Is low code really “dead”… or is the Power Platform just evolving? In this interview from the Power Platform Community Conference, I sit down with Microsoft MVP Victor Dantas to break down how AI is reshaping the way we build apps, websites, and automations.
We walk through a real customer service scenario and show how to go from blank canvas to working solution using AI across the stack.
In this video, you’ll see how to:
✅Use Power Platform Plan Designer and prompts to generate a full data model and model-driven app
✅Turn a blank Power Pages site into a Dynamics-style portal using AI-generated CSS, HTML, and
✅charts Offload 90% of the heavy lifting to GitHub Copilot while still staying in control of the architecture
✅Use MCP connections to Dataverse and Azure DevOps so Copilot understands your tables, work items, and context
Want to learn from more pros? Learn from some of the top talent around the world at Microsoft Community Conference, and get $150 off with discount code VLAD150! Have questions? Check out my Microsoft Conference FAQ!
OR if you live in Europe, I highly recommend the European Power Platform Conference (EPPC)! Good news! I also have a discount code for this event too! Save 15% on any EPPC pass with code VladSpecial!
Watch my 100+ courses on Pluralsight
Video Summary
- “Low Code is Dead” – But Not Really. Microsoft shook the community by announcing “Low Code is dead” during the Power Platform keynote. The real message? Low code isn’t gone—it’s evolving. The Power Platform is being reborn in the era of AI.
- AI Is the New Superpower for Makers. Instead of dragging controls and writing endless formulas, makers can now leverage AI tools like ChatGPT and GitHub Copilot to generate CSS, HTML, and even full components for Power Pages—turning a blank template into a polished, Dynamics 365-style site.
- Prompts Are Your New Programming Language. Detailed prompts are key. Victor demonstrated how he used AI to create case deflection logic, dashboards, and reusable components—all by describing what he wanted in natural language. Knowing how to prompt is now as important as knowing code.
- MCP Changes the Game. The introduction of MCP (Microsoft Copilot Protocol) allows GitHub Copilot to understand your Dataverse environment and Azure DevOps tasks. This means AI can generate context-aware code, suggest solutions, and even help you complete work items without leaving VS Code.
- The Future of Power Platform = AI + Makers. The Future of Power Platform = AI + Makers
For more information, read the transcript blog below, or watch the video above!
Transcript
Low code is dead. At least that’s what Microsoft just announced at the Power Platform Community Conference during the keynote. And I know that might be scary for you. It’s scary for me too. But that’s why I’m lucky enough to be joined by Victor Dantas, Microsoft MVP and author of The Power Platform Playbook for Digital Transformation. Victor, thank you so much for joining me today.
Thanks for the opportunity to be here. I’m excited.
What was your reaction when Microsoft said in the keynote that low code is dead? I was shocked to hear it from the VP of Power Platform. But the way I took it, when he said low code is dead, I think we could tie the bow in that particular sentence if we had a little bit of a sentence that said, “But Power Platform isn’t. Power Platform is reborn.” I think that’s what they said. Isn’t that a beautiful way to say it?
But today, what I would like you to show me and everybody else here that kind of saw that slide “Low code is dead” is how it is reborn. And I know you’ve shown this week some beautiful demos on how you can use AI and not only in the Microsoft ecosystem, but outside as well, to build cool stuff.
That’s correct. Yes. Yes. Let’s do it. I’m excited to present it. There is a lot of chatter in the community about whether or not low code is dead. I believe it’s not necessarily dead, but things are changing for sure, and it’s important for people to stay in the know so that they don’t get left out when the boat leaves. The only constant in IT is change. That is correct. A lot of industry leaders and community leaders posted a similar sentence a few months back, in which they said low code is dead. And it was a thought-provoking set of posts to instill all of us to rethink the way we do our work. So if low code is dead, what is the new thing? What is the Power Platform being reborn into, if you will? So this is something we’re going to take a look at here today.
And I see we’re starting in the famous Plan Designer. That is correct. So what we are going to do—and this is actually a segue or a takeaway from the sessions that I presented here at the Power Platform Conference—is build a fully functional customer service solution that includes a model-driven application and a website with the help of AI. Of course, because we are bound by time here on this video—I’m sure you can talk about it for eight hours. You technically did in your workshop. I did. So this is my eight-hour workshop in, what, twenty minutes maybe? Let’s set it as a challenge. What can we build in twenty minutes?
Hey, let’s take a look at it. But in all fairness, even if we don’t show all the things that are involved in this process, I’ll show you the key pieces so that you can play with them yourself. And at the end of the video, I’ll show you a link where you can download some of the materials to do the same thing yourself. Okay. Awesome. Is that cool? Sounds like a plan. Are we ready? See what I did there? Very good. Very good. Very nice.
So, for this solution, what we are doing is I have a developer environment. And I have crafted a prompt or a set of prompts ahead of time. S,o for this prompt—and I hope Microsoft will not get mad at me—what we are doing is this: we are telling Plan Designer to create a customer service solution similar to Dynamics 365 for Customer Service using the following tables or attributes. For attributes, I’m referencing Microsoft’s Learn website to give me inspiration for what I need to build. Awesome. So I’m using intelligence that’s already available there. That’s the goal, right? You have the intelligence. You already have the data. Leverage it.
If you do that at home, you end up with a model that has a few tables that resemble what you would get from the Dynamics 365 Customer Service experience. So we’ll fast forward to the model-driven app itself. Here on the second tab in this model-driven app, I have cases. Cases reference customers, and then you have a field for description, and so on—everything that you would see in Dynamics 365 for Customer Service. Of course, with the Microsoft product, you get their own Copilot and a lot of other automation. But for a company that is starting in their journey within Power Platform, maybe this solution would work for them—a simple customer service application. Here you have the description of the issue. People can provide a resolution. If the resolution is associated with an article, they can select an article. They can navigate to the business process.
And this is something that—I know we did a bit of a cooking show, I like to call it. We kind of started it, and then we got the app out of the oven. This is what you got right after you ran the prompt with Plan Designer, right? Exactly. So the Plan Designer will suggest the set of tables that you want to create, but it will also provide you insight as to the applications and automation or even a Copilot that would help you fulfill the mission of this particular solution. Awesome. Yeah. So you get it all with one shot. Of course, you can refine things a little bit, and we’ll leave that for another day. But the mission is: the company now has a model-driven application that can be used for service request management or case management.
Awesome. All right. So the next part of the deal is: what if I need to build a website that people can use—my external customers can use—to submit tickets? Oh, that’s awesome because this would be the internal one, but you also need customers to interact, see their own tickets, and things like that. Exactly. So, Plan Designer gives you a starting point for websites as well. But because my workshop was centred on Power Pages techniques, I decided to use a blank template for it and build it from scratch. You’re very brave. But here’s the catch, though. I told this in the workshop and in my sessions: I don’t consider myself lazy. I think I like to get work done. I’m focused. But I like that AI can do a lot of the work for me.
So in that way, I take a little bit of the accusation of being a little lazy. But how did we start the process? You can create a site if you go to make.powerapps.microsoft.com. That’s how you get a site started. And again, we can look into detail,l maybe in another video. But what you end up with when you create a site is something that looks like this.
Not very pretty.
No, that’s about my skills in Power Pages. That’s how I can make it look. So the tagline for my workshop is “From Blank to Beauty.”
Oh, I love it.
Because you start with a blank template and you end up with something a lot nicer. Now this is what the site will look like in the end. So let me click on it and see the result. Do you think this is a website or a model-driven app?
It looks like a model-driven app.
But this is a Power Pages website.
Wow.
So the next stage for me to go from blank to a site that looks like a model-driven app is to use AI intelligence to help me fill in the blanks. So okay, I know that Power Pages runs on top of Bootstrap 5, so I have a little bit of technical insight into it. So what I can do is I can ask—be it ChatGPT or GitHub Copilot—you can ask it to give you a CSS file that will help your site look like a Dynamics 365 first-party application.
Nice. And as you said, you don’t have to use Copilot—you can use whatever you like.
Whatever you like—ChatGPT, whatever you like. And this is what my prompt looked like for it. I know people don’t really enjoy the old and famous Notepad, but this is what I have put in place for that prompt, and that was done in ChatGPT.
Okay.
Now, a little extra sauce here for folks who are starting this journey—be it for Power Pages or any development that they’re doing. You can upload images to ChatGPT and ask GPT to infer from the image while creating this asset. So for my CSS, besides the prompt, I added a screenshot of the D365 Customer Service. “Hey, give me CSS for something that looks like this.”
Exactly.
Nice.
Yep. And it will look at colors, do the translation—the colors that it sees in the image—to HTML code colors, hexadecimal colors. It will do all of it for you.
Wow.
Okay. So this is the prompt that I used, and again, I used ChatGPT to get it.
Okay.
In the end, I got a CSS file. I’m not sure if we should go into details on the CSS, and I’ll have this demo available for you—you can see that there are about 300 lines of CSS code that make your site look like a model-driven app built or used within Dynamics 365.
Cool.
All right.
Yeah.
So the next thing we said was, “Okay, I like this a lot. Give me some components that also look like Dynamics 365.” Meaning the ribbon—the ribbon area that you see here is very similar to what you see in D365.
It’s the same.
Yep. And so this is what we have done. We asked GPT to do that. It gives us results in HTML format. I took that, and I put it on the site.
Wow.
Okay. So the next thing that I showed people was, “Okay, you like the CSS, you like the way it looks—it’s already a game changer when it comes to Power Pages and how it’s presented—but what if I want to create extra automation that’s not shipped out of the box?”
Yeah.
So the first example that we had was: if I want to create a case, or if a user—an external user—wants to create a case or navigates to the Create Case page, what can I do to prevent the user from creating the case but still receiving information to solve the case without having to engage with our backend people, right? That’s called case deflection. We came up with a prompt for it, and let’s take a look at the prompt first.
You’re very detailed in your prompt. I love it.
And the reason why—and we’ll see why we do this, especially for ChatGPT—we have to be detailed because GPT does not have insight into the model, the tables that we’re using. So we are giving a lot of details there. But so here’s the prompt: in a Power Pages site, when a user types more than five characters, take the value on the input field and then run a query behind the scenes to find knowledge-based articles with this keyword.
Yeah. And I see you have results inside the div classes—everything.
So I’m giving all the details. Again, I’m giving the details to GPT because, again, it’s not yet able to see what’s behind the scenes. But there is a way that the prompts will get a lot shorter, and we will see that here today. That’s why I don’t think low code is dead.
Okay, cool.
But Power Platform is reborn. All right. So I took this prompt, ran this prompt, got the results, took the code, and the feature looks like this. So if I type “password,” then I get the results here, and users can see the knowledge-based articles.
That is beautiful.
Help them fix the situation.
That’s amazing.
Yep. And then the last demo that I did was dashboards. So we said, “I like the forms. I like that we can use out-of-the-box Power Pages forms and lists.” This is what you see down below—the list here. This is all out-of-the-box Power Pages. But I wanted to add some charts.
Yeah, I wanted to make this thing look nice. I want it to feel like I am in the Dynamics 365 first-party app. So I said, “All right, let’s craft a prompt that will give us these results.” Again, the prompt is very lengthy and verbose, and I’m giving lots of details here as to what the data model looks like, what the option set values for the statuses and different attributes. So I bake all of that into this prompt, and the result, as you can see, is that we got charts that are animated. If I reload the page, we see a little bit of animation, and it’s doing all the work for you.
That is really cool.
And all of that, you didn’t actually open up Visual Studio Code at all. Well, you needed a bit to know what div to put it in and things like that. So, there’s still some knowledge needed there, but you outsourced 98% of the work to AI.
Yes, 100%. I mean, a lot of the work of pasting the code there was manual. Yes, but it was just grabbing it from GPT and placing it within Power Pages. So the demo in Power Pages—you can edit pages in the browser using VS Code.dev. So that is the strategy we use. But the part that I’m most excited about—both the keynote and where things are going—is what you can do from VS Code for the desktop.
Can we do that still? How are we doing, time?
Oh, we got time. We go.
Of course.
Let’s do it. Let’s do it. All right. So, in Visual Studio Code—or in VS Code—what I did is I downloaded the contents of my website. The presentation will showcase how you can do that. And now all of your assets are in your local machine. For those who are careful with source code and all that, it’s a game-changer because you’re pulling it down to your machine. If the folder is bound to a GitHub repo, then now everything’s taken care of, right? But I did that. I brought the contents of the site to my local machine.
And then from here, now, because I have the files here, I have GitHub Copilot on the side, I can do all kinds of different things. The first thing—the first advantage of having it all here—is that you can include Power Pages AI intelligence as a participant in the conversation. So it knows when I start engaging a conversation with GitHub Copilot that the answers have to be somewhat bound.
Yeah. Like, don’t give me random code that will not work. Like, don’t use your imagination too much—it has to work in Power Pages.
Exactly. So we are putting that boundary there for the conversation. The other thing that I’ve done and I demoed is this—and this to me, if you’re watching this video and maybe you fast-forward to see some solutions or you watch it all and you miss some parts of it—the most important thing in this video is the keyword MCP. Remember this—tattoo it to your forehead or something. Don’t forget it. MCP. So I have configured an MCP connection to Dataverse here and another one to Azure DevOps.
What that allows me to do is I can now talk with GitHub Copilot with the understanding that it knows what’s going on in my Dataverse environment. And this is not only good for Power Pages, though. This is good for any kind of development Power Platform. So if you’re coding a plugin, if you’re coding JavaScript automation for model-driven apps, so on and so forth, it’s got that intelligence here.
So that is the Dataverse MCP. The other MCP here—the Azure DevOps MCP—the way it helps me is, at first, you know, at first level or level one, is that it can tell me what tasks I have to work on, right? If I have visibility into Azure DevOps, if I have tasks assigned to my name, I now know what those tasks are. I don’t have to open three, four, ten thousand browser windows—focused in one, focused in my work, right? So I can run queries like, “Show me all the work items that are assigned to me today.”
And then from there, the MCP has visibility into the details of the task. So I could do things like, “Based on the success criteria for this particular task, write some code.” Based on the people that exist—you have your objective—and then you work with GitHub Copilot to actually… You can have code to start with, and maybe the full solution without even knowing what was assigned for you for work.
This is amazing. Okay, this is crazy. So I have in the demo—I have some commands, not sure how MCP will behave here at the MGM—but some of the important commands, at least for me within the Dataverse MCP, are “describe table.” And let’s see if this is going to work for us.
I will reference the MCP here just in case it doesn’t know in context—help it as much as possible—and let’s see what it comes. But in theory, in your day-to-day, you don’t have to reference it like that. It’s usually smart enough, but no,w just because we’re recording, we want to help it as much as possible. We want to make sure that it works. I found it running “describe table.” It says the MCP server is there.
And okay, it’s helping us. So it’s listing the tables first. Ah, so it said, “Hey, you asked me to describe a table, but I’m not sure which one. Let me list them, and then I’ll describe them.” And guess what? In my environment, because it’s a developer environment, I don’t have the incident table, which is the out-of-the-box case table for Dynamics 365 for Customer Service. But it detected that I have a custom table called “case” as well. But it gave me the internal name of the table, the fields, the attribute types, the choices and whatnot. So from here, if I want, I can go back to my prompts and now skip to a prompt that’s a lot easier. I can say, “Hey, I love that it doesn’t only say, here’s the code.”
It’s like, “Hey, this is what you need to do. This is why I did it, and this is what I was thinking about.” So that part is awesome.
And for folks who are starting in Power Pages specifically, you could go into GitHub Copilot and say, “Okay, I see a lot of code here. It’s a full HTML page, right? There is quite a bit of code, but we could come back here and say things like, Hey, help me fragment this code so that they can become reusable components.”
How can I do that? And then it would explain what you need to do to achieve your result. Okay, so back to Charles’ declaration—if we go back to the beginning, low code as we know it, it’s dead in my opinion. I mean, I don’t think there is a way back to drag-and-drop and Power Fx formulas everywhere and trying to adjust that way. But I don’t think the Power Platform is dead as well. I think those are two separate things, as they put it very well. The Power Platform has been reborn. The Power Platform is reborn in the era of AI.
But thank you so much. I think this is really eye-opening on how many awesome things you can do now as a maker with minimal knowledge of code. You still might need to know a little bit of HTML, a little bit of CSS, but if you know how to prompt and you know the right tools, you can create something. I never thought I’d be able to create something as awesome just using AI. So that is amazing.
Thank you so much for your time. I really appreciate this peek into the future of low code, really. For those who want to know more about the book, can you tell me a bit what your book covers in 30 seconds, and why people need to read it?
Okay, so Chris and I—Chris Huntingford and I—we had this idea of putting a little bit of our knowledge, or maybe lots of our knowledge, into a book. We felt like every initial business conversation or even community conversation, we felt like we were repeating ourselves a lot of times. And we thought maybe we could put some of our lessons—the lessons learned—into a book format. So in this playbook, if you will, people will be able to see a lot of our story reflected there, and it’s centered on projects and people and building teams more so than the technology itself. I think the technology will evolve, and there will always be new things and new ways of doing the same thing, but how you form the team, what part or what key members you need, and so on and so forth. So that is the idea.
Awesome. Thank you so much. So, for everybody, I don’t know about you, but that sounds awesome. I’ll add it to my reading list. You’ll have a link to the book in the description below, as well as all of Victor’s social media. He’s super active on X, on LinkedIn, almost everywhere, and with really awesome pictures as well of Agent. So you’re not going to get bored if you connect with him—and lots of great insights as well.
Thank you so much for taking the time. Really appreciate it. And for those of you watching, if you enjoyed it, again, make sure you connect with Victor. And like the video and subscribe to the channel for more awesome content like this one. On the screen right now, we talked about the keynote. You’ll have a link to the keynote summary I did. So if you missed that, you can have all the announcements in less than 10 minutes as well. So I hope I’ll see you there.
