← Think AI Podcast
Loading summary
Transcript54 lines
- [00:01]
A
AI agents, LLM, Copilot, ChatGPT, Cloud, all of these things are kind of replacing workforce and that is a big concern for many people. But who does it replace? It actually replacing junior workforce. So I would advise people to go on that route instead of saying well I don't need to learn this language so I would re engineer. So at some point chatgpt would replace me. No, that is not the route I would say suggest people to go.
- [00:34]
B
Welcome to the Think AI Podcast. Each week we talk about the most exciting AI research tools, case studies and more. I'm your host Dev Goyer and I've been working behind the scene in data and AI for over 30 years. Whether you are an AI expert, skeptic or something in between, this podcast is for you. Today's guest is someone I followed for years and personally learned Fabric from when he came through LA two years ago, I think. Reza Red is the founder of RedAcad, Microsoft regional director, a 15 time MVP for data platform and AI, and the author of seven books on Power BI and SQL Server. He's based in Auckland, New Zealand and his training reaches practitioners around the world. What I wanted to talk to you today is the long arc Racha has taught through SQL Server, through Power BI and now through Fabric and AI. So we are going to ask what actually different this time, what has not changed and what should a leader with a Power BI estate already running to do on Monday morning? Reza, welcome to the show.
- [01:48]
A
Thank you. Thanks Dave. Thanks for the opportunity. Hello everyone. Really glad to be with you here. So my name is Reza, they've already introduced me. So back to your questions. That with the new era of AI and things like that world has actually not changed. I think the processes that we use, the people, the culture, those are not changed. Technology changes. Technology evolves as we go. We used to do things in SQL Server using the stored procedures, then we did things in ssis for transforming data, then data flow came in. These are all different technologies, but still there's a need for transforming the data, there's a need for integrating the data, there's a need for us to connect to a data source, get the data, prepare it in the shape that we want to analyze. And that is still part of the plan. Even with the AI, AI might speed up that process. Like we tell AI that this is what we want so that it speeds up the process, but still that part of the work needed to be done. So I think once we look at it from that point of view of the processes, we still have the same Processes as in the old days. Although we can do that more efficiently these days.
- [03:11]
B
Yeah, very well said. And we also saw the same thing in manufacturing and finance industry things stay stable at the same time there are a lot of innovations come through but the fundamentals are same, isn't it? Like SSIS may have transferred to ADF to fabric pipeline. The concept is still the same. You're still doing etl, you're still applying the transformation logic, just doing it differently. One thing I also wanted to see, you know with Microsoft there is a love and hate relationship. I'll be honest. People love Microsoft but at the same time they complain about certain things. With this fepcon this year I saw something amazing that Microsoft has done like a complete platform as we used to see in MicroStrategy Cognos back in our days 30 years ago. Now things have been pretty stable with Microsoft. What do you see? Customers or even people complain about Microsoft technology and how Microsoft or us should be marketing it or explaining it in a correct way. Because there are a lot of complaints which may not be the complaint, if that makes sense.
- [04:21]
A
Yeah, that is right. So one of the things about Microsoft roles in the market is that Microsoft is a big market player. Like I've been in the data and analytics, let's say market for like over 20 years. I've worked with like different technologies like IBM Cognos back in the days with Oracle. Not that many people are using it these days but still we used at those times, SQL Server, MySQL, all these different technologies that different companies provided, different vendors provided. And alongside with this you also have big players. You have Microsoft to have like other players in the market these days. We have a Google BigQuery, we have Snowflake, we have data breaks. The thing is that when you are a big player in the market you try to expand your wheels. And that was what Microsoft is doing. So Microsoft started the new analytics aura with Power bi. Before that Microsoft had a lot of technologies for analyzing the data like SQL on PREM analysis services, reporting services. But the one turning point I would say was Power bi. Power BI is a technology that many people are using these days even though they don't know how to go and build a database. It's a data analysis tool. It's a self service data analysis tool. Then from Power Bi Microsoft expanded because Microsoft had these other tools and services such as Azure Synapse, such as Azure SQL Database, Azure Data Factory. So I tried to expand it to a bigger feature, a bigger umbrella called IT Fabric which is a Natur control progress from Power BI when you look at it now, when you look at it from someone's point of view outside of this environment, they say, well, if I just compare, for example, synapse versus Databricks, I might get a better performance in databricks. If I just compare, for example, the visualization of Power bi, this particular visualization with a tableau, I might get a better visual experience in that. But then when you look at it from like a high level point of view, when you look at it as an umbrella product, instead of playing with a lot of different methodologies, dealing with a lot of different vendors and licenses, you could get one environment, one licensing platform, one capacity structure that would cover everything. Always. There will be always pitfalls and there will be always bugs to fix. But one thing that I would encourage people to look into is not to look for technology that doesn't have any bugs. We don't have such a technology. There is always a bug in the technology. But look into if the team behind that technology, if the product team behind that technology is taking time to improve that technology, is investing time to hear what the customers are saying, what their users are saying, and improve their products. And what we have seen from Microsoft, this is the case. They have the ideal form, they have the customer works, they have different channels that they get these inputs and that impact the way that they improve the products, which is also showing like the fabric that we have now is totally different than the fabric we had two years ago. It's much more reliable and it will go this way in the future.
- [07:53]
B
Very well said. And yeah, I think we used to say work with limitations, not work without limitations because there's nothing such thing out there and available. So you very well explained it. So I want to pull on the Power BI side again. And one of the things Microsoft started doing is launching Copilot and everything. So Power BI obviously have copilot and now it's matured and now it's going towards the agents. Now agents are arrived in fabric and also in Power bi. Where do you see customer actually seeing it wrong on the ground? Not in theory. So they may be thinking, okay, agent can do this, but agent might be for something else. That reality versus the dream. Where should they keep their feet right now?
- [08:42]
A
Yeah, so what I see that a lot of organizations are thinking at the moment, or that is their, let's say, idea of copilot in Power BI or agents using Power BI is that we ask the question, this gives us the whole response like instantly what we're saying is that this still relies on the good data model underneath the good data model is following all the rules that we define what the good data model is like. It is a star schema. It has fact Table Dimensions, Table 1 To many relationships between those. It has a rollover security setup. It has all of those configurations that we want this to become a good data model. On top of that data model, we could go and write our DAX calculation. If we don't have the right data model, our DAX calculation would be heavy, it would be low performance even if Copilot writes it right. But if I have a good data model, either me or Copilot go and write that calculation. Either me or Copilot go and build the visualization and that can be a good thing. So the agent use in power BI as of now, this of course would change in two years time. But as of now, the agent used in power BI is good when it comes to building something, something on top of a good data model. But you as a developer, you as a BI team data analytics team data analyst, you will still need to spend time to build that data model in a good way to make sure that your relationships are well defined. You have good scenarios for the agent to understand that. So at the moment you are building it to serve the agent and then later on you expect agent to serve you to give you better visualization. In the future, two, three years, who know this might entirely go and build a power BI model. But we are not there yet.
- [10:39]
B
No. Very well said. And we had a similar story. So we had a CFO ask copilot what was the Q3 margin and it returned the right number, but the stale data set nobody refreshed. So now the number which was right in his mind wasn't right. And the story behind data before AI and also the calculations have changed because they've acquired different businesses and nobody pays attention to that. So they think AI can bring a magic wand and start building the calculation. It could, like you're saying. But I would think the data will always remain the foundation. The thinking behind how the data needs to be modeled, what the calculation needs to be, the human touch, the human control needs to be there. Copilot is only going to expand or explain the wrong thing in a bigger and better way.
- [11:29]
A
That's right, yeah. Copilot would emphasize what you have in your model. If you have a good model, you get good output. If you have a bad model, you probably would get even worse output.
- [11:42]
B
The other thing you just touched upon is dex calculations and dexnm you know, I remember seeing Chris Webb and others from Europe and then you writing and teaching on DEXNM over a decade. I would say Copilot can now write a whole lot. But then there were like these best practices that's been taught through. So what does a power BI consultant or a BI developer actually spend their day? I've got a lot of juniors and they say, why do I need to learn the concepts of dax? I can just give a command to copilot and I don't have a good answer. So I'm pushing to you what would be the good answer for me to give it to them.
- [12:23]
A
Yeah. So the thing is, this is not just DAX or M, this is like everywhere in programming. Like if I'm a developer, why should I go and learn C Copilot and write it for me? Why should I go and learn Node js, what should I go and learn JavaScript, all of these things. But what I say is that of course Copilot can goes and write these for you and it actually improves. The copilot we use today would be much better in your time, it would even write better calls. But still this needs to be vetted. This needs to be checked by someone that is it correct or not? Now you may not go and check every single measure DAX measure that this creates for you, but if it happens that that DAX calculation that it generates for you has a low performance, then you need to go and check it. It doesn't work with just telling Copilot that well, this is low performance, go and write it another way. You need to go and look up the code and see for example, this use this invo function which is probably performing slower than or can't be part of that function, which might much better. How do you know that? By knowing the language. By knowing the functions, the learning are going to be different than the old days. You may not want to write the whole program yourself, you may not go and write the whole DAX and m, C sharp, JavaScript, all of that yourself. You get Copilot to write most of this for you, but you have the capability in case it is needed to go and code know that this is not the right way to do it and change it. And that is only happening when you know the language very well.
- [14:09]
B
No, very true. And we also talk to our consultants now that you need to know the subject area a lot more than before. You could get by just being a technologist before, but now because technology is acting on your behalf, you need to come in front and ask the right questions, do the right testing, give the right architecture, give the right advice to AI so that AI can perform better for you and validate to that level. And that shift is, I think we are in that transition stage, especially I work with India, a huge team, they need to, I'm sure you're dealing with the same thing. They need to have this mind shift towards, you know, having more thoughtful thinking, solution oriented structure than just a technology oriented structure, isn't it?
- [14:59]
A
Correct, Correct. What, what I think will happen, and it is already happening in many organizations is that AI agents, LLM, Copilot, ChatGPT, Cloud, all of these things are kind of replacing workforce and that is a big concern for many people. But who does it replace? It actually replacing junior workforce because like for example the junior workforce that wrote this media called code, now Copilot can write probably the same code or might be even writing it better. The workforce that this cannot replace at this stage is someone who has a good technical understanding, is someone who has good knowledge, more like a senior. So I would advise people to go on that route instead of saying, well, I don't need to learn this language. So I would renew engineer. So at some point chatgpt would replace me. No, that is not the route I would suggest people to go.
- [16:02]
B
Awesome. No, that's a great advice. I also want to switch gear a little bit. So we talked a lot about how to use the technology, where to use it, what to do and what not to do. But now the technology is out there, especially fabric, which I'm big fond of. Do you have any examples like one or two real customer problems that fabric is solving today? Using AI or power bi AI in last six months maybe?
- [16:29]
A
Yeah, we do have actually both. So in general we have customers who we had actually customers who been previously working with some other technologies. Combination of the costs for licensing, they paid for those now has improved a lot that we migrated them to fabric that is just like pure fabric, not even using AI. Now with the use of AI we have a little bit different story these days, the use of AI. So the general expectation is that this should speed up the process. But the fact is that nowadays we are not at that stage yet. Nowadays we have to prepare our data well for the AI to consume. So we actually spend more time in building our model. So let's say the semantic model that I used to create, for example previously in two years time, now I might even spend one week more like make it in four weeks time because I want to add all the synonyms, I want to make sure that this has all the logics that the AI can leverage later. So when you look at it from a customer point of view, their short term view might be that this cost them more. But the long term view, like after some months they get to see the results. They get to see the fact that now that we have built this model, now that we have prepared this data all in fabric, all in Power bi, OI can just sit on top of it and give them really efficient visualization reports from what we already have. So they save a lot of budget, a lot of resources on that side.
- [18:03]
B
It any KPI's you had with your customer? No, no need to name the customer, but just the KPIs like 3x faster or from days to minutes or something like that.
- [18:14]
A
Nothing specifically that we have measured it. But I remember like one of the projects that we have had this kind of things was not necessarily implemented using fabric because this was a little bit older days. We used some analytical engine and AI engine with machine learning. We had a customer who have been using a lot of like pre flight booking and they've been looking for finding out what is the right window to do the pre flight booking. And they realized that the right window for the pre flight booking is actually is the window. Based on this decision, three algorithms calculated that this is for example like six, seven weeks before the ticket time for international flight and domestic flight was different and the result was different. In a six month flight they saved pretty much like half a million dollars just for that particular change in the time that they make their flight booking.
- [19:20]
B
No, that's a great example. In fact we work a lot with manufacturing and we worked with one client. In fact we just completed the project where we were dealing with manual POS and email or printed and we started doing with the scanning OCR with Power, Automate, AI builders, things like that. And also using fabric on top to use the intelligence of data from their erp. And you taught us real time AI or bi. I wrote a book on it too. But this was near real time AI, right? So they were able to get this multi day approval into a near real time approval cycle and they were very happy about it. So those are some of the good examples that you have mentioned. I also want to touch upon how do you tell your customer whether AI investment in Power, BI or fabric is actually working? How do you balance speed against the governance? So we talk about F8, F16, F64 and most of our customers are talking about it, especially speed against, you know what they're investing but also the governance. So governance plays a big role over there as well. What do you have to say on that?
- [20:39]
A
Well, of course I would first like whenever a customer comes to us asking about utilizing AI in their environment, first I would challenge them to think about how much they are going to invest in the data underneath AI because like if we consider AI like tip of the iceberg, underneath that is all the data that we have prepared. That data needs a lot of work to be prepared for the AI to consume. Not only that, we need clean tables, clean data warehouse, things like that. We also need to have clean processes. We need to make sure that the workspace design is correct, we need to make sure that we have the medallion architecture, we have the enablement of self service users. We have all of these processes defined because AI is just one part of all of these aspects. If I have business logics defined by the BI team, but there is no business logic input in that, that's not going to work. How would I add my business logic by enabling self service users? Self service users. If I just list 100% self service I wouldn't have governance. So I have to find that sweet spot between so that I have good governance alongside with the self service. Now when you consider all of these, if we start with looking at y' all project from this point of view that let's first get all the foundation right, let's get all the architecture right, let's get the model right and everything else. Then investment on AI makes sense, then investment on AI I would say the more the merrier, right? We would get the better outcome of this. But when I have a potential customer coming in and saying well we want to spend much on AI but we don't want to go and build data warehouse, we don't want to go and do data integration, we don't want to go and work on the process that will bring data from other places, things like that that usually doesn't end well, like it might generate a couple of reports, a couple of dashboards for them that is just good for now for like a short period of time. But after a while they realized that this is just a report on top of a chaos of a data that unless we go and fix that chaos of data, it doesn't really lead to a good outcome.
- [23:06]
B
Now that's a great example. We have a parallel story. Most clients we work with obsess with speed so they can't articulate the success metric, but they can say oh, what does the success look like for 90 days and we put that question back to them. I said what does this look like for you in next 90 days so that we can support technology merely follows what your success looks like. If you don't have a metric then your project will drift. So something like that is crucial. If you don't have a definition of what is a definition of done, then you are always going and the speed, no matter what you have is not going to work out for you.
- [23:43]
A
You.
- [23:44]
B
That's right. Okay. I want to touch bases back onto your passion which is teaching. I know you have been doing teaching for 20 plus years. Amazing. I see your YouTube hits. I think you recently reached 10 million views which is phenomenal. So congratulations.
- [24:04]
A
Thank you.
- [24:05]
B
And you have survived every platform turn right from SQL Server to ssis to Power BI to now Fabric. And this changes every and it changes everything cycle as well. And life has thrown plenty of you at outside of work too. So I'm not going to go into the details but I'm mainly talking about mindset. What's the practice of mindset that keeps you teaching, shipping, meaning building products, ensuring with this positivity. I've always seen you very positive person despite of everything that I know about you. So what do you have to say to someone listening who is in a really tough chapter in their life right now?
- [24:50]
A
Yeah. So I would say, well, life happens. We have circumstances that does not always go in our favor. We have good days and bad days. Good days are not. Are not hard. Like when I have a good day, I'm productive, I go out with my friends, I go and build this piece of code, I write this code, I write this application and I am in a good mood. The main thing is not how do we handle our bad days. What I found myself that is working is that there are some tasks or activities during the day that I would consider them as the tasks and activities that would drain my energy. And then there are some tasks and activities that would give me energy. Right. If I do that, like for example, if I go for a hike, although this would like drain my physical energy, but it would give me a good mental energy to do something. If I go for work with my friends, if I go and play pickleball or tennis with someone, these are giving me energy. If I go and write a code that also gives me energy. For a lot of people this might not. If I go and work with particular technology, if I go and work with cloud and copilot to go and do application that gives me energy. But if I go, for example go and do some accounting. I'll be like that, right? That drain my entire energy. So then you have to find the right balance of doing these things in a way that you don't have in one day all your energy draining activity. Right. If I start my day with things that I don't like and then do two, three of those, then by midday I wouldn't have energy for many the things that I really like to do. I like working, working with technology. This technology could be like back in the days as a science now power bi fabric AI. I still love to work with technology. The other thing that I like is to teach others about what I learned. I started this like 15, 20 years ago with just a blog for myself. I wrote that blog just for myself to keep diary of. For example, I did this in SQL Server. And then later on I found people that would get benefit from this blog, come and read the blog and they realize that this is helpful for them. So this became more like a source for also teaching others. So that way I can I get some feedback from them that this is working for them. This is not working for them in a situation that I haven't worked before. Right. Because they are working in different environments. On the other hand side this, this fight that I'm helping others gives me energy. So one of the things that encouraged me to do a lot of these like blogging video videos on YouTube, going to different conferences, presenting training, all of those is the fact that I actually like to help people. But the way that I help people is different than the way that, let's say someone else might help people. I help people with teaching them the technologies and the tools that I use. And that gives me energy and I'm glad that that also helps people to get benefit out of of it.
- [28:20]
B
Now that's amazing. Can I say teaching is a stabilizing force for you. So discipline versus inspiration, Right? So discipline to run the business and do things, but inspiration to do the teaching, become a provider. Is that correct?
- [28:34]
A
Yes. Yeah, that is right? Yes.
- [28:37]
B
Awesome. This is very inspiring, Reza. I love the answers we got here. Thank you. I want to go back to the Microsoft front again. So you've been at the front of every Microsoft data platform. You and I both go to different conferences. You go in a lot more than I could afford. But looking forward for five years, where do you actually see AI going? And even data. I always say data and AI together. I never differentiate between the two. So where it is going for a practitioner like you and me, and it's not just a Headline. Because a lot of people chase AI as a headline and not as, you know, the electricity which will run the operation, the lights and everything. If you are advising a 22 year old yourself, me or someone else who just landed in their first BI data or even AI role, what would you tell them to spend the next two years and what to ignore?
- [29:37]
A
I would say that like AI is growing so fast, right? The AI that we know today is let's say combination of LLM and agents, things like that. A lot of us have seen like agents for doing different things now in a few years this would change completely. We would have like multi agent programs. An agent that like for example, I would say I want to have this project, this project manager agent would go and hire two developer agents, one tester agent. We would go all together and build something entirely without any interaction from me, which is kind of mid unnecessary because I would be more productive in the business because then you have workforce with you. So this is quite critical. A lot of organizations these days are thinking about what would happen in the future. Of course we're not looking in that future yet, so we don't know exactly what is going to happen. But there are some signs of it at the moment. I would say the first sign of it is that, that this would save a lot of time for organizations. Instead of hiring, let's say 10 junior people, they might use a combination of agents to do that, right? So my advice based on that to young generations is that they need to work hard, utilize AI to learn more about the subject that they are going to be expert on. To focus on a particular subject. I don't recommend then to go and do a little bit of everything, right? Won't be a whole Ocean with just 1 meter deep. Try to focus on a specific area, let's say data analytics, but then go and learn about that data analytics. AI can help you actually to learn that. Like for example, what are the top, let's say things to learn? I'm going to what are the top channels to follow If I'm going to learn updates about fabric, things like that, then utilize that and go and learn it. Not to tell copilot, go and learn this and give me a summary. They actually need to go and spend time and learn things. That way they become proficient. When you are proficient in your job, not only AI cannot replace you in a short time. The other thing is that you will find some areas of improvement that others who haven't been working with that much didn't find. For example, just giving you an example. But one of the things that we did in rabocad was to come with a tool called Power BI Helper. A lot of organizations are using it at the moment. The reason we came with that tool was that we worked with so many Power BI solutions that we realized the first thing we do when we work with a customer is that we can find out what exactly is happening in the Power BI file. And we built a tool that connects to the Power BI file, give us all the documentation, which measure is coming from where, which measure is not used at all, things like that. And that tool was a tool initially for us to use internally. Then we got other people to use it. Right. So when I work in that particular environment quite deep, I get to know that what other tools or services might be necessary then in the future will not take AI to go and build those for me. You would have that opportunity that you wouldn't have. If you don't go deep in that part of the technology. I would say say no.
- [33:20]
B
That's a great advice. And in fact, I tell young consultants the same thing I tell my own son. The tool will change every two years. The way you think won't spend your time learning to ask the right questions and explain the answer in one sentence. That's a skill. Being succinct is going away and I really hope and wish the youngsters now chase and acquire that skill more than any anything.
- [33:46]
A
Yeah.
- [33:47]
B
So thank you for that advice. I'm switching gears again. So you run the Power BI and Fabric Summit every year? I was fortunate to speak in that couple of times. It's a community led event in the space and I had the privilege of speaking not on the stage, but online, but from a listener who has never heard of the summit. What is it it, who is it for and what do you want for them to know in the next year's edition or even this year's edition?
- [34:19]
A
Yeah, thank you for mentioning that. So the Power Bi and Fabric Summit is a yearly conference that we run fully virtual. It's not in person. It is not running in a particular venue or a city or a country. It is fully online. We run it in two different time zones so that everyone in the world depends. It doesn't matter where they are. Like it might be in a different parts of Asia, Europe, us, Australia, New Zealand. There is always a time frame that would work for them. We have sessions in different rooms, like eight sessions, usually in parallel happening sessions after each other. Each of them is like about an hour, 40 minute session. The Q&A after that, the Q and A is live. The session itself is pre recorded. So have the best quality of presentation for the attendees. And who should actually attend this? I would say anyone who work with data analytics with Microsoft Data analytics specifically because this is focusing on Power BI on Microsoft Fabric. So as long as you do anything with any of these technologies, for example, you might want to have like your organization might want to migrate to fabric. This is the perfect conference for you to come here. Because we have a range of sessions. It is from beginner all the way to advanced. We have sessions that are like deep dive for someone who have been working with Power BI for like 10 years and we have sessions for someone who hasn't worked with Power BI at all. Right. If any of these technologies is something that you would work, then this is the right conference. The price of this conference compared to let's say a in person conference is not comparable. Like you pay thousands of dollars for in person conferences. Plus the fact that you need accommodation, you need travel plan, something like this. You sit behind your desk at your office or in your home and you watch the sessions. You can also interact live with teams. Q and you can watch these sessions as many as times you want. So I would say this is like great value for anyone. Not just for the learning experience, but also for the connection getting to know all these different speakers, all these different people who are working with the technologies that you are working with. The next one that we have is coming last week of February 2027. We haven't had the call for the speakers out yet, but soon we will have it and soon go. Your website for selling the ticket will be also open.
- [36:53]
B
No, that's great. And do they get access to the content post conference once it is done?
- [36:59]
A
Yeah. So we will have like four hour access to this content. We call it lifetime access. This means that even after the conference is completed, you can watch decisions as many as times you want through the online platform. But you have access to that online platform for.
- [37:19]
B
Yeah, no, that's great. And speaking from my own experience last two years when I was able to do it, I see more practitioners rather than more novice, meaning people who have already playing it and that shows the value because they want to learn something bigger and better. While you do say that people who are not into the state technology at all can come into and learn from it as well. But I see more advanced users also coming to the conference. And I made some great connections. So the density is unique and I have to commend you onto that yeah, thank you.
- [37:53]
A
Thanks for mentioning that.
- [37:56]
B
So let's move on to something lighter. One of my passion is playing guitar. I see you. You are an amazing guitar player. And I've always thought music and technology, especially like teaching materials, learning structure, they are very closer than they look. Both about pattern, repetitions, listening and improvising inside structure. Right. You can draw a lot of parallels without going into those details. Does that lens show up in how you teach Power BI or run redcat, or am I projecting?
- [38:33]
A
Yeah, yeah. I think that pattern is pretty much everywhere in the music. Also, like, if you want to learn language, with the language, it might be still little bit different, but in music, it's kind of exactly like that. So you start by learning a technique first. Then once you learn the technique, you go and play that technique quite a lot to make sure that you master that technique. But once you learn this technique, that technique, and that technique, then you start to improvise between these, that how can I stay in the rhythm, how can I stay in this structure, but improvise in a way that I would be more creative? So I might find some interesting things. That is exactly like the way that we do the music. I would say the same way that I would teach Power BI or Microsoft fabric or any other subjects that I'm teaching. I think that is the way that I learned myself built in, and that is the way that I try to teach others as well.
- [39:36]
B
Yeah, I see that. And great explanation. Practice is generally an unsexy core of any craft, whether it's music or dax. So you would learn different keywords, techniques, but at the same time, you need to be creative. So I think technology requires a lot more art than the science, even though science is very important. So is music. So that's a great parallel you draw. Thank you, thank you, thank you. You spent 15 years as an MVP and 20 years in this ecosystem. I was talking about the long arc, you know, which has been the throw line today. What does the version of Reza who started blogging at Radicat in like, SSIS days, in one sentence, what have you learned by surviving every platform that turns since? So, like, technology is changing, but you are surviving. What's the secret?
- [40:35]
A
Well, I would say start thinking out of the box of technology. Start thinking of what problem you are actually solving. Like, if you are doing the BI project, you are helping organizations to make informed decisions. So then you would need data. You would need clean data. You would need clean data to be modeled and analyzed in a way that can be consumed. This might be in the past with just SQL script these days with Power BI in the future, it might be a different technology are doing the programming. The programming itself is not what you do. You are helping other businesses to build applications that they use on a daily basis. This application might need a database to store some of the information and it might also need some forms of entering the data, making sure that the amount of errors when entering the data is reduced. So when you think out of the box, when you think that these are the problems, problems that I'm solving, then you just try to find how you do that with the technology. 20 years ago with SSIS, if I wanted to do a slowly changing dimension, it was a different process. Nowadays if I want to do that a slowly changing dimension in Microsoft fabric using pipeline and dataflow with a different process. But still it's the same thing. Still we are solving that particular challenge. The way that I looking at it, is that how we can help each other to get to solve that particular business problem?
- [42:13]
B
No. That's great. So building a content engine so you have learning and being very passionate about technology and also teaching. But what kept you going on blogging, writing books and running an academy at the same time for a decade where people quit at six months and everything requires dedication, hard, hard work, discipline, consistency and above all, inspiration. What kept you going on these areas? Which is more of an influencer area?
- [42:41]
A
Yeah, I would say just passion. I like technology, I like working with technology. So every day, every week that I work with technology, I start to record these. Sometimes as a video, sometimes as like a blog post so that I can refer to it later myself because I don't have like a memory of like 20 years of things. Like if I, if I want to find something that I even I have done in the past, it's better to go and look at my blog in the past or my video in the past. So consider it as a passion and also as a little diary for myself to read later. But others can also get benefit from it.
- [43:22]
B
And that's inspiring a hundred percent. I got inspired and I started doing this while I was thinking I'm not good enough to do this, this, but you have to start somewhere. And we are in top 20 on Apple podcast. The listening side is working great. Thank you so much. You are the inspiration. So we run this podcast for three types of audiences. AI, curious, enthusiast and skeptic. What's the one power BI fabric thing? Each of them should learn this quarter. So if you're an AI, curious what you should learn or enthusiastic what you should learn or skeptic even what you should learn.
- [44:02]
A
Well, if you are I would say starting from spot talk I would say you would better to get to know with MCP servers. These are giving you really interesting things about now that you can use AI agents to talk with these technologies. If you are like AI just enthusiastic. I would say start working working with things such as Copilot in Power bi. See how this can help you. And if you are in the middle layer, I would say try some of the technologies in Microsoft fabric such as Data Agent, Fabric Agent or even Prep data for AI of Power bi. That will help you to understand how AI result can become even better when combined with these.
- [44:49]
B
Oh no, that's very nice. Thank you for that suggestion. One of the thing I forgot to touch base on adcad. You're running a successful academy and it's a paid training work in the market where so much free videos, including yours, free documentation from Microsoft, certain free community events. What made paid training work for you?
- [45:15]
A
Yes. So as you mentioned we have a lot of material, a lot of content already free available. Like our blog has like thousands of blog posts. Our YouTube channel has like like almost a thousand YouTube videos, things like that. What makes it beneficial for someone to come and learn using a paid platform? The thing about YouTube or blog or things like that is that they are not structured in a way that we usually go and learn something. Right. For example, my blog one day might be about writing this tax time on intelligence calculation and another day might be related to what is the governance principles in Microsoft fabric. Whereas if I'm going to design a video course I would have it laid out entirely what is the syllabus of each? Where this starts, where this ends with some practical examples to go through as well. And that is where the main value is for someone to start from ground zero in that particular area to go through all of these aspects in a structured way. For example, the particular DAX function that I expl explained in that YouTube video might be the same function that I explained in this video course as well. But because now it is in the context of the course, it is in the right place, it is in the right order of elements. It makes much more sense. So people usually found this way of learning much better. Whereas the free content usually is good for someone who is already proactive, who already knows the basics, just wanting more tips and tricks here and there. Whereas the the actual structured learning is to actually go and learn that in a deep level.
- [47:02]
B
Yeah, better structure and sincere learning requires something to be Paid on. And that would make sense to do. And speed that up. Last question before I close. You are from New Zealand. You have worked in New Zealand community in Microsoft versus US and eu. What's different about how the community shows up? The people, the warmth, the organization? What's the difference? Difference or similarity that you draw?
- [47:31]
A
Yeah. So I will say in terms of the community, we have a really good community in New Zealand. We have been working with this community since like when I moved to New Zealand 2012, like 14 years ago. Like we had user group sessions that it was only three people in the community. Like in, in that room, me, the speaker and the attendee. Right now there's. If we run a user group meeting, we have like 70 over 70 people easily in the room and we have to close the book because otherwise we don't have enough space. So the community is growing, the community is nurtured. It's a good community around using Microsoft technologies. If we compare the overall usage of data analytics technologies in let's say New Zealand compared to us, I would say we have a slightly more progressive usage of AI in US and Europe compared to New Zealand, Australia. In New Zealand, Australia we have organizations who are using AI, but they are mostly still in the process of like leveraging what these AI functions can do for them. A lot of proof of concept. Whereas in our US customer base, in our, let's say North America customer base or Europe customer base, we have companies using it already in production. Of course Lithium and Australia will get to that stage as well. But it's just I think a little bit lag in the market to get there in terms of community. However, I think they both are really a strong community.
- [49:06]
B
Great, great stuff. Sureza, thank you so much. Before I close, anything you have to add?
- [49:15]
A
No, thank you. Thanks for the opportunity to be here with you and I hope everyone enjoys working with the Texas.
- [49:23]
B
Thank you. So that's Reza rat, founder of Radicat 15 time Microsoft MVP, continues since 2011. Wow. Just talking about it gives me a goosebump. And one of the most consistent teachers in Microsoft data platform community. Find his blog and the radicatacademyradicad.com I'm going to leave a link here. His books on Amazon, his videos on YouTube and you can obviously find him on link LinkedIn and run into him in conferences. If this video help you frame your own power, BI fabric or AI strategy, share it with one person on your team who's wrestling with the same question. That's the highest compliment a podcast can get. Thank you, Raza. Thank you again.
- [50:09]
A
Thank you, Dave. Thanks everyone.
- [50:12]
B
You have been listening to Think AI podcast with Dev. Take one idea from this episode and turn it into action.