we these These are so far already big big topics big potential changes within organization super helpful I’m going to bring it down to things that I’m going to implement right away in some of the the report updates so one of the biggest gripes from not only like internal but external customers has always always been around matrixes

30:28 always always been around matrixes tables of information like just I just can’t tweak it enough I can’t fix it I can’t so the the new updates to The Matrix layouts and selections I guarantee you are hitting the next Sprint in my team like as far as we’re going to revisit a whole lot of Matrix visuals for customers and reports because it it is a instant win when it comes to like a a a rolled out change that was just announced so I’m super

31:00 that was just announced so I’m super stoked about it yes it’s just a report visualization but that’s where like there’s a practical like thing I can I can deploy immediately out of all this and I’m super I’m super stoked about it yeah and yeah the big things from the Matrix layout that talking about is they’re just further aligning with Excel everyone wants an Excel layout Microsoft’s like fine here in powerbi your Excel layouts I think now you can pick the outline form form Taver form compact form all in powerbi which is and

31:32 compact form all in powerbi which is and a lot of the other enablement like turn show a blank row or not so really aligning with how people want to view tabular data or want to view data in a table form which no I agree Seth there’s a lot of things that I’m like all right this this one this one this one I’m going back to so it’s a so so one the the the announcement previous does pique my interest because all three of us had and especially Mike you and I met Miguel a long time ago well before

32:02 met Miguel a long time ago well before he was on the Microsoft team and was already doing some pretty astounding things with powerbi visualization yeah there’s a comment that’s made in the modern modern tool tips and I don’t know if it was directly out of his LinkedIn post but or or part of the announcement but it it says it’s modern tool tips are are changing so it’s just like a a white box now with some some text in it and they’ made this change because it’s the first step in a series of remarkable

32:34 the first step in a series of remarkable improvements these future developments promis to revolutionize tool tip management and customization offering possibilities that were previously only imaginable now for the for the normal person right like that would read this be like okay this is just hype marketing I don’t think I don’t think so I think this is I think this is Miguel behind the scenes blowing people’s minds and them using like this this type of

33:04 and them using like this this type of anac because like you don’t read a lot of that in Microsoft announcements right like hey it’s the next B thing blah blah it’s big but they don’t use revolutionize they don’t use like these are big words for a tool tip what I’m saying so it’s like I’m I I am super excited about the changes incoming and and that one in particular but only because I’ve seen what he’s done in the past and if it is anything like that

33:34 the past and if it is anything like that this this is this would be an area that I I I I think would significantly change I think some some reporting features and how people implemented things if it it starts to come to fruition the way I’ve seen things in the past from from him yeah he’s he even throws even so what you’re I think you’re referring to Seth is I put the link in the chat window so there’s a link on the Microsoft blog talking about the new tool tips and then from there there’s a link to a LinkedIn post that Miguel is

34:04 link to a LinkedIn post that Miguel is written and then in there there are some screenshots looking at the screenshot under the summary section is what he’s talking about there there’s a there’s a whole bunch of really he’s already showing like a different tool tip on a line that has all these different categories with details and information in there and you can see as the cursor is highlighting things even just the screenshot the teaser that he throws on there is is really interesting and it’s really going to be able to enhance this a lot so the the whole he’s the right person to be on the

34:37 whole he’s the right person to be on the visuals building team it’s just Bar None it’s so good yeah he’s I’m very excited about that one I’m gon I’m gonna pull up here a bit I’m gonna go you were going down to the some of the detail Seth so I I want to I want to pull up here I think Tommy’s gonna like this one I think this is is but that’s okay I know well if if we were giving out so every year around Christmas we give everyone gifts and this feels like an early Christmas

35:07 this feels like an early Christmas gift to Tommy so they are thrilled to announce a new AI capability called AI skills and so I’m still trying to W I’m still trying to wrap so I’m I’m wrapping my head around like what is AI skills how do I use it there’s actually a really good video that’s on on the blog post and it walks you through like okay here’s a thing called an AI skill and it gives you a guidance walks you through how that experience will work you have a data model you can pick which tables you care

35:38 model you can pick which tables you care about you can start asking questions of things and you’re it’s like an AI Builder experience a little bit and then you can utilize that to then give people insights or information about your data so what I’m a little fuzzy on still is like how do I implement it and what’s the value ad for people that are going to be using these AI skills but it it feels like we’ve had a lot of gripes with data and AI previously it feels like this is an attempt to start really trying to

36:09 an attempt to start really trying to provide some supplemental report author or data modeler information to an AI model that will help you skill it up faster and give better results more accurate results to what people really want and it’s great timing too by Microsoft because open AI literally announced last week that their new model is going to be be able to take any CSV data and analyze it for you do basically the same thing but no to the from a Microsoft Focus I’m curious too Mike because a huge Focus

36:40 curious too Mike because a huge Focus here is co-pilot co-pilot not just in the fabric spe but co-pilot from a Microsoft point of view looking at your business data the co-pilot Studio the AI skill part I’m I’m in the same boat as you incredibly incredibly intrigued and yeah I think more incredibly excited because the AI skills look to be building your Q& A but really hyperfocused almost like what we we asked for with the automated insights can I wait my tables can I wait my

37:13 can I wait my tables can I wait my columns and we never had that ability so I can actually choose the tables so kind I can actually choose the tables so a little background with the AI of a little background with the AI skills which again you’re gonna see a lot of content for me on this stuff a ton because I’m writing this why this is why I knew you’d like this one yeah the AI skills I can choose tables in a Lakehouse and I can selectively choose the columns and tables and I can basically start a prompt and start saying hey here what’s my biggest who’s my best sales rep well that’s not the result I want so let me modify what the

37:44 result I want so let me modify what the results are so it’s like a very like it’s like the kid version of training your AI model of training AI model it’s the kitty version but yes yes and no but but it’s it’s so the the part that was missing before was this context there’s additional context like so I ask a question it responds with something well it may not be all the things that I need and they have this ability to they have notes to the model so they have this other area that is like additional notes

38:15 other area that is like additional notes you send into the model so like look hey when I say customer this this this and this customer is first name last name I want to see their address I want to see their total sales from this table right you give it more context and it’s becoming Smarter on how the response is and okay cool but what’s really neat is it’s writing a SQL statement it’s it’s not only it’s not just and this is I think something that’s going to be tricky to to negotiate is if you use AI against a

38:45 negotiate is if you use AI against a data set how do it’s right how do what information it’s providing to you is accurate right what which customer has the highest amount of sales it may just present to you a list okay what good is that so what it does though is it writes a sequel statement based on the question and then you can I would presume I I haven’t seen it yet but I would presume you could go modify that or you could go tweak it there’s literally sequel there and I think to myself okay this is an attempt

39:18 think to myself okay this is an attempt to justify the answer with an actual code snippet that shows you how it’s going to get the data and return it and to me this is really cool like the fact that it’s doing that code generation and producing you the results in a nice easyto read format this I think could be much more impactful because now you’re going to not only learn how to to do simple statements talk to the model but you’re going to get more complex SQL statements out and you’re going to learn SQL too so the this is this is great because the business users who don’t know all this

39:49 business users who don’t know all this stuff yet or just getting into this they’re now going to start exposure to a more Rich query language and now when we go over to the SQL analytics endpoint no problem we’ve already been exposed to all these SQL statements I’ll start copying them over and start utilizing them in SQL endpoints this is yeah I think I think this part of this feature is much bigger than they made it to be in in the blog I’m very excited to see where this is gonna go have you built a custom GPT yet or tried building one I have not okay so

40:21 or tried building one I have not okay so it it’s very I know my Lane Tommy and that is not my Lane so well this this goes back to my V diagram of analytics skills and I think again a big one is going to be this prop thing so if you in open AI you can actually build

40:33 open AI you can actually build your own little thing but it’s the same workflow I’m telling hey GPT no I don’t want you to work focus on this focus on this the bigger difference is I’m choosing my columns but Mike not just isn’t showing you the sequel to verify what you’re saying which again I think we did an example in one of our episodes you’re like hey your number 16 you’re like I did the math and you’re not right but actually sequel but more importantly like I’m going to be able to basically produce these skills to my

41:03 basically produce these skills to my business to say hey I have a sales rep GP co-pilot or I have a I have a sales rep skill this is going to be in anytime this data is in a report or in in this particular lake house this skill will be available so the fact that I can train it into your point I think it’s the sequel is huge for us as the analytics side but from the business user that we’re being able to actually train this there’s so many capabilities here yeah I what’s what’s very interesting to

41:36 yeah I what’s what’s very interesting to me is Mike made a comment earlier about how we were we we’ve talked about how how does AI run well well all the data has got to be in the same place right so it’s this is one Lake it’s it’s fabric I think this I think this feature is very telling and I think encapsulates the story of what what needs to happen for AI to work and for the really big differentiator of what fabric provides that other others are not right now right and and the

42:06 are not right now right and and the reason is how do like this feature is basically saying here’s here’s an AI capability that you can get some incremental help with right like you’re you’re asking for a singular experience you wanted to explain something you can refine that right and then and there’s a little bit of value there or and or you can create this this object that is valuable to the business and then in the second paragraph as it’s

42:37 then in the second paragraph as it’s describing like whoa like what does the future look like well in the future you’ll be able to take these and then leverage them for other co-pilots right so all this is is to to Tommy’s point right like you’re utilizing your own organization to train the models right so these are all of the steps of we have the data we’re refining it I’m getting incremental value so there’s value for me to go do this but ultimately what are you doing you’re

43:07 ultimately what are you doing you’re you’re creating the the biggest challenge that everybody wants to create which is the ultimate chatbot for your business right right yeah and guess what you can’t just do it like it takes a lot of work to get there but incrementally if you think about if everybody’s using fabric right if everybody’s doing these things wow you can train things a lot faster than just a singular team can’t you yes you can exactly yeah and my last point I’ll say I think Microsoft if anyone’s going to realize that you have

43:37 anyone’s going to realize that you have to guide AI to say hey you can take a left you can go to any stores in the left you can’t go right and realizing that if it’s going to work you have to guide it I’m super excited about that guys I I’m surprised you guys the de developer nerds out there we haven’t talked about the framework here which I feel like to me is maybe the biggest announcement here or one of the I think largest impact announcements is the fabric framework that’s now available for isvs yeah yeah anyone well

44:11 available for isvs yeah yeah anyone well so this is interesting so there’s there’s a new large feature and this has actually got a full session in the the build group so again I was working the booth didn’t get a chance to go to it but I’m going I was actually talking to the PM who’s building this Aviv is the gentleman who was working on this feature and he said you might want say well explain it a bit more why would I want to use this so this is this is the idea of embedding external applications directly into your fabric so say you’re ezri and you want to do some Advanced mapping things right ezri

44:42 some Advanced mapping things right ezri is a great map but it’s a little tiny visual on a report page well ezri has a whole Suite of things you can do with mapping and if you have a subscription to it you can now you can basically full window that application so what’s happening to to your point Seth earlier about about fabric is the place for data well once you have the place for data what do you need to bring to it all the applications that can do things to the data do you want to go interpret images do you want to do some audio file processing well there’s Services already

45:13 processing well there’s Services already out there that do this there’s now going to be a Marketplace for independent seller people to come in and build their own stuff and I thought to myself this sounds like a great place for the theme generator to live I’ve got to figure out how to make work this is a good idea and and to the team they were like what are you talking about theme generator and then I had to show and they’re like you can do all this this is amazing so but but but it’s it’s the idea of like you can build

45:44 it’s it’s the idea of like you can build third-party tooling and bring it into Fabric and so okay look I have an exp like now you get a an area where I’m going to go grab a lakeh house and talk to its Delta tables all that can just happen you now just need to build an integrator between your application and connecting to the data like ready to go everything’s already loaded so instead of instead of bringing my data to the applic this has been the challenge for years right when

46:14 been the challenge for years right when I go to I go to Salesforce why do I go to Salesforce and have S Salesforce show up and and have to send all my data there well bring Salesforce to to my one link maybe that’s what I want to do so I think I think technically the name of this is called workloads I think it’s actually the name there of what Microsoft is doing and they’re working with Partners today to build out some applications and things that are going to be built into the into the program so I see a lot of potential here I see this is going to

46:45 of potential here I see this is going to be an opportunity that’s going to be really interesting I’ll I’ll put the link here in the chat window so people can go see the public announcement of workloads this is you’re you’re right Tommy this is I think a pretty major announcement as far as things that are going to change my world this one probably will not directly impact me but it’s going to give me a lot more opportunity to see other programs and figure out what do I need to pay for and bring other services into powerbi fabric right and and to the point I think as we always say words

47:16 point I think as we always say words have meaning they’re calling all the other products in fabric now workload synapse data engineering data Factory they’re calling them workloads so basically these are now in a elevated additional workloads that that you’re leveraging right yeah as part of your data ecosystem that’s yep so even even if you are a company let’s imagine you already are a company that has yeah well and I think there’s I think there’s also opportunity here I don’t know if this is isv workloads but imagine you are

47:48 imagine you are so more and more are conversations coming up around I am a company that makes data I I’m a data company I have things that are data related I need to give it to my customers I think there’s something coming here or brewing with fabric where I’m going to be more easily able to distribute my data as a company into clients worlds of data like let’s imagine they’ve got fabric I’ve got fabric hey you want an extract of all the data let me just build a a a solution a tool that lets your fabric

48:19 solution a tool that lets your fabric communicate to my fabric via shortcuts here’s here’s it already pre-loaded boom ready to go so I’m building the data I still own it I control it if the customer stops paying me I can turn that off but now I have this ability to like integrate very easily to other people by giving them shortcuts through these apps potentially there’s a whole bunch of really interesting use cases here that I think you’re going to be very creative here if you with nothing else if you’re if you’re a powerb developer this is not going to be as important if you’re a company that has data or you’re a data

48:49 company that has data or you’re a data company that Services data to your customers you better pay attention to this part workloads potentially could change how your business works yeah I like this one a lot all right one one more thing here I have to I know we’re getting close on time to the end here one of the major announcements that I really really like here I’m a huge data bricks fan I’ve said this a number of times on the podcast I’m a big fan of data bricks and the unity catalog has been data bricks

49:20 proprietary way of keeping you in the datab bricks ecosystem right it’s it’s it’s a a way to categorize your data it’s lineage it’s helping you manage all your models your ml stuff as well it’s it’s part of that guiding bundle of everything that’s inside your datab bricks environment well now you have they recently announced it they announced one you’ll be able to connect to the data bricks Unity catalog great so in just a few clicks you’ll be able to go into your catalog you can load

49:50 to go into your catalog you can load the full catalog and all the schemas and this is nice because it will automatically be taking care for you if you update tables if you update schemas if you add new data everything will just stay immediately up to date so for one that’ll be huge all my existing workloads just continue to work so my investments stay but now I get access to all the information so one thing I I think I’m feeling here is if you’re not in the datab bricks Unity catalog you’re going to want to be because it’s going to be so much easier to get it in and

50:20 to be so much easier to get it in and administer data into P fabric by using data bricks there’s something else that came out in that announcement and you have to be very like you have to read on a little bit the second half of that announcement is Federate one Lake as a remote catalog in Azure data bricks so how I read this is you can now create a

50:43 how I read this is you can now create a lake house and it makes a catalog in one Lake that you can then go directly access inside data bricks so this is a two-way street now create data bring it over into fabric make lakehouses and then if I want to go use it for AI machine learning training going back and forth I now have the ability of going back over to data bricks and using my existing like this

51:13 bricks and using my existing like this data still stored in one Lake one lake is not going to act like a Unity catalog back to data bricks that’s pretty huge I think so the fact that they’re going both directions data bricks to fabric and fabric to data bricks this reduces a lot of the friction and so whatever works best for your team you can now just utilize that so I I feel like i’ I’m not smart I’ll tell you this right now I’m a dummy I I make a lot of mistakes but I have made a very good decision by learning data bricks it has been a very good career move and it is

51:45 been a very good career move and it is incredibly wise for me to understand how it works it’s helped me a lot of this new fabric stuff i’ and understand things a lot faster because I’ve already been doing it for years so I’m very pleased to see that there the integration between the two platforms is getting much tighter yeah and I think ultimately you tighter yeah and I think ultimately there there are some questions in know there there are some questions in the chat about Donald like isn’t data breaks a competitor yeah I isn’t data breaks a competitor yeah but in this new world competitors mean but in this new world competitors are working alongside each other yeah especially in the world of analytics

52:16 especially in the world of analytics where the tighter those Integrations are between these systems the less reason there is to like fully migrate off one or the other unless it gets to a Tipping Point where all of the sudden you’re there’s a need but to to your point there’s not right like data stored like many many many datab bricks implementations are already storing data in Azure right you’re mounting stuff to an Azure data Lake the data’s ecosystem is creating the Delta tables managing it

52:47 is creating the Delta tables managing it doing all the the the things so fabric is also doing that to access my data via data bricks connection or whatever like there’s no need to m anything right does that mean I don’t have new workloads in fabric well well no because now I have an ecosystem where it doesn’t matter where my data lives and it what what this is doing is saying like whether it’s on premises for certain things it’s not everything out of the gate but in this world this is fabric is the mesh that’s

53:18 world this is fabric is the mesh that’s basically allowing organizations to have your third party app to have your random SQL database to have your data bricks over here and your fabric over here what it doesn’t matter because when we’re trying to bring all the data together to make it meaningful that’s what Fabric’s providing right it’s no longer a man we have all these ETL processes and these services and everything that we’ve got to plug in together and like all of the infrastructure and architecture now it’s just much easier to access these systems and that I think is is the the important

53:50 and that I think is is the the important part here it’s mattering less where we like what we’ve decided in terms of where we want to store and process our data and it’s mattering more that we have accessibility to all of that or larger swats of that in the fashion that we don’t like especially when we’re not paying for data movement right and and that’s in large part why a lot of these are challenging exercises for organizations because regardless if you wanted that data Lake right if you wanted all of your or data in a in a

54:21 wanted all of your or data in a in a place it costs money to move it all there yes and I think these are all moves to just ensure that hey yes like eventually you’re going to have some costs in here when you’re when you’re processing things but it shouldn’t be on the most remedial thing of like just accessing your data right so I think I think that’s one of these things that we may not embellish as much about so like architecturally yeah I think there’s a lot of mean I think there’s a lot of similarities between what data

54:51 similarities between what data bricks has done where fabric is going but that to Microsoft what do they do they come in right they leverage a lot of the things that have been going on that others have been Pioneers for and no it doesn’t necessarily mean that you’re just going to move off platforms that especially if you are a data bricks customer like you have large volumes of data I guarantee like typically you’re running large large big things those aren’t things that you just go oh fabric new thing let’s move correct but there is there is a lot less friction now than

55:21 is there is a lot less friction now than you used to be able to so like right you’re writing this and this is now it’s going to be okay now what are we actually competing on is there a feature set that we’re competing on is there a price point that we’re competing on like what what is the Now new competition that we’re looking at between which platform do I pick and what do I start with yeah so I’d be remiss if I didn’t do one last note here you’ll notice in your powerbi environment there is a new icon that you see called The Real Time Hub so

55:53 that you see called The Real Time Hub so one other very overarching I think theme of build that I’ve seen being here has been this idea of like things are now real time data is moving I have reservations about that concept I I like the idea of it but I really do feel that there has to be a concerted conversation around do we really need it do we really need the ability for us to have real time analytics are we making decisions fast enough that the extra cost every every

56:23 enough that the extra cost every every time you go to real time something’s got to be on your for some compute to do things to get the data to you fine if you can justify that if the value is high enough yeah do it so I haven’t again I haven’t played with the realtime Hub I don’t really know what’s going on there right now with yet so I can’t speak too much about like the platform the program how much compute you does it does it use or consume but I do think there’s a very interesting story around this whole Real Time data what data

56:54 this whole Real Time data what data is coming to you in real time and do leverage that so that could be really interesting actually if you and I have ever agreed on something more this is it so I I the Need For Real Time the need for real time if we’ve ever agreed more this is this is that moment because no I’m the same exact boat it’s cool but yeah you’re right there’s a lot of cost there I don’t know in terms of where the plays or the larger story here there’s obviously more coming here like right this is not the end here like obviously

57:26 this is not the end here like obviously this is a topic for another day right but just real real quick like I think we find ourselves in the world of it’s okay to have delays and process data because obviously we’re consolidating all this data together we’re doing a lot of things to it there’s a lot of processing and time that has to happen before you put that in front of somebody that’s not all reporting cases right there are very but and this is where I think real time you have to be very strategic about it because you’re G to it’s going to cost you do you have some reports that

57:57 some reports that require like views into production and that’s where I do see that there would be a fantastic use case of like small subsets or very highly used reporting where somebody needs access to what the real-time data is in the production system you just don’t want to you just don’t want to hit the production system you never wanted to hit production system right which is why they build availability groups and all these other readon things and ultimately all to all this is to me is like offering that other type of reporting that absolutely

58:27 other type of reporting that absolutely requires that you have the production changes happening in your reporting when you’re when you’re pinging them totally agree with that one so this will be definitely be we’ll need to dive in to some more videos there’s a whole bunch of new things here added into the fabric ecosystem all right Tommy we got our work cut out for us we got a whole bunch more learn fabric we got if if we didn’t have a lot already we got to do a whole lot more because there’s a whole lot of new things we got to play around with now which is exciting I like it but if I wasn’t

58:58 exciting I like it but if I wasn’t overwhelmed before with fck I’m getting more overwhelmed now with all the new options and things and how do I pick which one I’m using which which real time do I care about how can I leverage these things so very exciting Mir do I own oh man yes exactly right so anyways with that we want to say thank you very much for joining the podcast we hope you enjoyed this conversation around the build conference there’s probably going to be many more conversations around build and maybe we’ll even try and do a MVP’s react to the build conference as well so maybe we’ll when I get back in town and we get

59:28 we’ll when I get back in town and we get a little bit more settled we’ll we’ll watch the video together with you and we’ll react to it as as it happens live so I think we’ve had some requests to do it again from last year we did it last year as well so we might do it again this year as well if you want to see that let us know in the comments below also if you like this episode if you like what we’re talking about here from build please share it with somebody else let somebody else know that you enjoyed this content and we’d love to get the word out there that this is a pretty decent somewhat good podcast that you can use while you’re run Tommy where else can you find the podcast you don’t have to run and listen

60:00 podcast you don’t have to run and listen by the way you can find us you can find us on Apple Spotify or wherever get your podcast make sure to subscribe and leave a rating it helps us out a ton do you have a question an idea or a topic that you want us to talk about in a future episode head over to powerbi. com powerbi tips social media channels do you sound very New York we

60:30 you sound very New York we never you say all all all social media channel certain words that are very new yorky what’s what’s the matter with you matter for you all right with that thank you all very much we’ll see you next time time [Music]

Thank You

Thanks for listening to Explicit Measures. If you enjoyed this episode, please subscribe on YouTube and share it with a friend. You can also find more Power BI tips and tutorials at PowerBI.tips.