Latest / Microsoft Cloud IT Pro Podcast / Episode 435: Power BI Meets Git and MCP
Transcript
- 0:03- Welcome to episode 435
- 0:06- of the Microsoft Cloud IT Pro podcast recorded
- 0:09- live on 08/24/2026.
- 0:12- This is a show about about Microsoft three
- 0:14- sixty five and Azure from the perspective of
- 0:16- IT pros and end users, where we discuss
- 0:18- a topic or recent news and how it
- 0:20- relates to you. Today, Ben and Scott take
- 0:23- a lap around what's new in Power BI,
- 0:26- starting with the two MCP servers now in
- 0:28- preview, one local and one remote, that can
- 0:31- help you clean up your data model, write
- 0:33- decks for you, and turn a messy pile
- 0:36- of tables into an actual usable semantic model.
- 0:39- From there, we dive into how you can
- 0:41- start using Power BI
- 0:43- with GitHub,
- 0:45- DevOps, or other Git repositories
- 0:47- to help you manage updates, changes, and even
- 0:55- Another
- 0:57- sweltering
- 0:58- 90 degree day in Florida. Before the sun
- 1:00- even comes up, this is just not
- 1:03- I don't know why people decided Florida was
- 1:05- a good place to live.
- 1:08- Who knows? First place they landed and they
- 1:10- thought there was a fountain of youth here.
- 1:13- Yeah. It turns out it was not a
- 1:14- fountain of youth. It was a fountain of,
- 1:16- like, mosquito diseases, but
- 1:18- same difference. Anyways, we're not here to talk
- 1:20- about heat, humidity,
- 1:22- weather,
- 1:23- but all of that ties into data, which
- 1:25- we are gonna talk about. Well, how was
- 1:27- that? Yeah. You're getting so much better at
- 1:29- these. I'm trying, Scott.
- 1:32- Oh, and a trip down memory lane. Like,
- 1:34- this is a product that's been out for
- 1:36- a while, and I don't know that we've
- 1:37- talked about this recently. We haven't. We've talked
- 1:40- about well, we're gonna talk about Power BI
- 1:42- today. So we're gonna talk about some new
- 1:45- things that have come in Power BI. I
- 1:47- think you were introducing me to some things,
- 1:49- and I was introducing you to some things.
- 1:51- And it was just very interesting how much
- 1:54- it has
- 1:55- changed
- 1:55- over time.
- 1:57- And I think if you've been stuck in
- 1:58- your ways of using Power BI
- 2:02- in, like, one modality, like, there's a whole
- 2:04- bunch of new things that are out there
- 2:05- and available to you from
- 2:08- team collaboration
- 2:09- to MCP
- 2:11- servers,
- 2:12- like,
- 2:13- plural, that that can
- 2:14- help you out
- 2:15- and accelerate
- 2:17- your
- 2:18- pretty visualizations
- 2:19- in Power BI. Well, really, your, like, data
- 2:21- analysis, right, and and bringing things to bear.
- 2:24- Yeah. I would say just everything.
- 2:26- Like, for me, I don't spend a ton
- 2:27- of time in Power BI. Every once in
- 2:29- a while, it comes up or I'm working
- 2:30- on something or I want some reports, and
- 2:32- this kinda came up because I started playing
- 2:34- with it for a client that kinda wanted
- 2:36- a POC, and there are some things that
- 2:37- have definitely gotten easier. And then some things
- 2:39- that you were informing me about too that
- 2:41- I missed. So where should we start with
- 2:43- Power BI today? Why don't we start with
- 2:47- the MCP
- 2:48- servers
- 2:49- and kinda do a little trip around the
- 2:51- block there? Because there's some other stuff that
- 2:53- we wanna talk about later when it comes
- 2:54- to, like, collaboration and things like that. But
- 2:57- I think we talk about that later because
- 2:59- the MCP server, at least one of them,
- 3:02- can be super helpful for interacting
- 3:05- with that new mode for Power BI development.
- 3:08- So let's start with
- 3:10- the servers. Big overview, high level overview. There's
- 3:14- two MCP servers for Power BI. There's a
- 3:18- local and a remote. So you've got
- 3:22- different modes to think about when you were
- 3:25- interacting
- 3:26- with Power BI, the service,
- 3:29- versus
- 3:30- developing
- 3:31- Power BI through the Power BI app, and
- 3:34- maybe you're working with, like, PBX
- 3:36- files. You need help creating measures.
- 3:39- Maybe DAX isn't, like, your strong suit, but
- 3:42- you've run into it kinda thing if you're
- 3:44- just getting started with Power BI. So there's
- 3:46- a bunch of kinda niceties across both of
- 3:49- those servers. So maybe we'll start with the
- 3:52- local one, and then we can kinda work
- 3:54- our way towards the the service one. How's
- 3:56- that sound? Sounds good to me. And
- 3:58- the local one is the one that I
- 4:02- I think it was the local one that
- 4:03- I spent the most time playing with. Because
- 4:05- it was, to your point,
- 4:08- I had
- 4:09- a client
- 4:11- that wanted to start building some reports on
- 4:13- and it was actually QuickBooks data, QuickBooks online.
- 4:16- And Microsoft used to have this nice QuickBooks
- 4:19- Online data connector built right into Power BI,
- 4:22- and
- 4:23- they got rid of it. So I had
- 4:25- to go find other options and ended up
- 4:28- finding
- 4:29- a third party that I was able to
- 4:32- connect to the Power BI or the QuickBooks
- 4:34- online APIs,
- 4:36- dump content into
- 4:38- Data Lake, but then it was a lot
- 4:40- of data. Like and I will say QuickBooks
- 4:43- online from looking at the data is not
- 4:46- the most intuitive
- 4:48- in terms of the databases and how they
- 4:50- structure some of the data. And I very
- 4:52- quickly got overwhelmed. Again, I don't do Power
- 4:55- BI a ton. I don't do QuickBooks
- 4:58- Online data a ton. And I got to
- 5:00- this point where I'm like, I have, like,
- 5:01- 50 tables.
- 5:03- And
- 5:04- just a glimpse into it, like, this was
- 5:06- looking at projects in QuickBooks online. And projects
- 5:09- in QuickBooks online end up in your customer
- 5:12- table with, like, a flag that it's a
- 5:15- project and not actually a customer or a
- 5:18- project for a customer.
- 5:19- Like, super weird. And I was like, I
- 5:22- needed help. I was like, I don't know
- 5:24- how to sort through all this data. There
- 5:25- is a ton of data here. How can
- 5:27- I start digging through it? And I was
- 5:30- like, oh, I wonder if Power BI has
- 5:32- an MCP server. And lo and behold, here
- 5:35- is this local MCP server I could go
- 5:37- hook up. Yeah. The local one is super
- 5:39- super nifty. So for for those that haven't
- 5:43- played with it, you can go ahead and
- 5:45- wire this up to,
- 5:47- I I mean, a couple different things. So
- 5:48- it really just requires any harness that can
- 5:51- allow you to interact with a local MCP,
- 5:54- so
- 5:55- Claude on the desktop.
- 5:57- If you're I believe you have I believe
- 5:58- you still have to be licensed for for
- 6:00- those to come through. But Claude Code, GitHub
- 6:03- Copilot,
- 6:04- anti gravity,
- 6:06- codex,
- 6:07- things like that.
- 6:08- So you can just go out. There's instructions
- 6:11- on GitHub to wire it up. It was
- 6:13- a little bit of a rough start when
- 6:14- this first came out, but way easier to
- 6:16- kinda configure and wire up to everything that's
- 6:19- out there now. But
- 6:21- what the local MCP server does is
- 6:25- kind of cool. So if you're
- 6:27- sitting there and, say, you're working in
- 6:31- a PBYX file, so you've actually got Power
- 6:33- BI open,
- 6:34- and
- 6:35- you're sitting and you're working with an active
- 6:38- file, you can just go into
- 6:40- your
- 6:41- your LLM
- 6:42- interaction harness, whatever it happens to be, and
- 6:45- you can do things like, hey. Power BI
- 6:47- MCP server, connect to my local file.
- 6:50- And it knows that Power BI is open,
- 6:53- and it's able to go and interact with
- 6:56- that PBIX file. So you can do things
- 6:59- like you can say,
- 7:01- maybe
- 7:02- look at my schema and tell me anything
- 7:04- that I should fix with my relationships.
- 7:06- Like, Power BI can get a little weird
- 7:08- sometimes, right, when it comes to the whole
- 7:09- star schema thing and
- 7:12- one to one. And sometimes it just wants
- 7:14- to do many to many's, And you're, like,
- 7:16- sitting there fighting it. Like, what's wrong with
- 7:18- the model? What what's going on that I
- 7:20- just can't work with it? So it's super
- 7:22- good for things like that. Like, it's it
- 7:24- basically
- 7:25- has
- 7:26- inherent knowledge of Power BI modeling best practices
- 7:30- because it's kinda grounded and built on top
- 7:33- of all that. The other one that I
- 7:34- really liked with it
- 7:36- is I I can figure out Power Query
- 7:38- all day. Like, I'm used to it from
- 7:40- Excel and then in into Power BI, things
- 7:42- like that. I'm not good at all at
- 7:44- decks. Anytime I have to go look up
- 7:46- and especially get into, like, authoring, like, DAX,
- 7:50- I my my brain just doesn't work that
- 7:52- way. Like, it doesn't work for me. But
- 7:54- that Power BI MCP server, the local one,
- 7:57- it lets
- 7:58- you not only
- 7:59- validate DAX functions that are out there, but
- 8:02- it can help you
- 8:04- write DAX queries against your model. It can
- 8:07- do things like go write a bunch of
- 8:09- measures for you. So say say you've never
- 8:12- worked with measures
- 8:14- and they're confusing you for the DAX and
- 8:16- things that comes along, you can do something
- 8:18- like, hey. Go connect to my local file
- 8:21- and
- 8:22- suggest to me suggest what measures I should
- 8:25- have in this model. Or maybe you know,
- 8:27- like, kinda popular measures that you should have.
- 8:29- Maybe it's built, like you said, on, like,
- 8:31- Intuit data, like sales data,
- 8:34- invoices, things like that. So you'll know like,
- 8:36- you might want measures around, like, sales by
- 8:39- week, sales by month, sales by year kinda
- 8:41- thing. You just say, hey. Go create these
- 8:44- measures for me in this table or keep
- 8:47- the measures in the source tables, however you
- 8:49- wanna do it. And it will go and
- 8:52- spin those things up. Let's see what else.
- 8:55- It will troubleshoot calculations.
- 8:57- That that is super helpful.
- 8:59- If you've ever been working with, like, trying
- 9:01- to, like, round things out, convert,
- 9:03- say, from, like, a double to a percentage,
- 9:06- things like that. Super helpful for all that.
- 9:08- I have a public safety message going off.
- 9:10- Yep. You're getting on alert. I agree. And
- 9:13- part of what I was too was, like,
- 9:15- I needed to create a semantic model, and
- 9:17- I had to narrow down these 44 tables,
- 9:19- And that's where I started using it. I
- 9:22- had
- 9:23- a
- 9:24- I had a report, and it was interesting.
- 9:26- Like, I had a report in Power BI
- 9:28- that I wanted to replicate, and
- 9:30- I had all my data. So I was
- 9:32- like, no. I have all the data. I
- 9:34- have a report of what I want it
- 9:35- to look like, and I was able to
- 9:36- go feed it into
- 9:38- Claude, say, I have this report. Here's the
- 9:41- data. I need to essentially recreate this report.
- 9:45- Help walk me through it, and it was
- 9:47- able to go in and
- 9:50- help me generate
- 9:51- a semantic model and pull in the different
- 9:54- tables from
- 9:56- my
- 9:57- data lake and my lake house and all
- 9:59- of that and really narrow it down. So
- 10:01- at the end of the day, I had
- 10:01- to net 44 tables. It built
- 10:04- the, like, six I needed with the relationships
- 10:06- between all of them, and then it was
- 10:08- able to go in and here's the
- 10:11- and I actually told that I wanted to
- 10:13- understand it more. So I was like, just
- 10:14- give me the instructions. How do I take
- 10:16- this data and build my PBIX file? So
- 10:19- I didn't have it build the whole table
- 10:21- for or the whole report for me. I
- 10:23- just had it walk me step by step
- 10:25- of what did you what's in each table,
- 10:27- what table goes in which chart, what charts
- 10:29- should I use, how do I structure my
- 10:31- filters,
- 10:32- and actually almost had it teach me how
- 10:35- I wanted to take the
- 10:37- data and the models it had built and
- 10:40- build my PBIX report or the PBIX file,
- 10:43- build the report in Power BI so I
- 10:44- could understand it all. And then even some
- 10:47- notebooks that I needed running behind the scenes
- 10:50- to do some things from
- 10:52- data lake to lake houses and all of
- 10:53- that. So that's where I really found a
- 10:56- lot of benefit from that MCP server is,
- 11:01- hey, Scott. I love your puppy in the
- 11:02- background. Yeah. If everybody's here in squeakers, that
- 11:05- that that's it. He's No. Really going through
- 11:08- and helping me
- 11:09- go from all of that extra data that
- 11:12- I didn't necessarily need to just narrow it
- 11:14- down to the data I need and how
- 11:15- I wanna build the report and what notebooks
- 11:17- I need and all of that. Yeah. I
- 11:18- think it's a good way to go about
- 11:20- those things.
- 11:22- So you can use
- 11:23- I think there's two ways that you can
- 11:26- generally
- 11:28- use a AI and then certainly, like, these
- 11:30- MCP servers, things like that. So you can
- 11:32- either have them do the work for you,
- 11:34- or you can learn how to do the
- 11:35- work yourself and have this be kind of
- 11:37- an an aid or an accelerator for you.
- 11:41- And once you've made a decision, go in
- 11:43- a certain direction.
- 11:45- Sir,
- 11:46- so I think it is super helpful for
- 11:48- that iterative process
- 11:50- of coming through. And like I said, the
- 11:53- the MCP server is grounded in
- 11:56- things like the public docs. It understands best
- 11:59- practices.
- 12:00- It it knows what to do or ish
- 12:03- what to do. You might have to teach
- 12:05- it a little bit about your model. I've
- 12:06- been kind of impressed at what it's been
- 12:08- able to infer just by throwing some raw
- 12:10- tables and things at it. And Power BI
- 12:13- is generally pretty good about that anyway. If
- 12:14- I've got
- 12:16- sale date in this table and the field
- 12:18- date of sale in this table, sometimes it
- 12:20- can figure out that, hey. These two things
- 12:22- are in fact dates for the same tail.
- 12:24- May may
- 12:25- for the same sail. Maybe it was more
- 12:26- like a fact and a dimension,
- 12:28- things like that. So it can do definitely
- 12:31- do it and
- 12:32- and and get you there. So
- 12:34- so that's all good. So so you've got
- 12:36- a local MCP server. Author your files locally.
- 12:39- Work with them locally.
- 12:41- You get to be kinda human in the
- 12:42- loop and and go back and forth with
- 12:45- them. And then there's also
- 12:48- a remote MCP server. So
- 12:51- you can use that for
- 12:54- working
- 12:55- with your
- 12:56- semantic models that are already loaded up into
- 13:00- the Power BI service, things like that. So
- 13:03- if you're working, say, with, like I said,
- 13:05- like, a semantic model, it's already up there,
- 13:08- maybe you've got access to it, You wanna
- 13:10- go and maybe
- 13:11- create a new Power BI report dashboard, something
- 13:15- like that based on an existing semantic model.
- 13:18- It can go up there and do very
- 13:19- similar things. It can help you generate, execute
- 13:22- DAX queries,
- 13:24- natural language conversations,
- 13:26- all all that kind of stuff. The other
- 13:28- one where I think the remote MCP server
- 13:30- could help is
- 13:32- one of the things I've found
- 13:34- AI to be generally good at, I would
- 13:37- say doesn't tend to hallucinate nearly as much,
- 13:40- is just analyzing data. Looking over data and
- 13:43- saying, here's all my data. Like, help me
- 13:46- sort through it, sift through it, find certain
- 13:48- trends in it, that type of stuff. And
- 13:50- And I think that's where this remote MCP
- 13:52- server can could be nice too because like
- 13:54- you said, it can go connect to those
- 13:56- semantic models. So even if you're just trying
- 13:58- to understand your data a little bit more,
- 14:01- just using it to go query that data,
- 14:03- ask questions about your data, I found, not
- 14:06- to knock on Microsoft, but it's totally a
- 14:08- knock on Microsoft, is the Copilot stuff in
- 14:11- Power BI, I did try that first, and
- 14:14- it was trash. I could not get it
- 14:17- to work and do what I wanted to.
- 14:19- And when I switched to these MCP servers
- 14:21- for query tools and hooked it up to
- 14:24- Claude, I had a lot more success with
- 14:26- it. So if you've used
- 14:28- Copilot and Power BI and you're frustrated, I
- 14:31- would say go give this a try because
- 14:34- I definitely had a lot letter better luck
- 14:36- with this. And, again, the remote MCP server,
- 14:39- great for going in and looking through data.
- 14:41- I do wanna say too, Scott, both of
- 14:43- these are still in preview, so we need
- 14:44- to put our normal preview disclaimer on it.
- 14:46- These are both preview.
- 14:48- There's a comparison too between the road remote
- 14:50- MCP and local MCP that we'll link to
- 14:52- in the show notes. Right? Because there are
- 14:54- a few differences there in terms of, like,
- 14:56- what you can use for authentication.
- 14:59- It talks about the different tools that are
- 15:01- available, a few things there. If you are
- 15:03- trying to figure out, well, do I need
- 15:04- the local? Do I need the remote? Or
- 15:06- if one thing doesn't work in one of
- 15:08- them, what does the other one do, etcetera.
- 15:11- There are some additional steps along the way
- 15:13- as well. So
- 15:15- I I didn't run into some of this
- 15:16- until I read the documentation because you you
- 15:18- don't always, like, click all the buttons in
- 15:20- the ribbon and things like that in Power
- 15:22- BI. So there's this kinda
- 15:25- front matter or front loaded step preparation step
- 15:28- to go and prepare your data for AI.
- 15:31- So if you're working both with the local
- 15:34- MCP server or the remote one, Power BI
- 15:37- local at least has a button go and
- 15:38- prepare your data for AI,
- 15:41- and you can go and have it generate,
- 15:43- like, a AI ready schemas. So you have
- 15:46- to have Power BI q and a enabled
- 15:49- to be able to go and do this,
- 15:51- but you can just run through this set
- 15:53- of steps to kinda,
- 15:56- prepare your data for AI. It's what the
- 15:57- button says. It's what everything guides you into,
- 16:00- and it'll help you kinda optimize your schema
- 16:04- and optimize your semantic model for Copilot
- 16:07- or more broadly for LLMs
- 16:10- to be able to come in and understand
- 16:13- all of this stuff. Yeah.
- 16:16- Yeah. So I don't know if you've run
- 16:17- through that step yet. I don't think I
- 16:19- did run through that step. I'm gonna have
- 16:20- to go back and look at that. It
- 16:22- will do things like remember I was talking
- 16:24- about, like, help you with your schema, maybe
- 16:25- get it back to, like, a a real,
- 16:27- like, star schema kind of thing. So it
- 16:29- can help you simplify your schema.
- 16:31- It can remove redundant fields, redundant measures,
- 16:35- think things like that. And then part of
- 16:38- it is you're kinda, like, grounding it and
- 16:39- asking it along the or it's asking you
- 16:41- along the way, like, hey. I did this
- 16:42- summarization. Does this look right to you? And
- 16:44- you could say yes or no and and
- 16:46- go back and pivot. So it's another way
- 16:48- also to kinda not only clean up your
- 16:50- schema, but verify things like measure logic
- 16:54- and DAX,
- 16:55- and all of that is ready. And then
- 16:57- as part of that, you're gonna go through
- 16:58- and you're gonna mark things as
- 17:01- verified answers so that if somebody else is
- 17:03- using your model, they can also come in.
- 17:06- And you'll know that, like, consumers of your
- 17:08- model who are maybe using, like, Power BI,
- 17:10- q and a, or just Copilot, things like
- 17:12- that through through Versus Code, maybe they're using
- 17:14- the MCP server and hooking up, that they're
- 17:16- getting
- 17:18- the right
- 17:19- the right answer all along the way. I'm
- 17:21- glad to see these MCPs too. I think
- 17:24- I've talked about it before. I I
- 17:27- more and more as people create MCPs,
- 17:29- I think that's
- 17:31- you get so much flexibility
- 17:33- out of MCPs in terms of what you
- 17:35- wanna do with data. I would much rather
- 17:36- see a whole bunch of people build a
- 17:38- whole bunch of MCPs
- 17:39- than everybody build all their own AI stuff
- 17:42- and just let me get my data wherever
- 17:44- I want to. Let me choose my own
- 17:46- harness, choose my own AI platform, and give
- 17:48- me an MCP rather than, like, making me
- 17:50- stay in your tool or
- 17:53- your own specialized AI thing. I also get
- 17:56- it from a marketing perspective. Everybody wants your
- 17:58- eyes and their product, so they want you
- 18:00- to stay in theirs for their AI. The
- 18:02- more and more I can find MCPs,
- 18:04- it makes me happy. We could have a
- 18:05- whole another talk on We could have a
- 18:07- whole another talk on that. MCPs
- 18:09- and death of SaaS and things like that.
- 18:11- Should we move on from MCPs to something
- 18:13- else you told me about that I had
- 18:14- kinda completely missed when it came to Power
- 18:17- BI? Like I said, once you have these
- 18:18- MCPs
- 18:19- in place, now you've got kind of a
- 18:22- a a little bit of a collaboration
- 18:24- scenario going on, whether it's you as human
- 18:26- in the loop, just you and kind of
- 18:28- model and the Power BI service
- 18:30- going back and forth, things like that. But
- 18:33- there's also the world where maybe you collaborate
- 18:38- with with other users,
- 18:40- maybe other agents, things that are out there.
- 18:42- Right? You said talk kinda talking through, like,
- 18:44- hey. You enjoy being able to work with
- 18:46- kind of, like, model and harness of your
- 18:49- choice kind of thing. So so that's definitely
- 18:52- another consideration is now that you've got
- 18:56- these things out there, like, how can you
- 18:57- use them
- 18:59- with your existing harnesses,
- 19:01- existing agents, things like that?
- 19:03- So I forget when they added this. I
- 19:06- turned it on as soon as they did,
- 19:09- but there's an ability now in
- 19:11- Power BI
- 19:13- to
- 19:14- work on
- 19:15- collaborative files.
- 19:17- So if everybody's familiar with a p b
- 19:20- PBIX
- 19:21- file, the kinda just raw file that encapsulates
- 19:26- your Power BI report locally that you work
- 19:28- with before you go and publish it to
- 19:30- the service, or maybe you just hang with
- 19:32- it locally and you never publish it to
- 19:34- the service kinda thing, There there's this new
- 19:37- file format for Power BI,
- 19:40- which is
- 19:41- a,
- 19:42- it it basically
- 19:44- allows
- 19:45- it'll it allows you to collaborate on these
- 19:46- things, and it breaks down a what is
- 19:48- a single PVAX file
- 19:51- into
- 19:52- multiple files
- 19:54- that all end up in this
- 19:56- PBIP
- 19:58- file. I I do wonder what, like, like,
- 20:00- what they were thinking with the the the
- 20:02- acronyms on the end there. But, basically, what
- 20:05- a p b I b PBIP
- 20:08- fi IP file is
- 20:11- is it makes your,
- 20:15- it it makes
- 20:16- the Power BI
- 20:18- file
- 20:19- files,
- 20:20- basically,
- 20:21- human readable. So that now you can do
- 20:23- things like you can store this, say, in,
- 20:26- like, GitHub and collaborate back and forth with
- 20:28- people. So, like, rather than me writing a
- 20:31- PBX, and then I gotta shut down Power
- 20:33- BI and then maybe share you a link
- 20:34- to it, and then you go work in
- 20:35- it. And then the next time I open
- 20:37- Power BI, it's probably broken,
- 20:39- or we both open it at the same
- 20:40- time and we really screw each other up
- 20:42- kinda thing, now you can actually get into
- 20:44- a collaborative
- 20:47- workflow
- 20:48- across these things. There's no, like, there's no
- 20:50- more having to email these things around, storm
- 20:53- in a SharePoint site,
- 20:54- any anything like that. And because they're just
- 20:57- raw files, I said, you can storming in
- 20:59- a Git repository.
- 21:00- So now you get things like safe safe
- 21:03- rollback. You get actual, like, versioning versioning,
- 21:06- across things that are out there. But the
- 21:08- really kinda maybe, like, next level thing you
- 21:11- can do now is because you have these
- 21:13- MCP servers also understand that new file format.
- 21:16- So now you can have, like, agents that
- 21:18- sit alongside you.
- 21:20- So you could be, like, authoring one of
- 21:22- these things, and you could have an agent
- 21:23- say that, like, runs daily
- 21:25- and does, like, sanity checks on your DAX
- 21:27- queries or things like that for you. Yeah.
- 21:31- If you're managing file storage today, you've probably
- 21:34- felt the pressure. Refresh cycles are harder to
- 21:37- plan, costs and complexity are increasing, especially as
- 21:40- file workloads scale across regions.
- 21:43- That's why Microsoft customers trust Nasuni. As an
- 21:46- Azure certified software,
- 21:48- Nasuni provides a global file system designed for
- 21:51- enterprise scale environments,
- 21:53- keeping file data secure, governed, and accessible across
- 21:56- Azure environments without added infrastructure complexity.
- 22:00- It's the foundation to keep your unstructured data
- 22:03- stable even as your infrastructure
- 22:05- evolves. To see how much you could save
- 22:08- while keeping your initiatives on track, visit
- 22:10- nasuni.com/tco.
- 22:14- That's nasuni.com/tco.
- 22:20- This is cool, and I did look up
- 22:22- some dates, Scott. So this was
- 22:24- this has been a long time coming. It
- 22:26- was announced in May
- 22:28- 2023.
- 22:29- It hit public preview in June 2023, and
- 22:32- it has been in public preview for the
- 22:35- last three years. And it's still technically labeled
- 22:38- preview,
- 22:39- although
- 22:40- we do have for Microsoft that general availability
- 22:43- is planned. So while it's still preview, this
- 22:45- one is probably relatively
- 22:47- safe to start using.
- 22:49- Every once in a while, I had one
- 22:51- of these recently, Scott, a preview feature that
- 22:53- got deprecated and never made it to GA.
- 22:55- It does happen. It does.
- 22:57- And three three three years is a long
- 22:59- time to be in public preview. But, yeah,
- 23:02- I think I probably saw this, like, two
- 23:03- years ago or so maybe Okay. And started
- 23:05- playing around with it then. It I don't
- 23:07- I mean,
- 23:08- granted, I'm probably using it a way, like,
- 23:10- a high level, but it seems to work.
- 23:12- Like, it does what it says. It does
- 23:14- what it's supposed to do. Yeah. Yeah. It's
- 23:16- kinda cool. I almost like, like, when you
- 23:18- were starting to tell me about this, because
- 23:19- I haven't used it partly because
- 23:21- being in the position I am with consulting
- 23:24- and stuff, I don't tend to have a
- 23:26- ton of collaboration around Power BI stuff. But
- 23:29- it almost feels like when we made the
- 23:31- move from the doc files to the docx
- 23:33- files, where the doc files are kind of
- 23:35- all self encapsulated,
- 23:37- docx is I mean, really just a zip
- 23:39- file with a bunch of other files in
- 23:40- it so that you can do the coauthoring
- 23:42- and stuff in Word and all of that.
- 23:44- This feels very similar where it was like
- 23:46- PBIX
- 23:49- was that self contained PBIP, which is Power
- 23:51- BI project, Scott. That's where it comes from.
- 23:54- I also saw that.
- 23:56- It's the way you were describing it, it's
- 23:58- very similar to that where now it's almost
- 24:00- it's just a container with a bunch of
- 24:02- little files in it so that as you
- 24:03- are doing that coauthoring,
- 24:05- you're putting these in
- 24:07- source control.
- 24:08- It's able to go in and edit, like,
- 24:10- individual files and do diffs a little bit
- 24:12- better and all of that with that collaboration.
- 24:15- So that was kind of the analogy that
- 24:17- jumped out in my head. I'm like, okay.
- 24:18- Now it makes sense what those Power BI
- 24:21- PBIP
- 24:22- Power BI project files are. There's still some
- 24:24- proprietary stuff in there. So if you start
- 24:26- digging into the folders and things like that,
- 24:29- there there's definitely parts that are kinda just
- 24:32- custom binary format and not meant for you
- 24:36- to use, but
- 24:37- you yeah. But you you can definitely
- 24:40- get in there and start playing with these.
- 24:42- Super easy to get going with. They do
- 24:44- recommend,
- 24:46- and I don't know, I haven't used it
- 24:47- with anything else other than, like, GitHub, but
- 24:49- just Git repos. But you could do this
- 24:51- with, like, a Git style repo in Azure
- 24:52- DevOps,
- 24:54- anything like that along the way. So now
- 24:56- if you think, like, back to, like, say,
- 24:59- like, the agents thing.
- 25:00- So you have one of these
- 25:03- PBIP
- 25:04- projects.
- 25:05- You spin that up. When you do your
- 25:07- check-in let's say you are using Azure DevOps.
- 25:10- So you do your check-in. You can have
- 25:11- a pipeline run that goes and kinda validates
- 25:14- things for you. And you could potentially have
- 25:16- that pipeline
- 25:17- run and invoke something like an MCP to
- 25:20- go validate your schema, validate your DAX. Like,
- 25:23- you actually get a little bit more
- 25:25- a little bit more kinda, like, rigor and
- 25:27- governance, I guess, around release of these things
- 25:30- as they as they come out, which is
- 25:33- super nice to see. Right? It kinda makes
- 25:35- it a little bit more, I think, a
- 25:36- little bit more, like, enterprise developer
- 25:39- versus, like, enterprise power user kinda thing. I
- 25:42- agree. Power BI,
- 25:43- it did seem to be very end user
- 25:45- focused
- 25:46- originally where it was go install it on
- 25:48- my desktop, create it there, I publish it
- 25:50- up, I share it out to people. This
- 25:52- is very much that collaborative development. Like, you
- 25:55- were talking about, I could even think if
- 25:57- you had, like, scheduled agents
- 25:58- that were running your scheduled tasks in your
- 26:01- LLM that had to go in and do
- 26:02- certain things in your Power BI report, publish
- 26:05- it through a pipeline.
- 26:07- I feel like it this just greatly expands
- 26:11- the flexibility
- 26:12- and the flexibility
- 26:14- ability
- 26:15- aspects of what you can now start doing
- 26:18- with some of these Power BI reports, especially
- 26:20- because people always like them to be up
- 26:21- to date. And let's face it, when people
- 26:23- build reports on data like this, they kinda
- 26:26- want them to be accurate. So if you
- 26:27- can use some of these tools as well,
- 26:29- to your point, help ensure the accuracy, help
- 26:32- ensure the make sure everything is
- 26:35- working properly, working the way it should,
- 26:38- I think there's a lot of lot of
- 26:40- options here that you start getting. Or just
- 26:42- the fact that now you can work on
- 26:43- a report with your best friend, Scott. You
- 26:45- and I can go work on reports together.
- 26:47- Best friend, your best AI agent, all all
- 26:50- those things. Yeah. All the things. I will
- 26:51- say one of the I guess
- 26:54- and I I wonder for being in preview
- 26:57- for three years now, like, why this gap
- 26:58- hasn't been filled more by Microsoft.
- 27:01- So if you go and you read the
- 27:02- documentation for this, it kinda
- 27:05- stops
- 27:06- at initialize a Git repo. So it tells
- 27:09- you how to enable this. It tells you
- 27:12- how to
- 27:13- create the new folder structure format thing
- 27:17- and get you spun up and kinda how
- 27:19- to do
- 27:20- an initial load into a Git repo. But
- 27:23- it leaves
- 27:25- all the follow on stuff, and I think
- 27:26- the power of it, the the versioning,
- 27:28- CICD
- 27:30- capabilities, things like that, just as an exercise
- 27:33- to the reader. So you will have to
- 27:36- go out and Google Bing, DuckDuckGo,
- 27:39- Kage, whatever your search engine of choice is.
- 27:41- If you're looking for, like, specific guidance on
- 27:45- how to configure, say, like, a CICD
- 27:47- pipeline
- 27:48- or some GitHub actions and wire them up
- 27:51- around this, you're gonna have to go, like,
- 27:53- find that stuff and hope somebody blogged about
- 27:55- it or figure it out and blog about
- 27:56- it yourself because that guidance doesn't exist from
- 27:59- Microsoft. Yeah. I was looking through this, and
- 28:02- it is. Like, that ends up with, here
- 28:04- you go. Initialize your git repo. And then
- 28:06- I was looking at one of the other
- 28:08- ones, editing Power BI files externally, and it
- 28:11- talks about how you can open them in
- 28:13- Visual Studio Code. And it says, here's the
- 28:15- button. Just go open it in Visual Studio
- 28:16- Code. It doesn't really give you, like okay.
- 28:18- So once it's open in Visual Studio Code,
- 28:22- what am I supposed
- 28:23- to do with it? I now get access
- 28:26- to the report and semantic model definition files,
- 28:28- but what do I do from there? We'll
- 28:30- say it also does call out to your
- 28:32- earlier point, Scott, under some of the limitations
- 28:34- is
- 28:35- not all of those project files that you
- 28:37- may see open up in Visual Studio Code
- 28:40- or within this file
- 28:41- support
- 28:42- editing. It's not saying the blocks editing. Like,
- 28:45- there's some JSON files in there, a report,
- 28:48- a mobile state,
- 28:50- semantic model diagram layout
- 28:52- where it's JSON. You could edit it in
- 28:54- any text editor,
- 28:55- but you might break it here because that's
- 28:58- not supported in this. So there are other
- 29:01- areas of the documentation
- 29:03- too that I would go read before you
- 29:05- just start editing some of those text based
- 29:08- files in there. I think if you're actually
- 29:09- gonna do it, like, in a collaborative way
- 29:11- too, like, say, you and I were kinda
- 29:13- going back and forth on something, you do
- 29:15- need to decide upfront
- 29:17- about
- 29:19- kinda how you're gonna approach
- 29:22- isolation for, like, your changes versus my changes,
- 29:25- things like that. And, again, this is an
- 29:27- area that, like, Microsoft doesn't have, like, discrete
- 29:30- guidance for you or prescriptive guidance, but I
- 29:32- think there's plenty out there in the community.
- 29:33- So you can do things like
- 29:36- you and I could be working
- 29:39- in the same same report, same dashboard,
- 29:42- but we could have isolated
- 29:44- workspaces,
- 29:45- effectively, like, a dev workspace for you, a
- 29:48- dev workspace for me. Like, if we were
- 29:50- working in, say, like, a a a Git
- 29:52- repo,
- 29:53- we'd wanna do things, like, probably, like, a
- 29:56- feature branch per workspace. So, like, I'm gonna
- 29:58- go spin up like, I'm gonna go work
- 30:00- on the
- 30:01- new DAX measures for for blah blah blah.
- 30:03- You're gonna go work on the visualizations
- 30:05- for report page
- 30:06- a, b, c kinda thing. So you you
- 30:09- do end up into, I think, more, like,
- 30:12- the geeky, like, Git semantics
- 30:14- that are out there. And, again, that's where
- 30:16- I'm more, like, enterprise developer than enterprise power
- 30:18- user
- 30:19- if you do go down this path. But
- 30:22- there's definitely, like, some nicety there, I think,
- 30:24- even if you're just, like, a solo person.
- 30:26- Right? You wanna go and have kind of
- 30:27- that, like, go forward, go back, replay logic
- 30:31- that you get with things like a with,
- 30:34- like, a Git repo.
- 30:35- Could be worth learning even for, like, the
- 30:37- solo Power BI developer out there. Yeah. Because
- 30:40- even though this is collaborative, I'm assuming
- 30:43- because it is Git, you're still doing commits
- 30:46- and branches and all of that. And you
- 30:48- could still
- 30:49- very much end up with a conflict if
- 30:51- you and I both decide to go edit
- 30:53- the same thing and change the same thing
- 30:55- in the same file at the same time
- 30:57- where then you have to decide which change
- 30:59- is better, my change or Scott's change? Yours
- 31:01- is always gonna be better.
- 31:03- It's definitely something worth looking at if you're
- 31:06- in Power BI a lot. If you and
- 31:08- coworkers are working on Power BI reports and
- 31:10- you wanna take some
- 31:12- or use some of that functionality,
- 31:14- this is something I will probably do as
- 31:16- well just from a I lose Power BI
- 31:19- the PBI. I keep saying Power BI. I
- 31:21- lose PBIX files, Scott, all the time. I'm
- 31:25- like, where did I create it? Which desktop
- 31:27- documents folder did I put it in?
- 31:30- I for whatever reason, because you have to
- 31:33- save them locally and then publish Milan, I'm
- 31:35- always losing them. I'm like, you know what?
- 31:37- If I just started doing this and had
- 31:39- a git repo with all my Power BI
- 31:41- files or a git repo per client where
- 31:43- I'm creating these PBIP files,
- 31:46- it would just greatly help even to your
- 31:48- point as a solo person with
- 31:51- keeping track of my different Power BI reports
- 31:54- and where kind of that source
- 31:57- report source file is. Move them all into
- 31:59- PBIP and GitHub.
- 32:01- I never even tried to convert one. Like,
- 32:05- can you convert them? I think you can.
- 32:07- I think I saw the documentation in there.
- 32:09- I always just start in,
- 32:11- start in a modality and go from there.
- 32:13- PbIP enable PB. Save as a project. If
- 32:15- you're working on a new project or you
- 32:17- opened an existing
- 32:19- Power BI Desktop PBIX file, you can save
- 32:21- your work as a Power BI project file.
- 32:24- So you just do, like, the file save
- 32:26- as, and then in the drop down for
- 32:28- the save as type, you can switch it
- 32:30- to PBIP. That's an interesting so so, like,
- 32:33- like, these tools are so all over the
- 32:35- place. Like, I don't think I ever use
- 32:37- the save as dialogue
- 32:39- in Power BI once I initially create a
- 32:41- a a project and I go and publish
- 32:43- it. Once you publish it, it just asks
- 32:45- you to save again. Like, do you wanna
- 32:46- save this before you publish it to the
- 32:48- service? Yeah. I wanna save it. So I
- 32:51- would never see the save as dialogue.
- 32:54- Now you can go use the save as.
- 32:56- Because I did see I think somewhere in
- 32:58- this documentation,
- 32:59- it does say that PBIP is going to
- 33:03- I'm assuming once it hits GA, turn into
- 33:05- the default file type. I would imagine. Right?
- 33:08- Like, that that would be kinda like the
- 33:09- happy path for these things to go on
- 33:12- is
- 33:13- let it just be the default state. And
- 33:14- it really doesn't matter at the end of
- 33:16- the day. Like, it's gonna be a set
- 33:17- of files and folders locally, or it's gonna
- 33:19- be a monolithic thing.
- 33:21- Arguably better, just have it be the files
- 33:22- and folders locally, and then if you wanna
- 33:24- push to Git and be done with it.
- 33:25- Exactly. Just like we did with I'm gonna
- 33:27- go back to my office analogy. Just like
- 33:29- we did where doc and XLS and PPT
- 33:32- and the other ones all switched. They just
- 33:35- added an x there. Not to confuse you
- 33:38- because in Power BI, it had an x
- 33:40- and then it went to a p.
- 33:42- Naming, even when it comes to file types,
- 33:45- can be hard. Such is life. Yep. Alright,
- 33:47- Scott. With that, we are coming up on
- 33:49- our We got some new MCP servers, new
- 33:51- stuff for folks to go play within Power
- 33:54- BI. And we have meetings coming up that
- 33:56- we should probably go to keep people happy.
- 33:58- I do. I don't know if you do.
- 34:00- I have a meeting coming up. I do.
- 34:01- I'm gonna roll from this one just into
- 34:03- my next, and
- 34:04- and then let's go. We'll have to come
- 34:06- back sometime and revisit the whole LLM Wiki
- 34:08- thing because, like, I went down your path
- 34:10- and built, like, a whole dashboard for mine,
- 34:12- and I went up, like, a little I
- 34:14- went a little wild on the side. But
- 34:16- Do you have any tokens left? Yes. I
- 34:18- have tokens left. Yeah. We will because I
- 34:20- keep expanding. I keep trying to
- 34:24- like, I need to actually get work done
- 34:26- instead of playing with what can I build
- 34:28- with AI and my LLM Wiki and all
- 34:30- this other stuff? I tend to get a
- 34:31- little distracted with that. So we will come
- 34:34- back to that. But for those of you
- 34:36- listening, if you do
- 34:39- play with these, have any other thoughts, or
- 34:43- anything about Power BI and MCP servers and
- 34:46- Power BI or PBIP.
- 34:47- Would love
- 34:48- to hear from you. LinkedIn
- 34:51- or Reddit? We've had some more conversations lately
- 34:53- on Reddit as well. So I would say
- 34:56- those are probably the
- 34:59- two social media
- 35:01- platforms you and I are both the most
- 35:03- active on is either LinkedIn or Certainly for
- 35:05- the podcast. That's a great place to find
- 35:08- us. Yep. Join us over there. And it's
- 35:10- been really good to have the feedback over
- 35:11- on Reddit. So folks should keep that coming.
- 35:14- For sure. And if you are gonna be
- 35:15- at TechCon three sixty five in Seattle this
- 35:17- week, this will come out Thursday morning. I
- 35:19- will be there Thursday and Friday. So if
- 35:21- you're there and you wanna say hi, come
- 35:23- say hi to me. Scott won't be there.
- 35:25- Sad face. I will not be at TechCon.
- 35:27- Yeah.
- 35:28- Outside my wheelhouse these days. Yeah. Not so
- 35:31- much Azure storage blob stuff there. But for
- 35:34- now, go enjoy your Monday, Scott. Get some
- 35:36- work done. Enjoy the puppy, and we will
- 35:38- talk to you again soon. Thanks, Ben.
- 35:42- If you enjoyed the podcast, go leave us
- 35:44- a five star rating in iTunes. It helps
- 35:46- to get the word out so more IT
- 35:48- pros can learn about Office three sixty five
- 35:50- and Azure.
- 35:51- If you have any questions you want us
- 35:53- to address on the show or feedback about
- 35:55- the show, feel free to reach out via
- 35:58- our website, Twitter, or Facebook.
- 36:00- Thanks again for listening, and have a great
- 36:02- day.