- [Adam And Carly] Yooo! - What's up, I'm Adam. - And I'm Carly. - And Carly's gonna show
us the magic of Copilot inside of Microsoft Fabric, but more specifically, from
a Power BI perspective. For the folks watching, like
what do you do at Microsoft? - So I'm a Program Manager, and I split my time between
Copilot and metrics things. But today, specifically,
I'm gonna talk about how everyone within
Power BI can use Copilot. So how you can use it for
Modeling, for Report Creation, and then how you can use it in
the Consumption Mode as well. - Awesome, all right. Well, enough of all this talking. You know, how we like to do
it here in Guy in a Cube. Let's do what? Let's head over to Carly's machine. - All right, so you can see here, I am already in my Modeling View, and you may have noticed, looks
a little bit fancier here. So I have my Model Explorer
up in this new build. So you can see I have my tables here, and I have my model over here. Here I have all my
Measures co-located here. So one thing that we have done here is just given Copilot the ability to write measure descriptions for you. So I'm just gonna click on, let's say this measure right
here, Product Sales Top 3. And I'm gonna go ahead and
click create with Copilot in this description area. And Copilot is actually
just gonna go ahead and write this description for me. The Product Sales Top 3 measure
calculates the net sales for the top three products. If a product is not in the top
three, measure return zero. Okay, I know my model pretty well. That's looking really accurate, so I'm gonna go ahead and just keep that. So this is really good. It's gonna make the report
creator's life a lot easier. This is just a really
good documentation task. One thing that I'll say
is I can neither confirm nor deny that the next do
is bulk measure descriptions for all of this at one time. So this is looking really good, but now I wanna go and
do some more tweaking, and I wanna write some
DAX for my model here. - [Adam] Oh. - [Carly] I'm a little
bit better at SQL myself, but I wanna get into DAX, so I'm gonna- - We like SQL also. That's fine. - We don't discriminate here. We like SQL, we like DAX, we like it all. So I'm gonna have Copilot help me learn a little bit about DAX. So I have my editor here. The first thing that I see in
my DAX Query View with Copilot is it gives me a nice little
query here as an example. And if I don't exactly
know what's going on here, I can just say run, and
it plops a table down so I can have that as an
example to just look at. - [Adam] That's beautiful DAX, by the way. - Kudos to our engineering team and to our design team who made this look super, super simple when it is in fact not very simple as we all know. And if I still don't exactly know 100% what's going on here, I can actually just go ahead
and highlight some of this, open up my little Copilot button up here, and just type in natural language
a question to this query. And this is just gonna
explain in natural language what's going on. So I can open that up. And this is telling me exactly
what this function is doing. Okay, this is using a
lot of words to just say this is returning, you know,
the top end amount of rows here for this specific table. This is pretty, pretty cool. So now let's say I'm feeling
a little bit more confident, a little bit more comfortable. I'm gonna start my own
query on a new tab here. - [Adam] All right. - [Carly] So I'm gonna click
my little Copilot button, and I'm gonna say show me
the return rate per store for the latest month. And it's gonna turn my
natural language question into some beautiful docs. - [Adam] It's magic. - We like to call this
auto-magically on the Power. - [Adam] Auto-Magic, all
right, we'll take that. - [Carly] So I'm gonna say keep 'cause this looks pretty good to me. And then I'm gonna and run this. And right down here it has
returned me a nice table that did exactly what I asked for. - [Adam] Excellent. - [Carly] So I'm gonna
go one step further, and I'm gonna say,
actually, I want to know what month this is actually talking about. So I'm gonna go ahead and
highlight my query up here and invoke my little Copilot button again. And I'm gonna say, include
what month this is. And this is gonna do something very cool. - [Adam] Wait, that's doing a
diff right there in the DAX. - [Carly] Which is pretty cool. So this is showing me what's
different, what's the same. And I can go ahead and, you know, edit it in here or I can say keep. I'm gonna say keep because- - [Adam] All right. - [Carly] So I'm gonna go
ahead and say run that, and then right down here it says opp, this is happening in June, so- - [Adam] All right. - [Carly] It's looking pretty cool. - [Adam] I love it. - So the last thing
that I'm gonna show here is a way to get even
more familiar with DAX with the help of Copilot. So one thing I can do, tell
me about common DAX options. And this is gonna return just
the most common DAX functions. And I can read up on these, maybe learn a little
bit about the difference between Sum and SumX. And I can say show more. And here I have just kind of a nice yes. Wow is right. This is just a really,
really powerful tool. So it will write you DAX queries. It will explain DAX queries. It will explain DAX topics. It's just, it's really, really nice. So next, let's imagine that I have done all of this beautiful modeling. My semantic model looks like it's in really, really good shape. So I go ahead and I
publish the semantic model, and now I wanna step into the role of (words twinkling)
report creator. I wish I had a different hat. - [Adam] So dramatic. - Yes. So here we are. Let's say this is that semantic model that I was just working on, and I published it to the service. And let's say I'm kind
of a beginner creator. One of the best things that Copilot can do is explain to me what
exactly is in this model and help me know where to start. So I have this little button
here that says Explore. And for those who are maybe not familiar exactly with what Explore is, this is a feature that we
just recently released. This is in preview. If you're not set or committed to doing a full report creation, but you still wanna explore
what's in the model, you can do things here like
drag a couple different things here where you wanna put a filter on here. Let's see, I think I have
some revenue in here. I can drag this here. I can change the visual type and just kind of really
quickly just do some lightweight explorations
if I'm not committed yet to building and publishing a report, but I just need to get a quick answer. The way Copilot can help me is maybe I don't know exactly
what's in my data set yet or what's in my semantic model yet, and I need to get an overview. I have this little button up
here that says data overview. I can go ahead and just click that, and with Copilot, get a really quick natural language overview of
what is in my semantic model. And what's funny is when
I was actually picking the semantic model to do this demo, I used this to click
through a bunch of the demos that I like commonly do. And I was like, which one
should I use for this? Anyway, back to our scenario. I'm like, yes, this is
the one that I need. I start exploring my data, but this visual's not really
doing what I want it to. I think I need a little bit more help. So I'm gonna abandon my exploration here. I'm not gonna save this, and I'm actually gonna just
open this in a blank report. And right off the bat, I'm just gonna click my
Copilot button up here. And immediately the
Copilot pane is gonna open, and I'm gonna say suggest
content for this report. - So this Copilot is the
one that came out first, and like this is the
one I'm familiar with. - So you're a fan is what I'm hearing? - Yeah, yeah, yeah, yeah. - What this is gonna do is take a look at what's in your semantic model, and it's gonna generate
essentially an outline for your report that says, hey, based on your semantic
model, here are some pages that we think is gonna be
the most useful for you as a creator to build
for a consumer to view. These are the insights that are gonna be the
most useful to glean. So you can take a look through these. You can expand all these and
kind of read through them. It gives you a title, and it tells you kind of
what the content is gonna be for that page. Let's take a look at this first one. Sales performance analysis. Create a page to analyze the
performance of sales by region, product, and industry. I know from my model creation that region actually
is better as territory in this model, I believe. But otherwise this is looking pretty good. So I'm gonna go ahead
and see how this does if I just click create. So boom, just like that, in, what was that, like five seconds? I have an entire report page created here that's honestly really beautiful. And you don't wanna know how much time this would've taken me
to create on my own. - Way better than what I would've
created, that's for sure. - [Carly] And I can just
keep going down the line and doing this. Like revenue forecasting and goals, that sounds really interesting. Let's create that page
and see what it does. And I can keep doing
this for all these pages. And boom, look how fast that was. That was like whitening speed. And this is a really, really
amazing starting point, especially when I might not know exactly where to start
with my semantic model. And this next thing that I'll show, it's called the Narrative Visual. And so this allows a creator to actually place a
Copilot visual on the page, and it will offer the
consumers a permanent summary essentially once you publish the report. Let's go to our sales
performance analysis page here. And then I have this visual right here called our Narrative Visual. This is one that I was just talking about. - [Adam] Yep. - [Carly] And let's go
ahead and click Copilot. This is also in Public Preview. So here I have some standard
prompts that I can do. I can give an executive summary, answer likely questions from leadership, give a bolded list of insights. But let's just go ahead
and see what it looks like when I just summarize the data. And what this is doing is taking everything that's on the
current page and scanning it, and it's just gonna create a
summary of what's going on. And then when I publish this, like I said, this will be available to all
of the consumers of my report. - If you're slicing and
dicing on the report, it'll regenerate as well. - That's right. I can say when I'm slicing and dicing, there'll be a banner there that says, I wanna refresh this
and have it regenerate. - Nice. - And if I really like the summary, I can leave it there and
make sure that that's what's, you know, always gonna be there. So one of the things
that's really great here is this gives me references. So when I click on this,
it's showing me exactly what visual on my report
this is coming from. And it's showing you, hey, we didn't just pull this
information from the ether. This is rooted in your data. This is rooted in your
report, so you can trust this. So one of the other
things that we can do here is summarize over the entire report. Or if you want, you know, for some reason, some specific visuals to not be included, you can do that as well. So a lot of control here as the author. For example, if I wanted
this first page here to answer likely
questions from leadership, but for some reason I know
my leadership doesn't care this month about the sales
pipeline for some reason, then I can create this
visual summary here, not taking into account the
visuals on that last page. And then when I publish this report, they'll get the questions and answers from just those two pages. And I can click on those, and
zooms me to the next page, and see where they came from. - [Adam] That's interesting it can be aware of multiple pages. So it's good for like a summary page also. - One of the kind of main
use cases that we see for this visual is people
putting it on that main page as kind of an overview page that says, here are the key insights
across this entire report. Like here are the key takeaways, and it summarizes every
single page on the report. Okay, so let's imagine I
have tweaked this report exactly how I want it. I've created all the pages that I want, and I go ahead and I publish it. So now you know what I'm about to do. (words twinkling) I am now a new person. I have published this report into an app, and now I as a consumer
have landed on this beautifully created app, all
of which is created by Copilot. And I'm clicking through here, and I'm thinking, hmm,
I just started here, and I am very overwhelmed
by all this information. I wish I just had some kind of tool that could help me get the
main insights from this page. Gosh. - [Adam And Carly] Copilot. - Oh my gosh, you're so right! I wish Copilot worked over apps though. Oh wait, it will soon.
(both laughing) So here it is. Right now it works over reports. Very soon, in the next
couple of weeks, months, it will work over apps as well. So as a consumer, I can
come into my report. And I can just say, I wanna
just summarize the visuals that are on this page
and get insights here. So just click in that button, and boom. Again, lightning fast. Very cool. And again, those visual
references are coming with it. So I can click on that and see exactly where this
information is coming from. So as I read through this, I'm seeing this last sentence here. It says, the regions
with the highest number of opportunities are US South,
US West, and US Midwest. And I'm thinking that's kind of all of 'em except for the Northeast. Like what's going on in the Northeast? So let's just double check this and say, what is the region with
the lowest opportunity? So you can summarize across your report, you can ask for insights, and you can also ask
questions about your data. So it's telling me, yep,
this is the opportunity, and it is coming from the Northeast. So let's go down here. So maybe I'll just say something like, tell me insights on this page. And this is very cool
because in this Copilot pane, I don't lose the history
of what I've been talking back and forth to Copilot about. So when I go to another
report, I maintain that history and can go back in the
conversation and reference things, which is pretty cool. And now I might wanna say like, can you simplify these
insights into bullet points? And it's gonna think
about this for a second, and then it should give me
some nice clean bullet points. So the last thing that I'll do here, I do have a page down here that focuses on that Northeast region. Again, I maintained that history, so I can scroll up if I need to. So let's see if we can do
something for that region. And so for me as a consumer, I'm hoping that this kind of just tells me in this region that I
know is not doing well, what are the individual stores that are kind of my high
performers and my low performers? And then I can take that information, kind of analyze it
against the other reports that I have here and see if
there's any insights I can draw with what products are
maybe contributing to that and things like that. I think I've demonstrated
most of the capabilities here. Feel free to play around with this. This is in public preview now. So please go check this out. The app is not in public review. That is coming soon. And the next query view, Copilot should be coming to public review within the next couple weeks. - Love it. All right, so let us
know down in the comments what do you think, first off? Is this like insane, amazing? 'Cause I think it is. Maybe there's something else
that you're interested in that maybe is not being covered yet. Go to Ideas.Fabric.Microsoft.com and get it in there as well. Thank you so much for joining us, Carly, and thank you everyone
for watching, as always. And we will see you in the next video.