Power Platform Boost Podcast

Ignite IQ (#73)

• Ulrikke Akerbæk and Nick Doelman • Season 1 • Episode 73

Share episode
Ulrikke:

Let's move to power platform news because we got a very interesting announcement too this uh time around.

Nick:

Yeah, so the morning of Ignite, I wake up now. Here I am on the West Coast, which is nine hours behind Europe. And the first thing I do, and this is I'm a geek, is I open up LinkedIn. Hey, what's new? Who liked my stuff, you know? And I just see everybody's talking about vibe.powerappps.com as a brand new thing. And it's very funny for the Europeans because it's only available on US-based tenants or US environments within the tenant. And I'm like, well, what? Like, I thought, okay, this is this is exciting. And so I actually went and tried it out myself. And wow, you can. So if you've used the tools like Replit and Lovable, this is the Power Apps version. Welcome, everyone, to the Power Platform Boots Podcast, your weekly source of news and updates from the world of the Power Platform and the Microsoft community with your host, Nick Dolman and Litika Akeback. So welcome everyone to another episode of the Power Platform Boots Podcast. We can tell we're having fun. We're recording this on a Sunday afternoon.

Ulrikke:

Yeah. I've been out outdoors with the kids the whole day. You know what it's like in the back of a day when you're being outside and it's been like the most crappy weather you could possibly have. It's been raining and it's been four degrees. When you get inside after a day like that, your whole body is like hibernating or something's happening with the temperatures. I'm so hot and so overwhelmed and so tired. You're right. So this is gonna go.

Nick:

Best episode ever.

Ulrikke:

The best episode ever. Okay, so you want to try this again?

Nick:

Let's try this again. So, as as we were attempting, this is gonna be fun to edit. I'm just gonna probably where you muted me. So we were saying last time we recorded was the week of Microsoft Ignite. And if you've been living under a rock, Microsoft Ignite is Microsoft's big dog and pony show where they're showing off all their new shiny toys and everything. And when we recorded, it was actually the day before. Now we had the book of news, and there's we didn't really want to the the book of news before it officially gets released is pretty fluid, so it was kind of hard to say um what would be there. And the other thing was I think that version of the book of news might not have had everything that was even announced at Ignite, so it's kind of hard to see what the spotlight was going to be on. So um we will be talking a lot about what was announced at Ignite. Uh I was there. Um, I didn't see a ton of sessions in terms of just because first off, all the sessions are recorded, so I had that at the back of my mind, but just getting around the venue, the venue was massive. So just a little bit about Microsoft Ignite. There was like 17,000 people there. It was in San Francisco. The security was really high this year. There were protests outside of the venue. I didn't see too, too much. I did see a van driving by with a big shiny poster on it saying uh Microsoft Ignites Genocide. So there were people concerned about that. There were a couple people with megaphones handing out flyers, but of course they were kind of kept at bay with security. It wasn't in terms of any incidents, but in terms of getting in and out of the venue. If you had a regular patch, for those of you who've been there, you know this, you would have to go through um like almost airport-level security in terms of checking bags. In fact, I had a staff badge, or because I was working in the expert's booth, we we did have a separate entrance, so we were able to get in a lot quicker, a lot easier. But I really feel for anybody that had just a regular ticket because even to get from venue, like there were three different buildings to get from building to building, you would have to pass through security again. And I could see that being very tedious. So overall, the the the expo was good. I did work three shifts at the expert booths, which I really actually really enjoyed. I had some really good conversations with different people, people who didn't, I was kind of on the Dataverse booth, people who had no idea what Dataverse was, to other people that kind of cracked open their laptops saying, Hey, I'm having this problem and this problem. Can you help me out? Give me some directions. Um, to other folks too, because the day with the booth had something with Dataverse and Agent 365, which of course, Agent 365 was like the big announcement. So people are asking me, like, so K, what is the relationship between Agent 365 and Dataverse? And I'm kind of like, I only found out like a day ago about Agent 365. So here's what I think it means. And I'm almost like reading right from the from the documentation. But actually, uh a bit of a quick shout-out to uh Daniel Laskovich, who was there as well. And he actually showed me some things with Dataverse MCP server at Agent 365, which helped ground some of my context of more than what Agent 365 can do for us. So that was really cool. Thanks, Daniel. Greatly appreciate it. You're always always there to help out.

Ulrikke:

So the keynote is what, what, two and a half hours long or something? And this time I found it was a really nice narrative because they brought it from end to end, a red thread, where actually they so they have a new Contoso. Uh so what's like the new firm called? Zara?

Nick:

Zora, yeah. Contoso, Contoso is dead. Long live Zara Zara. Zara? Zara? Yeah. Yeah.

Ulrikke:

Zora Zara? Zara. Uh, which is also a frontier company, which we will talk about soon. But also they had a narrative where they ordered t-shirts for the whole audience at the venue at the actual keynote, which was fun. And at the end they had a guy come in with the big boxes delivering the t-shirts, which is a lot of fun. But also, it kind of brings the whole thing together, and it's actually the the devil's in the details, and it's the bread and the butter, and it's the thing you can do. And they kind of embodied something that was a bit abstract. So I think it was a smart move for them to work with something that tangible and that all the demos played into this the scenario and the story that they'd built, which I thought was a lot of fun. So um let's just kind of go through some of the highlights that we are gonna touch on and talk about a little bit more. First of all, I think we need to talk about what Frontier Firms is, what Frontier is. And I found it funny to see the when I picked the highlights that I wanted to talk about when I had an internal kind of tech talk about Ignite. And then I was like, well, Frontier, we talked about this just a couple of episodes ago. And then we also talked about a lot of the other things. The last episode was about the forward um uh forward deployed. Yeah, and so forward deployed, yeah, which also comes up a lot with the frontier firm. So I was like, ah, yeah, we're in the we're in the midst of it. We we are kind of one step ahead. Uh, and uh, and I like to see that. Frontier firms will talk about that, and of course, work IQ, Foundry IQ, and fabric IQ. We can't really not talk about that, which is massive. We are gonna touch on AGEN365, and of course, news for power platforms specifically with the WIBE stuff, and also I wanted to touch on the app migration agent. Do you have something that you wanted to highlight?

Nick:

Uh in the book of yeah, yeah. So basically all of these things, of course, there's like a ton of other announcements. Uh, I really want to dive into basically we can at the tail end talk about one of my favorite sessions, and that is the reimagine the human agent collaboration. Um, and that was put on, and we'll talk a little bit more about that because to me, that whole session really made made this real in a real world sense and saw how I'm gonna be able to work with this and how it's gonna affect my immediate customers and things like the MCP servers and things like that. So those are things I want to touch on. But yeah, frontier firms, I mean it's interesting because if we kind of pop back to the, you know, in terms of the news, I do have something in in our news and updates, so it all ties in, segueing nicely. But basically, it was an article that I found. Um, it was online. It was actually it was in um linked, but posted by someone called Ben Lee. And basically, his question on his article is what exactly is Microsoft's co-pilot frontier program? And so I can explainer blog, and he gives uh an amazing example of what Frontier firm means, how you can turn it on for your environment, how you can turn it on for specific users, what it all means, why it matters. Like at the end of the day, it is kind of beta testing or public preview, but overall, there is a whole strategy behind it, and there is actually some guidance on how you can deploy this in your own organization so you're not willy-nilly turning on half-baked software, but you're going through it through a strategic way to kind of monitor and try this stuff out. And so basically, by doing this, Microsoft is getting real-world feedback, which I think is critically important, but also giving administrators uh tools that they can actually carve off sandboxes and have smaller populations of their users interact with the frontier features that have been enabled within Microsoft 365. So we have that link in the show notes. It does talk about the different agents, it does talk about different aspects of Frontier. Um, that of course, that was a big thing at uh Ignite as well was the concept of frontier firms. And everybody was kind of, you know, for who didn't know it, are we, you know, I'm sure there's a lot of conversations where the where the CEOs were turning to their CTOs going, all right, are we a frontier firm? Like uh how do we get how do we do this? So hopefully if you don't know what it is, we do have a link. And I actually went through the steps because I am a company of one, so I am a I am a frontier firm. I'm always trying out new stuff, and I went through the steps to enable the frontier features in my uh my Microsoft 365 copilot tenant as well. So that's uh yeah, did you want to talk more? You taught what was your perspective on the frontier firms?

Ulrikke:

Yeah, no, just what you were touching on the fact that it is kind of a collection of all the programs because there's a lot of initiatives and programs now, and it's not just for copilot. It also includes something else that I saw pop up this week or last week, which is the Microsoft Agent Factory, which is another one of the programs that is included in the frontier framework mindset way of working-ish thing. Because that's how I read it. It's more of an approach to AI and the new world than anything. So it's a I brought a quote up from the tech talk that I did, and it says, you know, frontier firms don't layer AI on top of existing work, they rethink how work gets done from the foundation up. So it's a way to think about strategy or an adoption of a that Microsoft has already provided in a way. So it's a mindset and it's kind of a strategic shift more than anything. And also just wanted to touch on the Microsoft Agent Factory. It says, you know, IQ for every agent and choice for every maker and deploy out anywhere. It's a new program that's designed um to move from experimentation to execution. Because that's also something we've heard Microsoft talk about. And I know I was at a conference uh at Microsoft in Norway before the summer, and they were all about action. We're done with the POCs already, get to production with your AI tools and your AI capabilities. And this is very much a way to facilitate that with actually a single-metered plan. So this has a lot to do with licensing and also how you work with the new work IQ, fabric, IQ, and foundry IQ. So, and also it's on top of AI forward deployed engineers and and all of those buzzwords in one article. So, we're gonna also put it um a link to where you can learn more about Microsoft Agent Factory as well, which is now kind of part of the same two well, it's one of the child programs within the Frontier program. Yeah. And also touching on that, so from your perspective, being at Ignite, how big is the IQ family announcement things? How big was that compared to other things? Was that a lot of on people's minds? Did you talk a lot about it?

Nick:

Like everything, it's sort of what does it mean? I think to me, the biggest, the biggest sort of thing was the agent 365 more than the uh but the IQ stuff really definitely comes in second when you talk about work IQ. So, I mean, when you kind of look at the definition of of work IQ, and it really kind of makes sense. It's um, you know, it's an intelligence layer. Look, here's a description that actually Google Gemini came up for me. So AI is using AIs. Microsoft WorkIQ is an intelligence layer that powers Microsoft 365 Copilot and other AI agents, enable them to understand individual work patterns, roles, and organizational context to provide personalized and relevant experience. So, I mean, I think what this means is it's reading data from your emails, from your own files, from your team's chats and building a work memory around that. So I think already this makes people a little bit nervous in some situations because I've heard of folks saying, well, you know, you send an email basically like, hey, can you give me a follow-up on this project? And by the way, how is that foot rash? Have you been able to treat it kind of thing as an aside of the email? So these these kind of things you gotta are sort of could potentially be surfacing and bubbling up in the work IQ as well. So again, it's all about at the end of the day, I mean, AI is is as smart as it is, it's really dumb. It's really looking at all this data. So it's all the context you're giving it. So I like the idea that it's building a work memory, and I know I work with other tools, which I work a lot with ChatGPT as well. And it and even the other day I was working on a workshop or a session or something, and it said, well, based on based on the background and what you've done before, here's blah, blah, blah, the output. And I'm like, okay, yeah, it's I am feeding it system prompts in terms of my writing style and everything, but also it is remembering when you have those memory settings turned on, which it's a double-edged sword. It's good that it's because my memory sometimes sucks. Like I'll, you know, I'll forget things, but but it also is going to pick up all of these other things as well. So definitely interesting. I think this is also gonna be a work in progress. And then, of course, you know, the guardrails and the security are all around this as well, like that a lot. Like, you know, I could send, you know, if I send you an email to say, here's your bonus this year, we're giving you a bonus of blah, blah, blah, all of a sudden that email could now be part of that knowledge context that it's picking up, right? Which if I'm writing something or I'm generating content or generating data around all of this stuff because it's my email or it's in your inbox, could the AI pick up on that as well? Um, and then so who knows, right? So this is interesting times, and we and it's great technology, but like with great, you know, great power comes great responsibility as well.

Ulrikke:

Yeah. No, no, but and I don't think that Microsoft would ever even consider announcing something or launching something that would give would make people nervous about their email content. So what you had access to before is not gonna violate the that security for sure. So it's not about kind of sharing details from your email with everyone else. It has more to do with kind of personalized for you, but also if it does that with your colleagues and the other ones as well in your office, it's more of a collective memory and also has to do with culture and and kind of how you work and who you are as a company. I think that's what work IQ is trying to get at. What kind of company or who are you? Know how you work, what you do, how you do things. It is the collective kind of the things that are that in we embody as as co-workers and as employees that isn't always written down as well, right? We'll catch up on those things. But it is, it does consist of memory, the patterns, the workflows, the relationships, those kinds of things, inference. So it's predicting actions. And that's probably one of the most interesting things here. It's predicting patterns in terms of tools and and what you use and how you use them, and also how that's gonna be adopted throughout your organization. And then it's also the memory part. So it's very interesting to see how this is gonna tie into everything. And then we can move on to the fabric IQ, which is so if work IQ is how you work, it's my Microsoft 365, like you said, emails, notes, documents, SharePoint, all of the storage of that kind of data, then then fabric IQ is more of the the actual data that you're on top of. Imagine all the Power BI reports, all the data flows, all the data verse, data power platform, all the data you have in there is it's it's looking at that as entities. And suddenly the entity phrase or word is is coming back. It's now talking about entities and relational entities, how they relate to one another to get a cohesive and holistic view of the data that you are working with, the data you produce and the data you store. So that in terms of a factory, work IQ would be the emails, the processes, the workflows, the how you do things as a company. Then fabric IQ would be the IoT data from the IT devices, the the output of your factory, the the numbers for what you produce and when you produce it, that would be work IQ. So it has to do with the logic and the data. And then you look into fabric IQ, which is the knowledge and retrieval layer. This is where you have fabric, but all those AI capabilities, the rag style workflows and the and the agent patterns, all of that is in there. And all three of them together in one grid system, knowing about each other, that AI, the IQ foundation, that is the way of a frontier firm has that as the basis. That's the new foundation for the firm, and you build everything with that on top of that. That's the idea, as far as I can at least.

Nick:

I think you mispronounced the second not pronounced misword, missed you said the second one was fabric, but I think you meant foundry IQ as the second IQ, right?

Ulrikke:

I don't know if it matters which order I've said the fabric. So work IQ is the kind of the Microsoft 365 one, fabric IQ as the data layer, and then foundry IQ is the last one where you have all your AI capabilities and the rag patterns and all the LLMs and all that jazz. Yeah. Right. Okay, good. Thank you.

Nick:

Cool. Well, now that we have all of this IQ and all of these agents, how are we going to manage all of this?

Ulrikke:

It's not all agents, though. I mean, I one was really refreshed to see that it was not not all agents. And I think also that's a part of the communication this time. Is that it's not all about LLMs and chatbots and agents. This is actually algorithms at work structuring unstructured data. And we had a customer day the other day talk with about AI, and the I said, the if you're gonna take away one thing from this day, and I think this goes for Ignite as well, is that you've been collecting all companies throughout the world's been collecting data for a decade or more, and now is finally the time to capitalize on that data. This is when you actually get value out of the tera, tera tera, tera terabytes of data you've stored. So we're ready, right? We now have the tools, and it's not it's not all just about LLMs and chatbots. So yeah, where do you think that all of the new fancy agents are gonna be um managed? Yeah.

Nick:

Yeah. So of course, I think that's also what I've heard. This is of course we're leading into the other big announcement being Agent 365, which almost sounds like the next James Bond movie, like Agent 365. So which really interesting, it is that you know, that plane, that surface layer, that plane where you can if in an organization kind of get that centralized view of all the agents um within your organization, sort of managing those. And I think um not only that, it's also kind of a platform for like hosting MCP servers and things like that. So that was pretty interesting. That was something that really wasn't announced so loudly, but it was kind of the this is what Daniel actually showed me. He says within Agent 365, as opposed to creating uh a place to where your MCP servers will run, it will actually be a host for some of your MCP servers, which is really cool. Both Microsoft ones and then third-party ones as well. So if you're always wondering, okay, do I need to spin up uh um you know um an actual Azure resources for this? Well, you can now do that with an agent 365. But of course, managing all of these agents across the your organization as well, centralizing that, of course, you know, getting into DLP, you know, now we're beginning to weave in defender and purview and all of these things, which of course I talked earlier about security and data loss prevention policy. So this is all tying it in, making it much easier for admins, because I've even heard of stories of IT of IT um organizations or IT departments completely freaking out because all of a sudden people are creating agents, they're using AI. And if you shut it down, guess what? They're gonna use other AI tools around you. So you uh I have to say you have to embrace it and use these tools. And Microsoft is providing these governance and these security uh tools. Are they completely perfect? Not always, but they're definitely way better than letting someone use Chat GPT on on their own personal account to do work information or do um work analytics. Um and I already know this is happening. I was talking to a client and they're freaking out because they have external people that they can download data in Excel spreadsheets and they're doing analysis on their own AI tools, and they're like, okay, we need to stop this. We need to, but to stop it, we need to provide these people with their AI tools built within the context of our own sandboxes and security roles. So um AI is not taking our jobs, AI is loading on a whole bunch of new new jobs and workforce. Yeah, if you want to work in security, you know. Holy cow, tell me about it. Yeah. So um, I know the folk like people like Chris Huntingford, who's doing a lot of work on like you know, Chris will yell about purview till the top of his lungs and talk about red teaming. And if you ever get an opportunity to watch one of his sessions, this is what it's all about. And I think these are the things that's really important. Um, and you have people like that championing these these causes is really gonna make um organizations uh do this properly. So, of course, yeah, Agent 365 was a big thing. Uh, of course, with it a new icon, and we are gonna I think we have a news article about new icons as well. So if you've if you've seen the Agent 365 icon, it looks like a it looks like an A with a little dot in it kind of things, uh keep an eye out for that as well. If you're one of these sticker collectors, we were handing them out at the Ignite booth, and we had also Dataverse water bottles. So we actually had a bunch of these stickers and a bunch of Dataverse water bottles, so we were making Agent 365 water bottles.

Ulrikke:

It's like that with everything. If it's all sold out by Christmas, it comes around for Easter. That's always the same thing.

Nick:

Yes, exactly.

Ulrikke:

Just pretty much.

Nick:

Yeah, so did you have a different perspective on Agent 365, or did I miss something, or was that sort of No, definitely not.

Ulrikke:

And also we've kind of uh we alluded to the fact that there was gonna be big announcements uh at Ignite, and also there were some uh interesting and kind of leakages before, which was kind of prepared us a little bit for what was coming. But um no, I I saw the same thing you did, and also something that's been around since May, I think, uh the new agent um AD. So you'll have the an I an Azure Active Directory kind of thing, Azure ID, which means all agents get their own ID, as we do in the company. So you'll have agents on the same level as people, and that's kind of how it looks from the admin side, which is a bit stirring, I think, but also means that you can manage your agents. And I think the also another thing that I want to kind of that ties into what you were saying, you can manage all your agents across all the platforms from uh agent 365 within the Microsoft Stack. So even though you have agents from all around the other providers, you can all you can bring them in here, and then that's supposed to help with the shadow AI, which is a big thing. We talked about shadow IT for so long. Shadow AI is a massive problem just because of what you just said. And so the fact that you can bring this in, and also the new collaboration between Anthropic and Microsoft and Nvidia, those um collaborations was also mentioned in the keynote and also talked about adding night. These tech giants are now pulling together and collaborating, and you'll now have Claude Sonnet in uh Visual Studio and as a GitHub co-pilot, just as you would any other of those agents. It's gonna be a first party um helper, a first party copilot, if you want, um, to help you code, which shows you the direction that these people that these companies are growing in.

Nick:

And the other thing about the announcement there as well is that Anthropic will be running on Azure servers or of the Azure environments. So before they were strictly, I believe, on both Google and Amazon, but now Anthropic is going to be one of those tools that are going to be running across all different, all the main data centers, including Microsoft. And then the Microsoft ones will have unique capabilities that are unique to Microsoft as well. So definitely that was a huge announcement. That news that came out press release, I think the morning of the first day of Ignite. So of course that was highlighted as well as one of the one of the big news items coming out of Ignite.

Ulrikke:

So let's move to power platform news because we got a very interesting announcement too this uh time around.

Nick:

Yeah, so the morning of Ignite, I wake up. Now here I am on the West Coast, which is nine hours behind Europe. And the first thing I do, and this is I'm a geek, is I open up LinkedIn. Hey, what's new? Who liked my stuff, you know? And I just see everybody's talking about vibe.powerapps.com as a brand new thing. And it's very funny for the Europeans because it's only available on US-based tenants or US environments within the tenant. And I'm like, well, what? Like, I thought, okay, this is this is exciting. And so I actually went and tried it out myself. And wow, you can so if you've used the tools like Replit and Lovable, this is the Power Apps version. Now, so this is the other part that gets me a little bit wound up. So they talked about it.

Ulrikke:

I know what you're gonna say.

Nick:

The M365 app builder, which they talked about in Las Vegas, where it builds these little apps that talk to your SharePoint lists and la la la. Good, good, good. This is proper, it creates code apps. So we've talked about code apps before, but up until this point, you would create that code app as a React app and then upload it into the Power Platform environment. And yes, it could talk to Dataverse and do all the security and all that other stuff. So what vibe.powerappps.com is is really a front-end nice user interface to let makers build code apps, but using natural language. And I've played with this a little bit. And yeah, it's buggy, of course. It has to be run on a US-based tenant. It does, it doesn't always pick, it actually doesn't, it doesn't put things in your preferred solution in your environment. It does the little CR, whatever, underscores when it's creating tables. It doesn't always recognize all the common data model tables, like plan designers finally getting around to doing. And within a matter of minutes, when everything was working correctly, I was able to generate some pretty powerful apps. And it kind of blew me away. And yeah, wow. So what's been your experience out?

Ulrikke:

Yeah, I tried it out. And the and also it's if most of us have US environments, US early release environments. If you have that, you can try it out. And it's was available for me on the on the same day as it was released, and that's pretty fun. I think I I really do enjoy that they do that. So you can actually try it out. And in the beginning, the first run, the first part of the experience is very much like Plan Designer. And I expect that Plan Designer is behind it, or Plan for Power Apps, as it's now called, is behind it. It's also called plans in the UI where you go to to make your plan. And then the experience of the creating the app is pretty powerful. So it creates uh an app for you and you can see and you can iterate, of course, using natural language, but you can also double click and edit inline text if you want in the app, in the user experience, which is pretty crazy. And what I thought you were gonna say was. Where is the open in Visual Studio button? Yes. Where is the download code button? Yeah, you were? Okay. Because I was expecting you to go there pretty quickly.

Nick:

Because I've already rant ranted about that. Because this is something now with the Power Apps, the single page applications, um, where you can create a React app and upload it into PowerPoint within the context of PowerPages. You can download it into Visual Studio Code and start mucking around with it. This, I'm like, first off, I'm like, okay. There's no that was a thing. I gave the feedback at Ignite. Okay, I want an edit button and I want that edit button to open up Visual Studio Code. The people I talked to at Microsoft acknowledged that. They said, yes, the editor's coming, and yes, we have talked about. No, maybe I'm maybe I can't say that, but no, you probably can, but you did, and it's fine.

Ulrikke:

That's really cool. I mean, come on. So they do take feedback. Ah.

Nick:

Yes. So sorry, I should say yes. And so then the next thing I thought, okay, cool. I see. So here's here's the mad scientist in me going, okay, I can see that as a code app in my solution. So I opened up Visual Studio Code. I got my pack command. Okay, here's the pack command to upload a code app. Where's the pack command to download the code app? And I'm looking through Learn and I'm looking through the docs and I'm trying things and like I can't download, I can download a PowerPages single page application, but I can't download the code app. Because I thought if I could down, if I could download this code app into Visual Studio Code locally, then I can start because I have a few things I want to add some other libraries and things. And I thought, hmm, could I convert this into a PowerPages single page application and upload it into the context of PowerPages? Yeah. So but I can't do that. I can you can see the code. So if you want to do some copy paste crap, which I'm I think that's really clunky. But anyway, so these are the things these are the wheels that are turning in. I think not just my head. I'm sure everybody that's playing with this already, and already we're beginning to see with generative pages, people doing little workarounds and things like that, taking it to the next level. Um, and I think we talked about this last time, the fact that you know, feeding it parameters that it can show up in line within a regular record and things like this. I generative pages, I've been building something with that and just completely blown away. Still can't deploy to my Canadian environment, which drives me nuts, but we'll we'll talk about that. We talked, we ranted about that, yeah. So, anyways, but yeah, so vibe.powerapps.com realize it is in preview. So, yes, it is buggy, and no, don't create production apps. I put a very complex prompt that had a very complex data model, and it came up with a stinking pile of poop. But I was able to go in and basically within it's the very same kind of editor that you can move tables and add tables and those relationships. Spent quite a bit of time completely cleaning up, deleting half the tables it came up with, bringing in a lot of the common data model tables. And then once that's all wired up, you basically kind of you save that, and then the of course the tool will go through and you actually can see the reasoning and it can say, hmm, the user removed this table, and they said that was a requirement, so we'll just adjust this so it no longer takes that into account, blah, blah, blah. And you can look at that reasoning after the fact. So you not only get this magical black box creating you an app, but it gives you the reasoning and the breakdown of what it did to build it and put it all together. So I think it's still early days and still, I think there's still a world for model-driven apps, especially because it doesn't have the code apps to incorporate a lot of these features that model-driven apps have, even those little things and all that stuff, it's still very important. But we're, you know, we're getting to a point where it's going to be more of our system analyst skills, our architect skills are taking precedence in over for building these apps. And even from a maker perspective now, they don't have to mess around with the power effects and things like that. If they can extract the requirements in natural language, they can start building some pretty, some pretty interesting things. And Biff definitely, yeah, strap yourself like I've said this before, but strap yourselves in, kids who are in for a wild ride with uh vibe.powerapps.com. So I know yeah, so yeah, that was that was a big thing for me.

Ulrikke:

Yeah, definitely was. And also, but but also this is something we talked about, right? So the real world, even though Ignite comes around and these things are DA and it blows up in LinkedIn, the real world still goes on. And we had a customer that they need to go live before Christmas with a new, there's they're shutting off their old system and they need uh an a simple app for data entry for factory workers. We tried with multi-diven apps, it's not good enough. We know that when we try something in the US environment, we generate your pages, it works. We can use code app, it works, it looks great. We can't deploy it because it's not GA. We have to go with Canvas apps. And we've and we have one of our consultants, one of the guys on my team, shout out to Steag, he's done a fantastic job and learned so much about Canvas apps. He's like, yeah, if there ever was a case for Canvas apps, this is the that the this is it. This is gonna be used on a pad, on one of those secure cases in the factory where the guys are just gonna punch numbers. It's perfect. So there's still so if what I'm trying to say is if you feel like I should be embracing this brand new world and there's this whole bunch of AI stuff where we're here, but I'm still stuck using building mold ribbon apps and canvas apps, don't feel bad. We are still, because the machine has to go on and the real projects have to be done and the real apps have to be done. So if you're in a time crunch and you have a customer who has that requirement and has a deadline and you can't wait, yeah, you use canvas apps or mold ribbon apps, whatever it is that you need. So we're in this squish still, and it's just it's not gonna be until this stuff is GA and we can actually deploy it in production. And so, yeah, that's just the way how the real world works.

Nick:

So, as much as I love the concept of app.power apps or sorry, vibe.powerapps.com, the other what I think probably one of the best sessions I got a lot out of was this session that I mentioned it earlier, Power Apps Reimagining the Human Agent Collaboration. Um this actually is the recording is available online. It's actually free for anybody. You don't need to have registered even. Um, but it's been put, it was the it where did you go? Here you are. I'm sorry, I got screens all over the place. But it was presented by Mitty Jossie, Evan Liu, and our friend Claudio Romano, which he's a friend of ours. And this session just it just clicked on so many levels with me because it basically they were showing a model-driven app, uh, and they're using State Farm Insurance, which is one of Microsoft's customers who actually uses um a lot of this stuff. And of course, it was a model-driven app. We worked with model-driven apps. We've been working with model-driven apps for years, came from the dynamics CRM side, but it's now been more enhanced. So the very first thing when you look at it is doing sort of data analytics. So that's sort of when we talk about apps. So, yes, we have all this data in. We've been entering it through forms and views, maybe importing it or integrating with other systems. But doing analytics with this, we've always had to go and write Power BI reports, or we create charts, or we create SSRS reports. Well, now that data, and even if people are still working with it, we can now get agents to analyze this. And I can open up Copilot within my model-driven app and say, and this is what Claudio demoed. He goes, hmm, there seems to be an interesting trend around these certain dates. And it will actually look at this data and create summaries and analysis. And, you know, this way, in terms of reporting, as opposed to it used to be, hey, can you write me a report or give me a dashboard or something? Now our end users that want to see this data can actually ask it how many members signed up in the last two months, how much you know, project hours have been billed. These are the types of questions we can now ask our systems, which is really cool. So that was like the first thing they showed off within this session, which kind of like, yeah, this makes sense. So this to me is how AI is not taking over, we're not throwing out the old apps, but it's enhancing our existing apps, which is really cool. Now, the next one is something that we've already been playing around with, I've been using, absolutely love, is data entry. So it's very tedious for people to kind of key into all this information, get it into Dataverse or into Dynamics. Yeah, we've always had like the Outlook, Outlook app where we could track emails that get attached to opportunities. For some reason, way back not even that long ago, but Microsoft had the business card scanner built in as an agent builder thing. And like, who uses who uses this anymore? But what they showed was so there was the form filler, which is, I think, appeared in most model-driven apps now. If you haven't tried it, just try it out. You can actually copy a bunch of stuff and paste it in, and it'll try to fill out that form to its best of ability, which is pretty cool. So cut and paste, you know, Windows V key is your friend in terms of cutting and pasting. But now they've added, or there's new features that are rolling out as in preview, but you can, I think it's actually there where you can upload files. So if you have a bunch of a PDF or something or a document, you can click on that, upload it, will fill in the form that exists. But then they have a new button that's showing up. You can click on that and it can take a look at your emails, your team's chats. You can go into SharePoint folders. And then you can actually take that. Oh, there's an email from Claudio. I'm gonna click on this, I'm gonna feed that into my data filler. It's gonna fill up the form. So they were showing their insurance, their insurance agents that were sending in these quotes or these quotes from repair shops as attached PDFs. So they're able to go into the form filler. Hey, here's that email from this shop. There's an attached quote as a PDF, and it just filled in the form. That was cool. So then the next level, and Mitty showed this, which is really cool, is oh, but it fills in the number a certain way. So she was able to go in and you can actually teach the form to act in different ways. So every time the form fills in a certain value, strip off the prefix or modify this number a little bit. So now the form knows that it actually will take that data and manipulate that. So, and this was like they said it was hot off the press, and maybe I'm not I'm talking a bit out of school here because it was a demo. They said it was demo where they did announce this, but Claudio said the pull requests were done that morning before they did this session. So this is how new it was. But it was amazing that all of this worked.

Ulrikke:

So you kind of data transformation in real time on your forum, right? So data transformation is is always hard and clunky, and then now you got it real time on your forum as you're pacing in. The form knows about the data transformation before it's even entered. That's that's pretty cool.

Nick:

But but wait, there was more. So already like my mind is completely like, oh, this is so cool. So then they start talking about what the other announcement was the Power Apps MCP server. So they're like, that's great, but somebody like a state farm, they're getting they have a shared email box that gets thousands of these requests on a daily basis. So before, someone would have to go in and actually like re-key this all in. Like, okay, cool, now I can go in the form filler and click my email, but now they're gonna be clicking all day. Wouldn't it be great if there was a way that it would just do this all automatically? So they on stage created an agent, and this agent basically was triggered to run every time a new email showed up in that dedicated email box. What it would do is they gave it instructions and they gave it actually the Dataverse schema because they said this code is so new it really needed the schema, but they said when it goes more to public preview or GA, you can you don't need that level of detail. But then, so that was the triggering, and then they added a tool, they added the Power Apps MCP server for the data entry tool. And basically that's using the same data entry agent. So all the training that they did on the form entry, this agent will go through and start triggering every time an email goes into that box and fill in the form, fill in the data automatically. So if they get a thousand emails, it's going to create a thousand of these records and fill in all the data correctly to the best of its ability. And you're like, okay, but that's again black box behind the machine behind the wall, what's going on. So this is where the agent feed comes in, and the agent feed you can actually go through and monitor. So you could actually click on it, you could see here is the PDF, and here's how the form was filled out based on the PDF. Now, sometimes the agent would stop and go, hmm, I'm not confident that this was the right data entry. So it actually puts almost like a hold on it. So you have to go into the agent feed and see the records that require human intervention, the human then loop to review. And you can go in and review, okay, this stopped because it couldn't quite read the discount amount that was written in. So you can actually go and make that adjustment automatically. And once you're good as a human, you can click on it, it will continue on with the process. So, or you could even add approvals with that, that these have to go through a certain approval and that kind of thing. So it just to me, it was okay, this is where I can now go to my existing customers that are using model-driven apps and say, How long are you taking to fill in this information? Here's the form filler. Hey, is this coming in on email? Let's set up an automation or an agent that's going to read that email. You're still a little bit not confident about it, no problem. Here's the agent feed. So you can actually go through as opposed to re-keying, you can now go through and approve and build up that confidence and build up the training as well in terms of how the this model works and make adjustments and then get to a point where, yeah, but also kind of have it raise flags or instances where it doesn't quite meet the criteria. So instead of spending hours doing data entry, now your job's focused more to dealing with those exceptions, which always you had to do anyway, but all within the context of a model-driven app that you might have been using for years, but now you're using agents to enhance that. So to me, this was the reality of stuff coming into the power platform using agents and AI. And again, big shout out to uh Mitty and Claudio and Ewan, or sorry, Evan, Evan, not Ewen, Evan, who just did probably one of the best hidden gems sessions of Ignite, in my mind anyway.

Ulrikke:

That's just mind-blowing. And I mean, the level of detail that you have to get into to kind of because now it's it's I get the feeling that we've been generalists up until this point, right? So it's all been, wow, we can now outsource these tasks to agents. Now we're getting into the details because people are actually diving in this deep. And then I love to see that. And also, this also ties into, you know, when because it's not far off until they can do whatever we can do, right? So they can key in the data, they can key in the data based on all a whole whack of different sources. And also, if none of that is good enough or you needed to go grab data, fresh data from a, I don't know, legacy on-prem system somewhere. You used to have RPA for that, robotics, automations, and now you have computer use. So if if I had to pick something that blew my mind, it has to be so, and of course, because I'm meet, I get to choose two things. But I just want to highlight CUA. If you if you hear anyone talk about CUA, computer use agent, it is uh a way to have an agent act like you on a computer and understanding the user interface. You give it natural language things, log into this website, create a new, register this, uh, or grab this data, search for this or find this or filter like that and grab that data. And then you can have it trigger something else or put that data somewhere. So it's kind of on the extension of what you were saying. If if none of that solves your case, then you could actually use computer views. And the benefit of KUA kind of if you compare it to robotics automation, is that if the UI changed and you recorded the UI, pushed the button, it will push the black space because the button moved. CUA understands that. It understands that, ah, the say button moved or the registration button moved. It reads the information, it deducts what to do, and it's more independent in that way. So it can actually work with the changes. Um and also wanted to highlight one uh session that I saw. Um, I'm not gonna go into detail, but if you it's kind of it's got your advanced agent development with Copilot Studio, and it's if you need to see, if you have 45 minutes and you need to see everything that Copilot Studio can do, you watch that session. It goes through everything from the simplest uh configuration that you can do when you create a Copilot agent, and all the way to Kua in 45 minutes, and you follow along, and it's in the pure kind of Scott style, the way that he presents is always fun and light and understandable. And then and I also got the whole thing, and it's multi-agent orchestration, it's code app, it's uh code interpreter, it's the it's the whole kind of spectrum, it blows my mind. So, yeah, that's that's one to watch.

Nick:

Yeah, I did I did say I was gonna go and heckle him on that, but I was stuck on um expert booth duty that it over cringe the time. And then I realized even if I had time, I looked at the recording later, that room was packed, of course. Everybody had the headphones on and everything like that. So I mean, yeah, so this is why like the fact that there's recordings there, you can go back later and then I can just harass them in other ways. So it's all good.

Ulrikke:

Yeah, yeah, absolutely. And we'll put the links to that in the show notes as well with with your the session that you mentioned as well. And then I must just I I want to just touch on something that is not related to PowerPlatform. It's it's not related to anything. It's just I just I just it's all this in the keynote. I was like, wow, all right, so we're done. Right, so we got everything. Right, so gardening then, because did you see the migration agent with GitHub Copilot in the keynote? It's the far end of the keynote. They go in, they grab so there's this guy. The thing is with the scenario is we have a lot of legacy apps, right? And a lot of our customers have that. You have something built, something custom somewhere some uh some time ago, and you need to refactor it, you need to move it. What do we do with it? I was on a project three years ago refactoring 109 homegrown applications built on Delphi. Right. These exist everywhere. So what you do is it's tedious because you need to go through it, you need to have build the business case for it, you need to get the funding for it, and then you need to actually document it, what it does and talk to all the stakeholders, and then you need to refactor it and move it and deploy it and everything. Meet migration agent, which lives in Azure, and you give it the background story, you tell it in natural language what it is that you need to do, where the app is stored, what it is doing. If you have code, you can upload it. If you have documentation, you can upload it and just share what it is about. It helps you build the business case. That's the first thing it does. It helps you bring the build the business case and what you need to do in order to get your stakeholders and the sponsors to approve that this is something that IOI-wise is a sound investment. And then it asks, right, so would you like me to document this then? If you got it approved, done. You want me to document it? Yes, sure, please do. And it goes through all the code and it and it documents the app. And then it says, uh, right, so you want me to connect to GitHub um co-pilot to refactor this app and to rewrite it into something else. So we recommend this architecture and these libraries and these and this. And you go, oh, good suggestion, why not? And it goes ahead, it connects to GitHub Copilot, and it creates the new code, refactors the whole thing, and writes the code again, and it allows you to test it. So then you can t write test automatic testing for it, it has it all documented, and you can just deploy it to Azure and have the documentation and and all your stuff in your repo and GitHub and everything sorted for you. And and then I watch this on the keynote and I go, right, this is PowerPoint. This can't be real. This isn't because this is kind of if if this is real and this is usable now, we're done. Because all the new stuff agents are gonna do and all the old stuff is already moved, then there's nothing left, right? Where it's like this. And I talked to one of my colleagues, he's an AI and Azure guy, and I said that to him. I said, if this is real, then we're done. So I don't believe that this is real. This is just D aware, this is PowerPoint. And he's like, Well, actually, even if even if this is just PowerPoint, even if it's just MOWER, but it's able to what we know the LMs can do today or a general agents can do today, it can document very well, it can assess very well, it can build a business case really well. All of the things that is not code, if it does that well, like we know that it will, because that works today, if the code is 60%, it's better than anything you do today because that's so time consuming that this is just this is huge.

Nick:

Yeah. Yeah. And I was but yeah. But I still think there's having refactored apps before there's yeah, because it it can be tedious, whatever. But I mean, but if you have something that works, that runs on old technology, then for sure this is a and yeah, I've been I've been in projects where we converted old a whole pile of access databases and we brought them all into Dataverse and did the, you know, kind of replicated the user interfaces and everything. I think what but again, if you're if you're a solution architect and you're guiding that process, then you as the hu the value that you bring is going, yes, this is how the app always worked. And yes, AI can take that old Delphi app and boom, vibe.powerapps.com, and here's your here's your new working app. Yes, it does all the the business case and everything in between. And it can go through the whole process, like we know all the co-pilots and chat GPTs can do. But the real value there is cool, it did the business case, but this is where you want to even working with the AI. Okay, I'm gonna challenge you on this one data entry point. People have been collecting like fax, fax numbers on forms. If you just tell an AI to convert uh an old application, it's gonna create the new one. Guess what? That fax field is gonna be there because no one told it along the way. We don't need faxes anymore. We don't need the integration to the fax system. We don't need all this. So this is where I think you as a business analyst or solution architect can go through the process and say, okay, we don't need fax anymore, but we do need this integration to this other new service. So yeah, we're not writing necessarily the connector or that, but we're telling, we're guiding that AI to do that. And then, yeah, but then the the cool thing too is we think, okay, well, that's great. You're taking away my job. But guess what? All these companies that have a bazillion historic or legacy apps, why aren't they converting them now? Because they don't have the time, they don't have the resources. So finally, if we're actually have these tools that help us, and like I said, we're guiding this process. Hey, that old app that yes, it works, but it's running on that old clunky machine that's underneath someone's desk that could fail someday that you can't replace the hardware for. I'm this is an extreme example. But hey, with the tools we have now, we can actually convert that app in a matter of weeks. Where if we talked a year ago, it was going to take a matter of months and a lot of resources to do this. So yeah, it is this is the case of things are changing, but also I think the skills that we have as developers, as solution architects, as business analysts, and like we talked about last time, the forward-deployed engineers, that mentality, this is this is going to keep us overemployed. Like I said earlier in the episode, AI is not taking our away. AI is now adding a whole bunch of stuff. So now I'm no longer a maybe a developer that refactors old systems. I am a solution, I am a migration analyst working with the AI tools to migrate old systems. Maybe I'm completely wrong, or maybe I'm being completely naive and on this, but I also see it.

Ulrikke:

It's gonna take our jobs away. That's not my point at all. I think because what usually happens is it's a lift and shift, because usually the organization, and that doesn't have to do with technology or us. It has to do with the the people in the company because they can't be on top of lifting and shifting and also so refactoring, but and also improving. We're often, more often than not, if when we are in those uh projects, it's the let's just lift and shift this as is now, and then we'll make changes later. But by the time you get there, something happened along the way, and it was all these integrations that you didn't account for, and all these little small details that you couldn't know about, that no one knew about, and all the skeletons that come out of the closet, they eat all that time. And then you and then you get in kind of when you want to go in and and do the changes and make sure that you modernize and and take account for how things work today and the new capabilities are in the new platform, for instance, the motivation is gone and it's a bit sour and all the money is spent. So if this can help to get that tedious work out of the way, so that we and I had a had a conversation with a friend the other day saying she was scared for her job. And then actually, as we were talking, discovering that no, this means we get to do more of the fun stuff. And we talked about this before. You take the tedious things out of it, the the amount of hours it takes to manually go through a whole app and document it when you've never seen it before, just to get it to do the the groundwork for you so that you can actually read through and see all its contents and how it fits together. That work alone is worth so much compared to doing that work yourself. And so I think it's what my point is that this will this will actually propel us into new development and doing the good stuff and actually being able and having the time and the finances to make these apps better and customize them to the new processes or the new ways that they want to work. Um, so yeah, that I think that's what this was got me so excited because it is an enabler to get speed up and to to build that business case and to get up and running faster so we could get the fun stuff faster. I think that's what got me excited.

Nick:

Yeah. For sure. Yeah, so that yeah, I mean it's it's funny because uh just due to logistics and me on the expert booth, I didn't get to attend the keynote in person, but I did, you know, so I was but I yeah, because that was the other thing with Ignite, did the keynote was held on the other side of the city and there was busting and everything. But shout out to the folks that they're shout out to again, my friend Lisa Crosby, she got up at 6 a.m. to make sure she got her spot at the keynote.

Ulrikke:

Yeah. So yeah, I I must think it's worth it. You look at that stadium, that kind of that space, that physical space, and the screens and the people and the mu and the sound, and I just was just buzzing all the way from Norway looking at it. Oh, that's so freaking awesome. And I wish that I was there. But yeah, let's wrap this up because we're way over time and we've touched on so many new things and news and updates. And I think we'll just leave it at this. Um, and we'll leave the show notes for you where you can go in and you can watch these recordings back. It is worth I I found because I I've looked at the keynote highlight videos out there and the the sum-up. I I would recommend you actually taking the time to watch this keynote. For me, it was worth watching, worth the time.

Nick:

And you can do it at the double speed or triple speed too, if you wanna go quicker through it too. That's the benefit of the recordings.

Ulrikke:

Yeah. That's true. And then we'll cover all the news and uh the new um the events and all that stuff next time. You got a bit a big update last time, so you can watch that back if you need that. Next episode, you put the date in here, is gonna be November and 19th. No, December 19th, which is gonna be our Christmas Christmas episode. So keep an eye out for that and have a lovely time until we see ya. Bye bye. All right. Thanks for listening. And if you like this episode, please make sure to share it with your friends and colleagues in the community. Make sure to leave a rating and review of your favorite streaming service and makes it easier for others to find us. Follow us on the social media platforms and make sure you don't miss an episode. Thanks for listening to the Par Platform Boost podcast with your hosts, Ulrike Ackerbeck and Nick Dolman, and see you next time for your timely boost of Par Platform News.

Head Shot

Nick Doelman

Host
Head Shot

Ulrikke Akerbæk

Host