What if I told you that you can build

the entire Power BI report that to four

pages or five pages with all the DAX

measures, visuals, data modeling,

everything inside the Power BI Desktop

without going inside Power BI Desktop.

And that too in not more than 5 minutes.

So, that's what is the power of AI right

now. I'm not going into detail if it is

great, if it is not. I'll show you

practically and you can decide yourself

if it is going to grow or not. So, this

is what has been the output. So, I can

see one data set if you see over here. I

have dimension outage and then we have

one of the data set.

So, you can see it created four pages.

One one is the overview, then we have

time analysis, then we have

outlier analysis. You can see some of

the errors and then we have impact

intelligence and again you can see one

of the error over here.

So, this is not complete of course and

you would say that this is not good

looking and this might not be impactful

for the end user, right? But, what else

you can expect? This is just in 2

minutes, right? And then you can improve

by using better context and you can

provide more information about what

exactly you need, the color coding, all

the brand colors and everything. So,

it's up to us how we can improvise, but

this is great to start with, right? Like

2 minutes and you are good to go with

the basic visuals and all the measures

have been created here by itself. Right?

So, this is the video where I'll explain

you step by step how to use Cloud Code.

First of all, so how we can create the

skills in Cloud Code, how the Cloud Code

will utilize the skills to create this

kind of report. So, let's get started.

All right. So, before we deep dive in

Cloud Code, let me explain you the

architecture of how the data is flowing

so that you understand from the top

level how the things are going on and

how the Power BI Desktop reporting is

done automatically. So, first of all, we

have Power BI

Desktop file. So, if you see over here,

so this is Power BI Desktop file. It is

running locally, you can see. Right?

Now, it should have a semantic model in

memory. So, once you have all the data

fed in, all the tables inside, now we

have Power BI Desktop, right? Now, you

will connect this to

Cloud Code. How you can connect to Cloud

Code? Through MCP server. So, once you

have connected MCP server to Cloud Code,

it will as an interface. So, just think

about this as an API and if you don't

know how to connect MCP server to Cloud

Code, I have a video explaining how to

create this so you can see on the top

right corner. So, you will see how to

create the connection. So, let's assume

you have already connected the

connection MCP server inside Cloud Code.

So, now my Desktop file is able to do

communication with Cloud Code. Right?

Now, if I mention that hey, can you

connect to the Power BI Desktop file, it

will connect, right?

Now, we will create some of the skills

inside Cloud Code such as Power BI

scaffolding, Power BI connect, Power BI

data modeling, Power BI reporting, Power

BI beautify. For example, so Power BI

data modeling, what it does? It

understands how to create star schema.

For example, it understands if we have

five or 10 tables, how to connect them,

all the relationships, all the DAX

calculations, everything from the

modeling perspective. So, this skill

will be useful there.

For example, Power BI connect, it will

connect through to the Desktop and this

is how we can utilize these skills and

ultimately it will create a PBIP file.

So, if you have not worked on PBIP file,

it is highly suggested that you

understand what is PBIP and then

it will it is more like

disks format. It comes as a folder and

you can easily connect PBIP to your

Power BI Desktop file. Right? So, this

is more about development side. For

example, if you're working with GitHub

and you want to track the version

controlling and you want to do all sort

of things. If there are multiple people

working on the same project, then

definitely PBIP is the way you have to

do. I have one more

um

document on this, anatomy of PBIP. So,

I'll show you this as well after this

diagram so that you understand how it's

done.

So, ultimately once you have created

these skills and you mentioned that hey,

using these skills, can you create a

PBIP project or file for this particular

data set? So, it writes all these JSON

and other files and then creates this

folder. Once this folder is created,

then I will utilize one of the file in

PBIP which is the

report name PBIP and I will open the

Power BI Desktop file and then from

there I will open this PBIP and then

final report is ready. So, this is how

the data will flow in, right? Let me go

to anatomy of Power BI just quickly. So,

this is my report, right? Under this

folder,

I will have my report.report. So, this

is

what user will see. This These are all

visuals.

My report.semantic model, this is all

data,

tables, DAX, everything. .gitignore, you

can ignore for now, but this is more of

Git configuration. So, if you're working

on GitHub, you're connecting. So, for

all those things, so this is the file

required. My report.pbip, this is like a

connector. Simple, you can expect this

as a shortcut key and then just connect

this and it will redirect to which

semantic model it is taking and all

those things. Right? So, this is root

project folder and under these we have

multiple projects and

folders and files and this is how the

PBIP is created and there is a huge big

document to understand what exactly it

is and I can share this as well with

you.

All right. So, coming back to this

diagram, so now we have understood how

the data is flowing. Now, the first step

is to understand this aspect, Cloud

Code. To understand what is Cloud Code

first of all and secondly, how to create

these skills. So, once we are able to

understand Cloud Code, once we are able

to create the skills, then we would be

easily able to utilize the Cloud Code to

say hey, can you create complete Power

BI reporting using these skills and then

it will give me PBIP file and then I can

utilize that to show in Power BI Desktop

file. Easy. So, now the next step is to

understand Cloud Code and creating some

skills.

All right. So, first of all, quickly

let's understand the Cloud Code

interface. So, first of all, we will see

what are the options in Cloud. So, if

you see on the top center, there are

three options, chat, co-work and code.

I've already explained this before. Chat

is most like a normal conversation that

you do with ChatGPT as well. Co-work is

more like a system where you can work on

your pipelines. If you want to clean

your Desktop, for example, you give the

exposure to the folders and then it can

access the folder. You can ask anything

to do and it will do in those particular

folders. Code is mostly about how you

can utilize your natural language

into coding. So, how to convert that.

For example, you are just typing in

English and then it will give you a

beautiful code to create some websites

or applications, all those things. So,

this is what code is used for. So, this

is what we are using today, code. Right?

So, if you're toggled over here, you can

just easily toggle switch towards code.

Right? So, if you see on the left hand

side,

there is a new session. So, if you want

to create a new session, you can search

in your historic sessions, what is the

scheduled ones, dispatch.

This is important customize where you

will see all the skills, all the

connectors and everything and this is

the

historical chats that you're doing with

the

Cloud. Right? Currently,

this has been updated to this version. I

just need to relaunch. I've not done it

so far. And this is where you can go to

the settings. So, if you click here, go

to settings and then you can see your

account system, billing, usage,

capabilities, all those things. Right?

So, I'll just go back. Now,

first step is to create a skill, right?

For that, I'll just go to customize,

right? I'll just click on customize and

then we have two options, skills and

connectors. So, I just need to click on

skills,

right? And then it will show me all the

skills that I've already created inside

Cloud. It will show under my skills

section and there are few of the

examples that you can utilize. Right?

So, there are three ways to add your

skill inside the customize capabilities.

So, first of all, you have to click here

on this plus icon. So, once you click

here, there are three options. One is

create with Cloud. So, easy. Just it

will redirect to Cloud conversation. You

will just text and I want to create

a skill that will help me to

write a Python code, for example, blah

blah blah. So, you'll just mention

explain it explicitly what exactly that

skill is used to do.

That is one way.

Second is write skill instructions. So,

if I click here, it will ask you what is

your skill name, then describe that and

give some instructions. So, this is an

example. It says weekly status report.

So, description is generate weekly

status report from recent work. Use when

asked for updates or progress summaries.

Instruction is summarize my recent work

in three sections, wins, blockers and

next steps. Keep the tone professional,

but not stiff. Right? So, this is how

you can create skills. Third option

is uploading a skill.

So, this is how I did. So, I started

with a create with Cloud.

Right? So, it will redirect me to this

option. Now, make sure that it is a

Sonnet 4.6.

Right? And then over here, you don't

need to work on anything else. Auto

accept edits is fine. Right? So, now

here you will type in I want

to create a skill.

This skill

should be able to help me

with Python scripting.

You are an

Python expert. You can mention the book

or other references so that it refers to

that and then create this and all those

things. So, once you're done, it will

give you skills.md file, that is skills

.md file.

So, MD means

markdown. So, this is one of the

file format and it is straightforward.

It is a text format. So, if I go back to

customize and go to skills and click on

plus. So, once you create, it will give

you MD and you can upload from here as

well. So, once I click on this

and you will see over here this is

skill.md under Power BI scaffold. I'll

go back here. So, these are some of the

skills which I created. So, if I go to

data modeling skill.md, this is what I

have already added over here.

Right? So, these are all MD files.

Now, whenever I go to code

as for example, I go to new session. I

mention

slash and then I mention Power BI

connect.

So, it will show me. So, it says it

connects to a Power BI semantic model

either a local Power BI desktop instance

or a fabric workspace model. Supports

fuzzy name matching with confirmation.

So, this is in project. Right? So, this

is helpful. So, now I can easily write

connect to

the Power BI desktop instance and then

it will connect and then we can connect

go ahead on

creating the reports. Right? So, now

that you know how to create the skills.

Right? So, that was the most important

part creating a right useful skills. So,

this will take some time and there are

some of the documentation. I'll just

show you after this on Anthropic's

website. There are some tutorials. So,

let's go there. All right. So, now in

order to understand where the tutorials

are, you have to go to this URL

cloud.com resources tutorials. Once

you're under tutorials, you will see

all the options. So, you can filter out

which category you're looking for, which

features are you looking for, which

product you're working on. So, I'm

working on cloud code. I can filter on

cloud code and see what are the options

and then straight away I could have

selected skills because I'm looking for

skills. But in cloud code you will see

troubleshooting skills is an option,

sharing skills is an option. There are

so many other ways. Like creating your

first skill as well. So, let me click

here.

And it will redirect to how to create

your first skill.

Okay. So, this is a video and then you

can easily watch and see how it's done.

But uh

if you go here on teaching and creating

the new skill, it will give you a lot of

information and it is very easy. If I

toggle to what are skills, it will just

say that whatever you are doing and if

it is repeated and you don't want to

repeat whatever you're writing as a

prompt, then create a skill and then

utilize again and again instead of

typing because we are moving towards

that space where where your prompt

should be smaller instead of, you know,

giving lots of information. It should be

more crisp and short.

Right? So, this is how you can create

your skill. So, as it has been saying,

save the skill that cloud creates in

settings, go to capabilities and inside

that it is skills. Now, it has been

changed from capabilities to customize

as I have shown you. Right? So, once you

have created the skill and you will see

and also this is important the skill

directory and this is how it's done. So,

first of all, it will give you

my skill name and then inside this

directory you directory you will have

skill.md that is your markdown file and

there would be some more optional files

such as scripts. Inside scripts you will

have a Python script and then you will

also have a reference markdown file as

well as the assets. So, these are the

optional files and this will enhance

your skill capabilities. So, in case for

example, I want to create a data

modeling skill and if you add some

references such as data modeling

concepts, star schema concepts and go to

Kimball uh

uh PDF and attach that as well. It will

have more context. It will have more

understanding on what exactly you're

looking for. Right? So, this is how you

can enhance the skill and this will

work like a charm. All right. So, enough

talking. Now, let's go straight into

cloud code and work with the prompt to

see if we are able to create the report.

So, I'll just go to new session. Right?

From scratch and let me change the

folder.

First of all, inside this, let me create

a new folder

as a new project

Power BI

demo

for YouTube.

All right. So, this is my folder.

And now one option is I can attach all

the skills over here, but I already have

that. So, let's see if it works.

So, first of all, I'll ask it to

connect to the Power BI desktop file

named

dashboard of

It is

cloud one.

Let's see if it works. I trust this.

So, it takes time. It sees if there is

an option if there is a desktop file

open. For this to work, your Power BI

desktop file should be opened.

So, it says since you mentioned Power BI

desktop file, I'll connect locally. Let

me check for running instances. Now, it

found it. Right? This is great.

So, now it found the Power BI desktop

file. Now, we can work on the reports.

Perfect.

Would you like me to sync the live model

into PBIP semantic model folder?

Uh not yet. Uh now

yes, you can do that.

I want

you to create four-page

Power BI

reporting

for this data set. Ask me questions

to get the

best possible

KPIs and relevant information.

All right. So, let's see what happens.

I'll also fast forward in between if it

takes a lot of time because it generally

does. So, because it's doing everything.

It It is checking all the skills. So,

I'll say allow all

for this session because it happens

because

works on

the data modeling, it works on

reporting, it works on so many things

internally. Right? Because this is what

you actually do and it takes you a

couple of days for example to create

these four reports or four five pages.

So, this is what it's doing in five

minutes. So, definitely it will

you know, time and it will ask me for

all the

uh access things. All right. So, now it

has given me some questions. It says, I

now have a clear picture of your model.

It's a power outages data set with even

details, customer impact, duration,

geographic info, etc. etc. So, now I

need to answer these questions so that

it can be more relevant. So, I'll just

type in uh who is the primary audience?

I will say executive

with the

who needs

high level

as well as

in-depth

understanding

of what's happening.

The most important KPIs

would be

total

outage events

total customers

impacted

average outage

duration.

These are all already there. So, we

don't need to write because it says

common options for this data set. Uh

these will always be there, but if you

need something else, just type in. Just

I'm just mentioning these.

And you can copy paste just in case. And

now

I want

four

page

analysis.

First is

you can say, let's say,

overall

analysis, then

customer-wise

analysis.

I want to get outliers as a page as

well.

Then

time intelligence,

that is

year-on-year and month-on-month

analysis.

Also, I want geographic

analysis. So, I can say I want five

pages.

So, it's all good.

For geographic-related

focus on

both

state as well as LGA.

Let's see.

It also says the file is a PBIX, not a

PBIP. So, there is no PBIP folder to

sync to. I'll skip the Tim Dale export,

that's fine.

It will again take few minutes to

process this and then we would be able

to see the outcome.

And in fact, if you want to make it even

faster, which is like 100 times faster

than this, then I would suggest work on

VS Code and then it integrates with

GitHub. So, that's good for version

controlling. Second, you can

make the changes in the

permissions in VS Code. So, there is a

settings. So, if you just toggle it off,

so it will never ask or toggle it on for

allow and never ask for that particular

session. So, it will not ask you before

editing. So, it works seamlessly. So,

even if you're away from your laptop,

you don't need to always click on allow,

allow, allow and it will work.

All right. So, it asked me one quick

clarification before I start building

for this and this. I need total customer

served.

As I don't this is in do you have fixed

number?

Yes.

I want this

for [snorts]

No, there is no fixed, actually. It is

better to have a what if

parameter would be better.

All right. So, it took us some time. It

took somewhere around 7 minutes. So,

what I did was what if parameter would

be better and then it ran various

commands and then finally, you know, it

gave me, close your current PBIX and

then open the dashboard of Cloud 1.pbip.

So, once I open, hopefully it works. If

not, then we can reiterate and see how

to work around.

All right. So, now we'll open the Cloud

1.pbip file in the desktop. So, I'll

just go straight to Power BI, click on

file, click on open report, browse.

And then I'll click on PBIP file and

then it will load some data and then try

to

create some

uh reports or load the semantic model

and reporting. So, it will give some

errors, I believe, uh like it can't read

semantic model, all those things. So,

this is what we have to fine-tune our

uh skills and improve on how to, you

know, get the development mode even

faster. So, let's see what the error is

and then we can, you know, modify that

as well.

All right. See, this is what it comes

with. Issues were found. Cannot read

this folder, this file, and property

semantic model has not been defined and

schema doesn't allow additional

properties. So, required properties are

missing from the object, this one. So,

artifact line nine, position five. So,

I'll just copy details to Cloud board

and paste in my Cloud. And just straight

away, like without

you know,

putting so much effort on your mind,

just paste and see what Cloud gives you.

And once we get the solution, we can

update our skills. So, that's the best

part. So, it says the error is clear,

the PBIP schema only allows report

entries in artifacts, not semantic

model. So, now it's fixing the file. So,

it is fixing the PBIP schema.

And now it says, try to open it again.

All right. So, I'll close this. I'll

also close this one.

Over here, I'll click on file, click on

open, and click on browse.

And then let's see if I open this and if

it works this time.

So, again, it will load Power BI Desktop

and then it will try to load all the

data, that means all the tables. And if

it works, amazing. Otherwise, we'll

again iterate to get the final solution.

All right. So, now again, it says report

definitions. So, the error changes. So,

previous error was rectified. It says

required artifact is missing in

this PBIR. So, I'll just go back to my

Cloud and paste

and see what it does.

All right. So, once we give this error

again, it gave me the updated version.

So, let me open this file and let's see

if it works this time.

All right. So, it's loading project,

loading data model. And once we are

crossing this step, then it should be

fine. So, let's see if it crosses this

step. Loading data model and then it

says loading queries, so that's good.

It's a positive sign for us.

So, once everything is loaded over here,

then we can see, you can see on the

bottom we have four new pages created by

itself. Overview, time analysis, outlier

analysis, impact intelligence. Great.

Again, same at the right-hand side, we

have all the tables which were there.

So, currently you can see visuals are

there, but it's all blank. You can see

it's all blank.

So,

let me hit refresh and let's see if it

works.

All right. So, it says could not find

this path. Let me change.

Or maybe I can just go to data source

settings and change the

path for that.

Over here.

So, I need to have the same path over

here. So, I'll change the source.

And then over here,

I will paste that.

All right. So, it should be fine now.

Click on close and then refresh.

Once we apply this change,

and the same step we can easily update

using Cloud so that it can update your

expression Tim Dale file in your

semantic model so that it remembers the

latest data source. Right. So, this has

been updated. Let me refresh it once

again and let's see if it works.

So, once the data set is loaded and

updated over here, then our visuals

should work.

All right. See, you can see over here

some glitches. You can see these are my

filters which are not really looking

nice, but these are my

data cards which are looking fine. This

is my bar chart. This is my column chart

which is month-wise total outages and

this is again bar chart for total

top 10 LGAs. And then this is cumulative

customer versus outages trend. So, this

is again, I can see. And if I go to time

analysis, again, we have all these

charts. Then if we go to outlier

analysis, some of them are not working.

We can fix them depending on how you

want to do it. And then impact

intelligence is also there. Just one

data card is not working. We can check

that. I mean, we can, you know, improve.

You can,

you know, give the brand colors and you

can change the default background size

and everything. But this is a great,

great advancement. I mean,

I didn't do anything on Power BI Desktop

and itself when I did [snorts] use

Cloud, I just mentioned all the skills

and that too, that's the one-time

effort. Now, you don't need to update or

add the skills again and again. You need

to update whenever you feel something

has improved. Right? But then this is so

important and interesting for all Power

BI developers. So, it's so great. Just

need to type in as a natural language.

Think like a businessman. Think Think

like your CXO that how they want to see

the reports. All you need that is just

type in that, "Hey, I want to do the

analysis of customers, how the stores

are growing, degrowing versus last year.

What is the forecast? What is the demand

for particular SKUs?" So, you just need

to type in and they will create the

Cloud will create the visuals for you.

And then this is a start. I mean,

it doesn't mean that Power BI developers

are gone and anyone who just need to,

you know, type in Cloud and they can

work because it takes a lot of knowledge

about data modeling, about creating

reports. So, whenever there is an issue,

you need to modify. Now,

it's even more important to know DAX and

data modeling because if

randomly Cloud is using a particular DAX

measure which is not

inclined towards performance and it is

giving you performance issues then it's

really hard for a beginner to debug,

right? So you have to be an expert. So I

mean, the doors are opening but

this is a game changer I would say in

terms of developing some reports. So you

can see and now from here you can so

much modify and at least you have a

start of way to go. And I have

cross-checked these values are all

correct. So yeah, that's a great start.

Also, if you need to enhance your

agentic AI to develop Power BI reports,

I would strongly suggest you to utilize

VS Code. So if you go to VS Code,

install Claude and Claude Claude Code

use that.

So once you have installed, also install

GitHub, connect through that and that

will help you and over there you can

just edit the permission that you don't

need uh

the Claude or VS Code to give or ask for

the permission again and again. So once

it's accepted, you do you can go away

and it will work for you behind the

scenes. So that's amazing development

and I would strongly suggest you to

practice this, try this at your end and

then

there would be so many new developments

coming in the next few weeks. So stay

tuned. All the best. Thanks a lot for

watching this video.