Claude Just Changed Power BI FOREVER [Connect Claude to Power BI - Full Setup]

Learn how to connect Claude AI directly to Power BI for faster DAX creation, metadata management, and report documentation.

Key Takeaways

Claude AI can connect directly to Power BI models, enabling powerful and efficient report development.
AI significantly speeds up DAX measure creation and metadata management, reducing manual effort.
Generating detailed report documentation with AI saves hours of work and improves workflow.
Using AI tools with company data requires careful consideration of security and approval policies.
AI will augment but not replace data analysts, requiring continued cognitive skills and oversight.

What the video covers

Demonstrates a real connection between Claude AI and a Power BI model, enabling direct interaction with tables, measures, and data models.
Shows how Claude can inspect data models, understand relationships, and provide detailed overviews without manual input.
Illustrates rapid generation of multiple DAX measures and organization into a dedicated measures table.
Explains how Claude can bulk add descriptions and metadata to columns and measures, saving tedious manual work.
Demonstrates creating comprehensive technical documentation for Power BI reports in Word format within minutes.
Discusses practical considerations for using AI tools like Claude safely with company data.
Addresses common concerns about AI replacing data analysts and provides a balanced perspective on AI augmentation.
Includes tips on permissions and setup challenges when connecting Claude to Power BI files.
Mentions additional AI connectors and sponsored content related to AI-assisted writing tools.
Shares real-world experience as a senior data analyst and Power BI developer using Claude extensively.

Chapters

How does Claude connect to a Power BI model?

Claude connects directly to an open Power BI pbix file, allowing it to inspect tables, measures, and relationships without manual data input.

Can Claude create DAX measures automatically?

Yes, Claude can suggest and build multiple DAX measures quickly, organizing them into a dedicated measures table for easy management.

Is it safe to use Claude with company data?

Safety depends on company policies and approvals. It's important to check if Claude is an approved tool and follow security guidelines before using it with sensitive data.

Full Transcript — Download SRT & Markdown

Speaker A

All right, look. I just connected Claude directly to my Power BI model. Not talking about copy-pasting DAX in between chat windows. I'm talking about a real connection between Claude and my Power BI model. This way, Claude can see my tables, my measures,

Speaker A

my entire data model, and it can help me build. I work as a senior data analyst and Power BI developer for a Fortune 300 company, and this has got to be one of the coolest things I've seen

Speaker A

happen in Power BI in a while. In this video, I'll show you what it can do.

Speaker A

I'll show you my exact setup, how you can set it up, too, and the mistakes that people often run into during that process. And at the end, I'll talk about the topic that will definitely be on your mind after we go

Speaker A

through this, and that is, will this replace data analysts? Because this is a really powerful workflow, but it's also a little scary. So, we'll talk about that. And then finally, I'll close out with my honest answer concerning using

Speaker A

this kind of tool or workflow with company data. So, I'll admit, I'm a little late to the game in creating a video on this topic. This is an update that's been out for a little while now, but I've been using

Speaker A

it extensively and I'm super excited to talk about it. So, let's dive in. Okay, so before we dive into the actual setup, I want to show you what it can do because this is very powerful. So, we have Claude desktop opened up on one

Speaker A

side of the screen, and we have a fresh Power BI report with just a few CSV tables loaded in to keep things

Speaker A

easy. The first thing we're going to do is we're just going to ask Claude to

Speaker A

connect directly to the Power BI report. So, check this out. Connect to the open pbix file titled Power BI plus Claude demo. So, we're just asking it to connect

Speaker A

directly to the open pbix file. And when you're doing

Speaker A

this, especially for the first time, you might get asked multiple times for permission to allow Claude to do certain things. You can either allow once or you can allow always or for the entire session. So, that was pretty quick, and

Speaker A

we can see that it has connected to the open PBIX file. First thing I'm going to have it do is tell me a little bit about the data. And we can see that that's even one of the suggestions that it

Speaker A

gives us. So, I'm just going to tell it to inspect the data model and all connected tables. Again, pretty quick and it's done. It gives us an overview, gives me a breakdown of the tables and each of their columns, even breaks down

Speaker A

the relationships. And this is a pretty simple data model. So, there's not a ton to cover, but you can see that very clearly it is connected to the data model because it can tell me about it directly without me giving it any

Speaker A

context. It just knows from inspecting the

Speaker A

file. Pretty crazy. Now, I'm going to ask it to advise on some DAX measures that it can create. So, I'm just going to ask it, "Please advise on some DAX measures we can build for this

Speaker A

report." See what it comes back with. And I mean, literally within seconds it just shot out all of these DAX measure suggestions

Speaker A

that we can build.

Speaker A

And it does give some caveats stating true profit margin is impossible yet and that time intelligence is running on the hidden auto date table. I'm not going to worry about that right now. But, I want you to watch this because this is where it

Speaker A

gets really insane. "Please build every suggested DAX measure suggested except for those mentioned in

Speaker A

your caveats. Place all of these within a measures table." I want you to see what it does next here.

Speaker A

Okay, boom. That probably took like literally a minute and you can see here now that we have a dedicated measures table with folders that contain every DAX measure

Speaker A

it suggested except for those mentioned in the caveats, which were the time intelligence

Speaker A

function and the profit ratio. So, what's this? This is like 15 measures that I made in literally 1 minute. That's crazy. I'll say, if you're doing real development work here, probably want to take it a little bit

Speaker A

slower than that and I would suggest building one DAX measure at a time making sure

Speaker A

it's accurate, it's the right thing, doing some quality checks along the way. But, for the sake of demonstration, I just wanted to show

Speaker A

you how fast you can move with this. You can in practice build dozens of DAX measures in a matter of minutes.

Speaker A

So, even if you're taking these one at a time, you are still moving so much faster than you would

Speaker A

without it. And the fact that it even organized these for us, built the measures table, is so incredible. And so, we just kind of thumb through these in the formula bar, you can see that they look pretty legit. And in my experience, especially

Speaker A

with these simple measures, and this is a really simple data model. So, I'm pretty sure that this is all kosher. A couple more things I want to show you. I want to show you how you can bulk add

Speaker A

descriptions into all of your measures and

Speaker A

columns. So, here I am in the data model view. One of the most tedious things to do when it comes to Power BI reports and Power BI development is adding like properties

Speaker A

into each of your measures or columns. Not something you always need to do, but sometimes it is necessary, especially if you're completely handing off a report file to someone. And we can do this pretty quickly with Claude, as you might

Speaker A

have guessed. So, we can see that in the DAX measures that it created for us, it already filled these out, which is pretty cool. But, the base columns that we started with don't yet have that.

Speaker A

Well, guess what? Claude can build those out for us. Please fill out all properties fields for columns in our customers, products, and sales tables.

Speaker A

Okay, and after a couple minutes, we are done. You can see it took the liberty of organizing each of the columns, too, into folders, which is pretty cool. You may or may not want that. But, the fact that it can organize

Speaker A

so simply is pretty cool. So, if you look at each of our columns, we see that this metadata is now filled out properly. Boom. Something that would probably take you a little while to do on your own, be pretty tedious, we can

Speaker A

now do, and again, a couple of minutes with Claude. It just went through all my columns and did that. So, crazy, right?

Speaker A

Like this stuff is so powerful. I'll show you one more thing here, and that is creating documentation. And I'm literally just going to say, "Create detailed documentation for this Power BI report." It's asking what format we want that in. We'll do it in a Word document.

Speaker A

Uh, because chances are if you're doing this at work, you'll want that in like a clean Word file. "How deep should the documentation

Speaker A

go?" Let's say a comprehensive technical spec. Just to see how detailed they can get. And then while you're waiting for

Speaker A

Claude to work, you could just scroll social media on your phone. Okay, and a few minutes later, we have a nine-page report on our complete Power BI report here. Breaks down the tables, the measures, the relationships, the data

Speaker A

profile, filing notes and recommendations, and it is incredibly well-formatted, looks really good. There's some awkward stuff here, like a blank contents

Speaker A

page. So, it's not perfect, but I mean again, it took I think it took 3 minutes. We got a

Speaker A

detailed nine-page document that probably would have taken you hours to do on your own, if not longer. Because this is very detailed. So, even if you're not just shipping this as is, which you probably shouldn't, you should definitely read through it. But even if

Speaker A

you have to make some edits, just the speed at which we got started is truly phenomenal. So, those are just a few of some of the most powerful things that you can do with this setup. And I've done this with much, much larger data

Speaker A

models, and it really is crazy the amount of insight that it can get into your reports,

Speaker A

and how much smarter you can go about the problem-solving process with.

Speaker A

Right click that and open this with Notepad++. You could use any notepad though. You won't have all of this, but you're going to paste in this first section. So, mine has my dedicated pathway here. In the description of the video, I'm going to

Speaker A

put a template that you can use and fill out with your own [music] information and I'll show you kind of the original format that it looks like in there. Key thing here is to make each of these [music] backslashes double backslashes

Speaker A

when you're filling out your information. From there, you're going to close and restart Claude desktop. And then when you navigate back to your local MCP servers here in the developer tab, you should see the Power BI Modeling MCP. If you run into any

Speaker A

issues, I found that literally just [music] asking Claude about those issues works pretty well and it can help you troubleshoot what your issue might be.

Speaker A

It could be something with the file path or something else. One thing you might be wondering is if you need a paid Claude plan for this. You don't need paid. I personally use Max, but that's just because I use Claude a lot. But

Speaker A

free should be enough here. Just keep in mind that you won't have as much usage on [music] a free plan. Okay, so once you get this set up, you can follow through the steps that I took at the

Speaker A

beginning of the video and you can start having some fun. And while we're on the topic of connecting things to Claude, there's one more connector that I've been using lately. And this part of the video is sponsored by Walter Writes. So,

Speaker A

Walter Writes just launched the first AI humanizer for Claude. It's called the Walter Writes MCP and it lets you humanize AI text right inside of Claude.

Speaker A

Again, you don't have to do any copy-pasting between tools. The setup takes about a minute. You go to your Claude connectors, you name it Walter Writes AI, paste the Walter MCP server URL. I'll leave that on the screen for a

Speaker A

moment so you can pause and copy it. I'll also throw it in the description so that you [music] can copy-paste it. Once that's in, you hit connect and you're pretty much done. So, if you write reports, summaries, or documentation

Speaker A

with Claude, this is the best AI humanizer for Claude available right now. Walter Writes has four modes, it works in over 80 languages, and there's a free trial. Link is in the description, or go ahead and scan it

Speaker A

right here on the screen. All right, lastly, let's talk about a question I get anytime I talk about AI tools with company data. And that question is, is this safe to use at work? And the answer is always it depends. And you want to

Speaker A

make sure you're asking the right questions to leadership. So, here's my rule. I don't connect anything to production data until someone who gets paid to worry about that tells me it's fine. You don't just want to start using

Speaker A

AI tools that are directly connected to company data willy-nilly. If you're in a larger organization, there will be a lot more data governance in place that completely restricts you from using certain tools. But if you're at a smaller company, you'll definitely want

Speaker A

to ask those questions. And honestly, either way you should be asking those questions. Many companies now are very AI forward, and most AI tools have a lot of data governance in place. [music] So, implementing these tools is not a

Speaker A

difficult task, especially for a tool like Claude, which a lot of companies are already using. So, things you might want to check in with your manager or IT about before connecting anything to real data would be things like, is this tool

Speaker A

already approved? Of course, some companies might even have a list of the tools they do approve of. If they do, ask for that list. But again, just ask outright, is this tool okay? If Claude is on it, then you're probably fine.

Speaker A

Another thing to know is where does the data actually go? So, the MCP server runs locally, but the model context gets sent to Anthropic. Your table names, your column names, your measure logic, know that before you connect. [music]

Speaker A

And then, who owns this decision? Because it's probably not you, it's IT, security, or data governance, depending on your company. Just send an email, ask if connecting [music] AI to a client or local Power BI model is permitted, and

Speaker A

get that in writing. And I'm curious, would your company allow this? Tell me in the comments. I'm genuinely curious which companies out there are restricting tools like this, and which ones are okay. My impression has been that more and more are becoming okay

Speaker A

with it, but I'm always curious to know what's going on out there with other companies. So, again, three things to ask before you connect AI at work. Is the tool approved? What data leaves my machine, and [music] who signs off on

Speaker A

this decision? So, it's powerful, and you can use it safely, which leaves the question everyone is really asking, will this replace data analysts? [music] I'll be the first to say that this technology is scary good, but I have been using

Speaker A

this exact [music] technology for months now, and I can tell you straight that no, it is not going to replace data analysts anytime soon. I spend days working on giant enterprise reports that are directly connected via the Power BI

Speaker A

modeling MCP, and just because I have AI assistance that plugs into my data model, does not mean that reports are just getting made in a snap. It still takes considerable effort. I'm still pounding my head against the wall some

Speaker A

days trying to problem solve, and just because you have AI help, doesn't mean the work just [music] gets done in an instant. That's just not the reality, especially when you are working with [music] massive data sets that require really

Speaker A

complex problem-solving. So, for me, what it's been looking like so far, and where I see it continuing to go, at least in the short-term, is that am I using faster? Absolutely. Is it making the work easier? Yes and no. It

Speaker A

certainly helps, but the job is still really hard. And even being very technically [music] proficient in using AI literally all day, every day, which is what I do, it still [music] takes just a high level of cognitive function.

Speaker A

So, with where this stuff is going, I mean, who knows where we're going to end up. I don't have a crystal ball. I can't guarantee where things are going, and at the speed at which AI is going, I have

Speaker A

no idea, and no one has any idea where we're going to be a year from now. But, with my experience in the field, I can tell you that for now, and I think for a while, data analysts are going to be

Speaker A

safe. I think we're seeing less junior roles because the work that juniors used to do can be handled more quickly by those at higher levels, but I say this all the time, the number of data analyst jobs on the market is not decreasing,

Speaker A

it's increasing. So, what I do think is that the role is going to continue changing [music] without a doubt. It's already changing.

Speaker A

It will continue to do so with AI augmentation, but I don't believe it's going away anytime soon. And something I get in the comments a lot is, you know, is it worth it to get into data analytics now if the job might go away 3

Speaker A

years from now, which no one can tell, but even if it did, if you got in now and you had 3 years of experience, you'll probably be in a spot where you can jump into a more advanced role

Speaker A

anyways. With that said, I think we're good. And when you're using tools like this and getting comfortable with these, this is huge. Companies are looking for people who are able to proficiently augment [music] AI into their work. And

Speaker A

if you bring this into the workplace, this is an advantage. Okay, so be proud about it. Be bold about it. Keep learning about it. So if you're trying to break into data or level up your career into a more senior level, that's

Speaker A

the whole reason this channel exists. I was a middle school teacher. I taught myself this stuff. I got the job years ago and now I do it for a Fortune 300 company. I try to make videos about what's going on today and what we're

Speaker A

actually doing in the field. I'm not a tech influencer who is completely removed from this stuff. I do it day in and day out. And tools like this and advancements like this are going to keep coming out every few months. I test them

Speaker A

so you don't have to waste a weekend on the ones that aren't worth knowing. So subscribe to keep up to date on technologies like this. So you have two options now. You can wait and see if this AI stuff blows over and if we are

Speaker A

indeed in a bubble, but that's probably not going to happen. Or you can spend 20 [music] minutes tonight, set this up and be the person on your team who already knows how it works. And if you're looking to get hired right now, watch my

Speaker A

video on the Power BI questions interviewers actually ask. Thanks again for watching and I'll see you in the next video.

Topics:Power BIClaude AIDAX measuresData analyticsAI in BIPower BI automationData model documentationAI workflowData analyst toolsBusiness intelligence

Get More with the SozAI App

Transcribe recordings, audio files, and YouTube videos — with AI summaries, speaker detection, and unlimited transcriptions.

App Store Google Play

Or transcribe another YouTube video here →

Free tools: TXT to SRT · SRT Validator · Merge SRT · Subtitle to Text · All tools