Exploring AI, Power Virtual Agents, and the Power Platform: A Journey with Sarah Critchley, Microsoft's Principal Program Manager

WEBVTT
00:00:02.505 --> 00:00:04.350
Welcome to the Power 365 show.
00:00:04.350 --> 00:00:10.467
We're an interview staff at Microsoft across the Power Platform and Dynamics 365 Technology Stack.
00:00:10.467 --> 00:00:16.486
I hope you'll find this podcast educational and inspire you to do more with this great technology.
00:00:16.486 --> 00:00:18.887
Now let's get on with the show.
00:00:18.887 --> 00:00:26.650
In this episode, we'll be focusing on the amazing tool that is Power Virtual Agents.
00:00:26.650 --> 00:00:30.685
We might touch on things like co-pilot, ai and all those lovely things.
00:00:30.685 --> 00:00:32.506
My guest is from Canada.
00:00:32.506 --> 00:00:35.488
She works at Microsoft as a principal program manager.
00:00:35.488 --> 00:00:37.627
She's a writer, speaker, author.
00:00:37.627 --> 00:00:39.767
Multiple books to her name.
00:00:39.767 --> 00:00:41.286
Check them out they'll be in the show notes.
00:00:41.286 --> 00:00:46.789
She truly believes that customers and employee experience should be at the heart of business.
00:00:46.789 --> 00:00:50.545
You can find links to her bio, social media, show notes, etc.
00:00:50.545 --> 00:00:53.067
In the show notes for this episode.
00:00:53.067 --> 00:00:55.067
Welcome to the show, sarah.
00:00:55.661 --> 00:00:56.524
Hi Mark, how are you?
00:00:57.462 --> 00:00:58.887
Good Good to have you back.
00:00:58.887 --> 00:01:06.067
I think last time we had you on the podcast was before you worked at Microsoft.
00:01:06.067 --> 00:01:07.203
You were being an MVP.
00:01:07.203 --> 00:01:27.186
Just as a bit of background, sarah and I have spent some time together our families in the UK when she used to live there, but so the audience can get to know you, sarah, tell me about life in Canada, how you ended up in Canada, and are you safe from only the bushfire reports that we're hearing going on in Canada at the moment?
00:01:27.822 --> 00:01:28.525
I am safe.
00:01:28.525 --> 00:01:31.368
It's a pretty terrible thing what's happening in Canada.
00:01:31.368 --> 00:01:33.006
It's actually happening all over Canada.
00:01:33.006 --> 00:01:36.426
There's a huge amount of wildfires, unprecedented in the scale.
00:01:36.426 --> 00:01:50.064
It's definitely good to raise awareness about what's happening in Canada Because, even though I'm in a good place in terms of we don't have any drastic bad things happening in Vancouver, you can see the smoke.
00:01:50.064 --> 00:01:59.227
You can actually see the smoke traveling so far, which for other parts of the world, that's, people don't really understand that and how far these kind of things can travel.
00:01:59.227 --> 00:02:01.367
But Canada is a beautiful place.
00:02:01.367 --> 00:02:07.945
It's so amazing to be able to just go out on a hike on a weekend rather than spend an immense amount of money to do so for a holiday.
00:02:07.945 --> 00:02:10.167
It's a true privilege, so I love it here.
00:02:10.920 --> 00:02:13.087
How did you end up moving from the UK to Canada?
00:02:13.800 --> 00:02:21.400
I was lucky enough to get a job offer to move actually, so I moved for work and truly grateful for that opportunity.
00:02:21.400 --> 00:02:28.829
It's amazing actually and, mark, I mean you've done it yourself right in terms of lived in another country and definitely there's things you take for granted.
00:02:28.829 --> 00:02:36.350
There's definitely lots of life lessons learned when you move and so, yeah, it's been truly life-changing and amazing.
00:02:37.021 --> 00:02:37.283
Nice.
00:02:37.283 --> 00:02:40.679
Tell me about your journey into Microsoft, because you're working for Microsoft now.
00:02:40.679 --> 00:02:41.865
How did that come about?
00:02:41.865 --> 00:02:43.278
Is it amazing?
00:02:43.278 --> 00:02:44.965
Is it the you know?
00:02:44.965 --> 00:02:46.068
You hear?
00:02:46.068 --> 00:02:50.606
The term that you've landed a job at the mothership is a common term out in the community.
00:02:50.606 --> 00:02:52.525
You've become a blue badge.
00:02:52.525 --> 00:03:00.967
You lost your MVP not because of an NDA violation, but you have joined Microsoft and Microsoft employees cannot be MVPs.
00:03:00.967 --> 00:03:02.525
What was that journey for you?
00:03:03.241 --> 00:03:04.519
Yeah, it's kind of funny, isn't it, ray?
00:03:04.519 --> 00:03:10.503
It's almost like people see it as like kind of like a career path, like especially as an MVP right, it's almost like okay.
00:03:10.524 --> 00:03:12.169
So your next step is going to be Microsoft.
00:03:12.169 --> 00:03:18.727
My experience has been truly amazing, Like it has been again super fortunate to work where I work and in the team that I work in.
00:03:18.727 --> 00:03:27.623
You did interview my colleague Dwayne recently as well, so it's an amazing kind of experience, because one of the things I truly value about working here is the collaboration.
00:03:27.623 --> 00:03:34.680
It's definitely a massive difference from where most of my career actually was, which is professional services, right.
00:03:34.680 --> 00:03:44.039
So coming from that kind of world into primary tech firm is definitely a change, One that I'm actually grateful for, right.
00:03:44.039 --> 00:04:00.026
So it's definitely really really good and it's quite nice where I've been able to leverage my background in dynamics and actually kind of bring that over to the conversational side, to the conversational AI side and everything we do in dynamics, right, and in business applications and all of that, and that's been truly useful for the role.
00:04:00.721 --> 00:04:11.659
Awesome we touched on it at the start there on your books that you've created, and always find it intriguing the life of an author and just how wealthy they become from writing books.
00:04:11.699 --> 00:04:12.625
You don't do it for wealth.
00:04:15.526 --> 00:04:16.610
You've created a tech book.
00:04:16.610 --> 00:04:18.346
Would you do it again?
00:04:18.346 --> 00:04:19.752
What were your thoughts?
00:04:20.860 --> 00:04:26.492
Oh, that's a hard question, you know, because, like this day is, I wake up and I'm like I've got so much to say.
00:04:26.492 --> 00:04:28.283
I love writing.
00:04:28.283 --> 00:04:34.276
It's how I learn myself and I don't know, to be honest, it depends what kind of day you.
00:04:34.276 --> 00:04:35.439
It would depend on what.
00:04:35.499 --> 00:04:38.524
I don't think I've write a purely technology book anymore.
00:04:38.524 --> 00:04:42.271
Like kind of the book it was really a guide on dynamics.
00:04:42.271 --> 00:04:48.951
It was around the time dynamics suddenly became very app focused Right, kind of back in the day it used to be fairly flat, right.
00:04:48.951 --> 00:05:02.661
You know, you would have your modules, you could learn the entire like kind of dynamic stack right and be an expert on dynamics 365, even though it was called something different, right Whereas it was around the time when it was really more about the depth of knowledge in each of the apps.
00:05:02.661 --> 00:05:07.560
So the book was meant to be something that you could hold in your hand and be able to at least be able to get started with it.
00:05:07.560 --> 00:05:09.930
But I don't think I would do that anymore.
00:05:09.930 --> 00:05:17.471
From a tech perspective, if I was going to write another book, it would probably be something a bit more kind of human centred or something a bit more long lasting.
00:05:18.081 --> 00:05:22.720
Yeah, so how to survive an office romance or lifestyles of the rich and famous at home?
00:05:23.019 --> 00:05:24.324
Lifestyles of the rich and famous.
00:05:24.324 --> 00:05:25.608
That was a good Charlotte song, wasn't it?
00:05:27.160 --> 00:05:33.271
Yeah, I just think it's a labor of love really writing when it's you know you've got a publisher on your case.
00:05:33.271 --> 00:05:45.247
And that tech sector the lifespan of a tech book right, is that the technology is changing and back when you know it used to be every three years, would get a product update or release back in the early days of dynamics.
00:05:45.247 --> 00:05:47.733
Fine, at least you got three years of runway.
00:05:47.733 --> 00:05:49.701
Yeah, Now you get what?
00:05:49.701 --> 00:05:51.345
Two wave releases a year.
00:05:51.345 --> 00:05:52.846
It changes a lot.
00:05:52.846 --> 00:05:58.735
It's hard to maintain that a book in a physical form when it changes so much.
00:05:58.959 --> 00:06:00.081
That's the problem, right?
00:06:00.081 --> 00:06:15.329
It's funny because I remember when, back in the dynamics days, I remember when and you're spot on right you would have such a massive amount of runway, and I remember just like you would be waiting for the day that the feature that was promised would be dropping because all your customers would want it.
00:06:15.329 --> 00:06:20.607
I remember that distinct to area was business process flows when they came into the dynamic.
00:06:21.028 --> 00:06:30.293
Yeah, everyone was like, oh wow, these are finally going to be in the product, the new like kind of UI, right, we're going to do so much stuff and you know it was that excitement that can be seen.
00:06:30.293 --> 00:06:33.165
You know, it's almost a bit different now in relation to that.
00:06:33.165 --> 00:06:34.749
It's a continuous update cycle.
00:06:34.749 --> 00:06:41.672
It's definitely something that people need to kind of like switch their thinking around when they're thinking about like upskilling, a learning.
00:06:42.480 --> 00:06:52.648
Yeah, interesting those business process flows, I just think, you know, in their day they were absolutely phenomenally amazing and even just to hit a project recently that used them and I still think they're amazing.
00:06:52.648 --> 00:07:04.754
I think they could be extended on massively from a BPM, you know, business process management perspective and more, you know, visually enriching, but, yeah, cool feature.
00:07:04.754 --> 00:07:14.161
A lot of people found it hard to get their head moving away, remember, from the old dialogues into this kind of metaphor as a way of grabbing multiple step.
00:07:14.281 --> 00:07:15.923
You know engagements.
00:07:15.923 --> 00:07:33.375
Anyhow we digress, let's talk about the amazing technology of power virtual agents and why every person in the power platform ecosystem needs to gain experience with PVA and think about interface design in a different way.
00:07:33.375 --> 00:07:42.396
Like there's been a massive index on power apps, for example, let's build an app for that and that we're wanting a screen based experience.
00:07:42.396 --> 00:08:04.315
And with PVA, it opens up different dimensions of engagement which, yes, you know, you've still kind of got a micro app experience in that you can put adaptive cards and stuff, let's say, in a Teams context, but it's a different way of communicating with data, extracting data in a really timely fashion.
00:08:04.315 --> 00:08:08.367
When you explain PVA to people, how do you explain it?
00:08:08.367 --> 00:08:11.093
What do you suggest they do with the tech?
00:08:11.500 --> 00:08:14.005
Oh, wow, how long we got Like kind of I would.
00:08:14.005 --> 00:08:20.379
I can explain it in a few different ways, but I think probably the best way to think about it.
00:08:20.379 --> 00:08:37.118
And actually, mark, you talked about something very important and that is that the difference between an app right so you know, we're so used and this isn't just a power wrap thing, right, this is something that, when you would make WinForms apps and even a website right, is technically an app.
00:08:37.118 --> 00:08:38.640
Right, you've got an interface.
00:08:38.640 --> 00:08:44.389
You use essentially a keyboard, a mouse, a screen reader or whatever to essentially use it.
00:08:44.389 --> 00:08:56.735
Right, and the very important difference, when we're talking about conversational AI and as much as that word can be super intimidating for people who are not in the industry it shouldn't be right, like, definitely recommend to kind of get an educated on it.
00:08:56.735 --> 00:08:59.261
You're watching this, you listen to this, so that's step one.
00:08:59.261 --> 00:09:04.876
Right, it's you're using your voice to interact with it, right, so this is about not using your hands.
00:09:04.876 --> 00:09:11.346
Right, this is about being able to almost like, just use language to get what you're looking for.
00:09:11.346 --> 00:09:18.043
And if you think about that from an enterprise perspective, there's something just truly amazing and accessible about that.
00:09:18.043 --> 00:09:24.664
That is so, so, very important and that should really always stay at the heart of everything we talk about.
00:09:24.705 --> 00:09:27.871
When we talk about conversational AI, because it's about enablement that.
00:09:27.871 --> 00:09:48.514
You said it kind of like micro app experience, for sure, because most people will probably be used to like chatbots on a website, right, but the idea, and especially as you start to, you know everyone would have heard Satcha talking about co-pilots, right, and you know this idea of this co-pilot experience and how AI is embedded in so many different Microsoft applications today and it's the same thing, right.
00:09:48.514 --> 00:09:55.332
Well, what we're trying to do is we're trying to make the fact that you don't have to spend a year trying to learn an app right, like, kind of.
00:09:55.332 --> 00:10:02.696
I mean, everyone surely can empathize with this as their own consumer, right, and you know, as a consumer, you're going to have, like, an internet service provider.
00:10:02.696 --> 00:10:06.562
You're going to have, you know, other service providers where they're probably going to have an app.
00:10:06.802 --> 00:10:13.421
Right, you know your bank will have an app and you need to learn that interface to figure out oh, how can I ask for help on this app?
00:10:13.421 --> 00:10:15.283
Where do I check my balance on this app?
00:10:15.283 --> 00:10:23.024
And the whole idea is a fact that you can just say what's my balance and you can just say, hey, I need to raise a support ticket, right.
00:10:23.024 --> 00:10:32.784
And the truth is like Mark, you and I will say those things slightly differently, maybe completely differently, but the whole idea is that the from a conversational AI perspective, that should not matter, right?
00:10:32.784 --> 00:10:39.003
Like you know, you're not hard coding slightly different differences in the fact that, oh, sarah said it this way and Mark said it this way.
00:10:39.490 --> 00:10:57.302
The fact is that the technology, the AI behind it, knows and can understand and derive the intent from what's being asked and then can then go ahead and perform the action right, and that's a truly fundamental difference as opposed to what should I put on this UI to make this kind of understandable for a user.
00:10:58.070 --> 00:10:59.836
How do people get started using PVA?
00:11:00.610 --> 00:11:01.852
Oh, wow, that's like I mean.
00:11:01.852 --> 00:11:08.004
There's so many different ways, so one of the biggest, the easiest ways is our PowerCat team did a playbook.
00:11:08.004 --> 00:11:29.720
So if I think about this from the consulting element and also from a customer perspective, to me honest this is so useful for everyone, not just implementing organizations that want to use it internally, and if you're like a Microsoft partner or something like that, the PowerCat PVA playbook will be the best way and we'll put those in the show notes for your audience and that actually takes you from start to finish of a PVA implementation.
00:11:29.720 --> 00:11:37.070
So, and there's actually a massive part of that playbook that covers the actual use case, like the actual business analysis.
00:11:37.070 --> 00:11:39.057
I can't again stress that enough.
00:11:39.057 --> 00:11:54.070
This is the make or break of any software development like project, and it's when people do not spend enough time understanding what business need is being worked upon right, and so the playbook actually talks about that quite significantly and we actually embedded that playbook into our PVA in a day our latest update for PVA in a day.
00:11:54.070 --> 00:12:03.070
So that's also another way that if people are a bit more hands on, if people like Sarah, I don't really want to read like a 10 to 20 page PDF document please go ahead and take a look at PVA in a day.
00:12:03.514 --> 00:12:08.059
We actually kind of really squished that down, we actually summarized that and made that really succinct.
00:12:08.059 --> 00:12:13.844
It's not dependent on each of the modules anymore either, so you can like kind of just pick up a module and kind of go.
00:12:13.844 --> 00:12:24.232
If you really are limited in time, go ahead and start with module three, right, go ahead and start with module four, whichever one will get you the biggest ROI, and then the final one will probably be.
00:12:24.232 --> 00:12:39.070
There is actually a conversational UX guide to the Microsoft have made, which also has, like again, we actually talk about this in a few of our materials, but again it goes back to the human element, right, because ultimately, you're still designing an experience for a human being, right?
00:12:39.070 --> 00:12:53.648
And so as much as we're trying to take away the kind of almost the technology interface and make this very kind of you know, familiar with language, there's still lots of differences that people can bring, and so the actual CUX guide is really good as well.
00:12:53.648 --> 00:12:53.990
Yeah.
00:12:54.850 --> 00:13:04.083
Soundbite that I just took away from that is the conversational UX which is I had to even picture that out of my mind.
00:13:04.083 --> 00:13:10.817
You know, and I think it's such a great way of looking at this, that it's not just a, you know, it's not just a very programmatic.
00:13:10.817 --> 00:13:12.802
You know, ask a question, get an answer.
00:13:12.802 --> 00:13:21.364
It's the nuances around answers and it's a nuances probably more around the question, especially when people ask multi-part questions.
00:13:21.364 --> 00:13:36.116
Yes, and understanding that, and one that comes to mind if I would want to validate an AI is I'd like to book flights to Vancouver on the 20th of the month and I want to find first class.
00:13:37.070 --> 00:13:41.224
You know, there I actually asked a whole range of things.
00:13:41.224 --> 00:13:58.873
Yeah, I asked scheduling things, I asked destination things, I asked service level of product things and that was just one question, and so many of the historic bot technology out of there needed question answer, question answer, question answer.
00:13:58.873 --> 00:14:03.734
Rather than understanding there were multi-part question and answer is a multi-part answer.
00:14:03.734 --> 00:14:11.683
How is that changed now in the context of large language models where it can understand that nuance of conversation?
00:14:11.683 --> 00:14:16.647
It has that context to know that that was what you were asking.
00:14:16.647 --> 00:14:20.625
Are you seeing a bump in experience from that?
00:14:21.518 --> 00:14:22.080
Yeah, for sure.
00:14:22.080 --> 00:14:27.844
It's really interesting because the example that you just provided is very similar to what we actually demonstrated at Build.
00:14:27.844 --> 00:14:30.302
We actually launched one of our latest features in PVA.
00:14:30.302 --> 00:14:32.019
It's in limited preview right now.
00:14:32.019 --> 00:14:32.822
That's progressing.
00:14:34.157 --> 00:14:40.000
It's called Generative Actions, so that is essentially really what you just described, which is like chaining technology.
00:14:40.000 --> 00:14:48.817
This is around how we're using this technology to really create again an even more natural experience for the user, because the truth is, you're spot on right.
00:14:48.817 --> 00:14:59.605
You're not going to and you shouldn't have to pass your own questions to a bot, because then again you're having to be trained to talk and it's just the same as using an app.
00:14:59.605 --> 00:15:04.806
We have to manage these natural differences in ways we talk.
00:15:04.806 --> 00:15:05.668
I'm the same.
00:15:05.668 --> 00:15:09.395
I would tend to ask 20 questions in one big paragraph on Teams People.
00:15:09.395 --> 00:15:12.604
My colleagues probably hate me for that, but it is what it is.
00:15:14.196 --> 00:15:19.219
That chaining technology is really important because they almost infer context from one thing to the next.
00:15:19.219 --> 00:15:27.081
So the other thing that's really important is the things that you didn't say in that statement, so you didn't actually say where you were flying from.
00:15:27.081 --> 00:15:36.225
There's actually missing information for the actual technology to complete the query, so there's lots of things that really need to be factored in.
00:15:36.225 --> 00:15:45.043
One of the amazing things and again we'll put it into show notes to our build video because it's actually amazing it's near the end and it was booking a cruise.
00:15:45.043 --> 00:16:05.447
This is where it was missing the information the departure port and it actually used generative AI to actually generate the question of the missing API information from a required component in the API that was needed to make the cruise booking, and it's that again, it takes the effort of the user.
00:16:05.715 --> 00:16:07.259
That is the biggest thing here.
00:16:07.259 --> 00:16:10.046
That's so important is you as the person making that query.
00:16:10.046 --> 00:16:11.941
You shouldn't have to do the heavy lifting.
00:16:11.941 --> 00:16:14.803
You shouldn't have to bridge the gap between technologies.
00:16:14.803 --> 00:16:18.465
You shouldn't have to make do with what you have.
00:16:18.465 --> 00:16:28.903
You should actually be served and you should be able to just be asked that question, but then, at the same time, as an enterprise, you can't hard code that information to every single request.
00:16:29.495 --> 00:16:37.216
Yeah, tell me about the building blocks of PVA, because you mentioned their actions, was it I think you said?
00:16:37.277 --> 00:16:37.678
actions.
00:16:38.100 --> 00:16:43.865
Yep, what are the other building blocks when you're thinking about using PVA?
00:16:43.865 --> 00:16:57.428
And I wanted to go out and I want to do gain a level of understanding around anything from large language models to generative AI, to what are those building blocks in your mind?
00:16:57.428 --> 00:17:03.586
What do you need to go out, understand what it does and then you'll have, kind of like, the tool set, the understanding to start creating.
00:17:04.095 --> 00:17:08.767
Well, you come in at this from like a perspective of like a consultant or an architect, mark, yeah, yeah, sue.
00:17:09.035 --> 00:17:12.143
They're going to build a PVA solution for their business.
00:17:13.326 --> 00:17:13.707
Yeah, okay.
00:17:13.707 --> 00:17:18.286
So the first thing I'll definitely recommend is go read those resources that we talked about a moment ago, right?
00:17:18.547 --> 00:17:18.686
Yeah.
00:17:18.855 --> 00:17:35.467
If you was to begin, one of the things to be mindful of in today's technology landscape from a business application perspective is there is a lot of interoperability between the technology Right, let's just be upfront about that Like power vector agents, has a dependency for our generative AI features on Azure Open AI service, right.
00:17:35.467 --> 00:17:47.265
And so one of the things to be very mindful of is there is this kind of element of people thinking they have to stay in their lane, and that's not really how I would recommend people thinking about it.
00:17:47.265 --> 00:17:52.461
Right, you know you don't need to become an Azure expert, right, but you do need to know what it is.
00:17:52.461 --> 00:18:13.269
Right, you do need to understand, because, ultimately, once you understand even just the some of the fundamentals, you can do a better job explaining about how that service is used in power virtual agents and also why and we kind of cover this in one of the blogs that we just launched, which kind of has a little bit of a matrix as to well, you know, these are some of our major services from a generative AI perspective.
00:18:13.269 --> 00:18:17.009
You know what do you get with one of them and you know what don't you get with them.
00:18:17.009 --> 00:18:19.122
Right, like, how can you compare and contrast?
00:18:19.122 --> 00:18:26.445
Because that is one of the challenges, especially when you come at this from a concern and an architect perspective is you're so worried about what?
00:18:26.445 --> 00:18:35.044
If I, like you know, lock myself out of doing something and stuff like that You're kind of trying to future prove something already You've not even built the basic thing.
00:18:35.044 --> 00:18:41.877
So that's one of the things I would say to understand is to try to just understand the kind of actual application landscape.
00:18:41.978 --> 00:18:42.378
First.
00:18:42.378 --> 00:18:52.940
I am you know that's power virtual agents has power automate right that some of the Azure open service right to is literally like one docs article I'd probably recommend, which we can also put in the show notes.
00:18:52.940 --> 00:18:59.789
And then it's really about understanding what power virtual agents does out of the box, right, like what are the fundamentals?
00:18:59.789 --> 00:19:03.792
You know what's the author and canvas, what can I do and what are our generative AI features?
00:19:03.792 --> 00:19:07.055
The power virtual agents team have worked so hard to deliver.
00:19:07.095 --> 00:19:12.979
You know generative actions is limited preview, generative answers and our copilot feature, right.
00:19:12.979 --> 00:19:19.446
So those are our kind of like flagship kind of components when it comes to generative technology, genitive AI technology.
00:19:19.446 --> 00:19:22.810
But ultimately it's like how can you actually create your own topic as well?
00:19:22.810 --> 00:19:40.056
Because what we gotta be seeing kind of what we're seeing now is where, sure, you can use generative AI technology, like generative answers point and a website, but some customers are still gonna want to, you know, go ahead and actually author their own topics that they have control over and they can manage the messaging of.
00:19:40.056 --> 00:19:43.201
So understanding the fundamental authoring canvas will also be key.
00:19:44.146 --> 00:19:48.935
Yeah, we've touched on Azure and the open AI service.
00:19:48.935 --> 00:19:52.743
Tell me about how you'd get that lit up and working inside PVA.
00:19:54.346 --> 00:19:56.347
Yeah, so there's almost two ways that we can look at that.
00:19:56.347 --> 00:20:05.394
There's actually like a complete hands free way where you'll never gonna have to go to the Azure portal so you can leverage Azure open AI service, the service, directly from PVA.
00:20:05.394 --> 00:20:10.679
So again, just crystal clear, right that you don't need to go to the portal and Azure.
00:20:10.679 --> 00:20:13.421
You don't need to go ahead and provision Azure open service.
00:20:13.421 --> 00:20:18.025
The generative answers feature, for example, utilizes that service and you can even see this when you start using.
00:20:19.407 --> 00:20:26.421
If you start test, if you hook this up and if you turn this on, you're gonna be able to start, you know, pointing out a website and ask a question.
00:20:26.421 --> 00:20:37.182
Once you get an answer, you're gonna see like a little bit text says you know, provided by Azure open service, and so that is one of the key things is the fact that I've heard from a few people it's like all to connect it.
00:20:37.182 --> 00:20:38.724
You know you gotta have to go, do a manual connection.
00:20:38.724 --> 00:20:40.848
No, you don't have to.
00:20:40.848 --> 00:20:44.494
You can certainly go ahead and just literally use our own service.
00:20:44.494 --> 00:20:53.369
Now the other thing you can do, which is what we just launched, is that you, if you're developing in the Azure open studio From your service, from your Azure open service.
00:20:53.369 --> 00:21:02.625
There is actually a deploy to PVA button now, so it's specifically on the Azure open AI, on your data service, and you can literally, just literally deploy, and that's another way you can get started.
00:21:03.686 --> 00:21:09.075
So I don't have to worry about things like GPT 3.5 turbo or 16 K turbo.
00:21:09.075 --> 00:21:12.159
No, because you know, when you jump into a zoo you've got those options.
00:21:12.159 --> 00:21:12.579
You can go.
00:21:12.579 --> 00:21:14.722
Okay, which model do I want?
00:21:14.722 --> 00:21:15.904
It just handles that for you, right.
00:21:17.106 --> 00:21:30.237
Yes, so if you're going from the PVA side, yeah, absolutely right, and it's that's the kind of irony, is that, like when people have tried to do this on the Azure open AI side, they're just like I have no idea what I'm doing and it's like you probably shouldn't be in there, right, like kind of that's.
00:21:30.237 --> 00:21:35.868
You know you, this is why we've developed these things, and the fact that you know we want to make it easy.
00:21:35.868 --> 00:21:37.090
We want to make it.
00:21:37.090 --> 00:21:46.325
We want to still provide business value, we still want to be able to make this so simple, so easy and cost effective that you know you can absolutely just provisioned power virtual agents and you're going to be able to get the generative features.
00:21:48.147 --> 00:21:49.753
Wow, pretty compelling.
00:21:49.753 --> 00:21:53.144
Tell me about the relationship between PVA and adaptive cards.
00:21:54.750 --> 00:21:56.838
Oh, yeah, for sure, I mean they like each other.
00:21:56.838 --> 00:22:05.805
It's funny, adaptive cards and one of our was actually one of our most anticipated features when we just dropped the latest release that build this year.
00:22:05.805 --> 00:22:10.333
So Our previous version did not support adaptive cards.
00:22:10.333 --> 00:22:21.112
You would have to go ahead and, like you know, install composer and stuff like that so you can now literally, you know, make some very, very rich adaptive cards as well as it's also the behavior around them.
00:22:21.112 --> 00:22:22.914
So I just want to be super clear.
00:22:22.914 --> 00:22:44.344
You know, especially especially for maybe those that haven't played around with adaptive cards in power, virtual agents, when you add the conversational layer on top, you know, it's not just you have to be really kind of like, kind of mindful of the channel, of what that's being kind of deployed in right, because the actual kind of how adaptive cards are constrained by channel is quite critical because of the interactivity involved and when you've got a conversational interface.
00:22:44.364 --> 00:22:44.704
It's important.
00:22:46.909 --> 00:22:48.271
So you're talking about channel being?
00:22:48.271 --> 00:22:49.794
Is this going out in an email?
00:22:49.994 --> 00:22:51.277
exactly inside teams chat.
00:22:51.277 --> 00:22:51.758
Yeah, I love it.
00:22:51.758 --> 00:22:52.318
I love it.
00:22:52.439 --> 00:22:55.265
Okay, good, so you're saying you need to know that context of where that's going to be surfaced.
00:22:58.630 --> 00:23:03.818
Yeah, yeah, for sure, because that you're going to have like different kind of things available to you and right.
00:23:03.818 --> 00:23:07.045
And if you've got like a web chat with adaptive card, are you making an interactive adaptive card or is it an informational type adaptive card?
00:23:07.045 --> 00:23:12.215
Right, because, again, you've kind of gone switch to an app type experience where you might be expecting someone to like click something.
00:23:12.215 --> 00:23:15.623
And that's also really important when you're thinking about voice technology as well, right.
00:23:15.623 --> 00:23:22.343
So, like, if you've got kind of if you've got your bot enabled for voice, things like adaptive cards are not going to be relevant in that area.
00:23:22.343 --> 00:23:24.588
Yes, yeah, interesting, interesting.
00:23:24.588 --> 00:23:25.109
You touched on voice.
00:23:25.109 --> 00:23:27.939
I wasn't going to go there, but let's do it.
00:23:32.048 --> 00:23:54.744
What's your familiarity personally with things like Alexa, google Home, those type of products just checking my one didn't light up and do you see, and maybe under the co pilot moniker that we're seeing, it's going to come into windows and AI on windows, you know, is in preview, I think at the moment Do you see that the potential interface out will even move maybe further away from the keyboard?
00:23:54.744 --> 00:23:58.345
No-transcript.
00:23:58.345 --> 00:24:09.481
As we go forward, as in so on my Windows device, I assume as soon as it goes public, my Windows 11 computer will update and it will have co-pilot for Windows on it.
00:24:09.481 --> 00:24:21.387
Do you think in the near future we will get to that point where we will become much more tactile in voice engagement rather than using the bot from it?
00:24:21.387 --> 00:24:31.561
I'm still typing a question into a keyboard and so therefore all my bad spelling, my bad grammar, my dyslexia still comes out in that, but it doesn't when I talk.
00:24:31.795 --> 00:24:33.741
There's a really interesting question, actually.
00:24:33.741 --> 00:24:39.564
Even if you just look today, people will use Siri 24-7.
00:24:39.564 --> 00:24:42.643
Some people won't use Siri at all.
00:24:42.643 --> 00:24:46.785
They'll just turn it completely off, and it's a similar experience.
00:24:46.785 --> 00:24:49.304
You get some people that love Alexa being in their home.
00:24:49.304 --> 00:24:50.660
Other people don't like it.
00:24:50.660 --> 00:25:07.884
I'm a big fan and I talk about this fairly a lot on various platforms about I do truly believe in the ubiquitous nature of conversational AI technology, because I think that I mean, even if you just watch films, you don't see someone really using a chatbot.
00:25:07.884 --> 00:25:10.343
You see someone using their voice.
00:25:11.477 --> 00:25:16.768
You see the actor or the main person using their voice and asking the computer what to do.
00:25:16.768 --> 00:25:24.242
I think how much the world has now become aware of conversational AI technology from chat, gpt.
00:25:24.242 --> 00:25:34.961
I think that that's just going to continue, the people being more comfortable with talking to things, because the other thing to be mindful of that is it's not particularly great.
00:25:34.961 --> 00:25:36.265
On customer calls.
00:25:36.265 --> 00:25:40.545
You have to take into consideration noisy environments.
00:25:40.545 --> 00:25:47.488
There's new things to consider when you change the interaction model of the technology.
00:25:47.996 --> 00:25:51.826
Yeah, babies, that definitely affects things.
00:25:52.496 --> 00:25:53.198
Yeah, for sure.
00:25:53.198 --> 00:25:55.945
If you've got a screaming child or something, yeah.
00:25:56.474 --> 00:26:02.638
I'll be halfway through a prompt and then my daughter will talk to me and of course it's now picked up on.
00:26:02.638 --> 00:26:05.144
That part of the conversation is trying to respond.
00:26:05.144 --> 00:26:12.523
I just want to turn the lights off and turn the lights on or something like that, not set a new timer or play.
00:26:13.518 --> 00:26:14.622
Because then you're backtracking right.
00:26:14.622 --> 00:26:16.919
You're just like no, no, no, no, don't do that, don't do whatever you're going to do.
00:26:18.477 --> 00:26:24.872
Honestly, as in my almost three year old, she is pretty close to saying all the commands.
00:26:24.872 --> 00:26:34.363
Really, I'm pretty sure that once she every now and again, she'll definitely get Google to respond, because we're mainly a Google Home house, Although I do play in all ecosystems.
00:26:34.363 --> 00:26:35.840
I've got a couple of extra devices as well.
00:26:35.840 --> 00:26:47.147
But yeah, at realizing, oh my gosh, I can actually control the projector turning on, I can get poor patrol by saying something rather than knowing how to control buttons and stuff.
00:26:47.147 --> 00:26:48.882
It's only a matter of time.
00:26:48.882 --> 00:26:52.404
Even her bedroom lights won't be long until she works out.
00:26:52.404 --> 00:26:52.997
Do you know what?
00:26:52.997 --> 00:26:54.643
She doesn't need to reach the light switch.
00:26:54.643 --> 00:26:56.280
She can actually just say.
00:26:56.434 --> 00:26:59.722
It's natural for her, though, right as well, which is like the truly magical thing.
00:26:59.722 --> 00:27:01.561
She's growing up with this technology.
00:27:01.561 --> 00:27:08.198
Can you imagine when she's older, like putting like an old interface or an app in front of her, Like she's probably going to be?
00:27:08.198 --> 00:27:09.804
Like what are you like, dad?
00:27:09.804 --> 00:27:10.657
Like what are you doing?
00:27:11.320 --> 00:27:12.596
Yeah, yeah, no, I'm going to.
00:27:12.596 --> 00:27:18.880
I'm going to get a little morse code key and teach you how to start Like Blip, blip, blip, blip, blip, blip, blip.
00:27:18.880 --> 00:27:22.041
Let's just change directions.
00:27:22.041 --> 00:27:24.502
We've talked about PVA for a bit and we're almost on time.
00:27:24.502 --> 00:27:28.184
Tell me about side projects that you are into.
00:27:28.184 --> 00:27:35.423
I see you often post under your favorite feline around careers and things like that.
00:27:35.423 --> 00:27:40.150
Tell me what tickling your interest outside of PVA.
00:27:40.813 --> 00:27:41.494
Oh yeah, for sure.
00:27:41.494 --> 00:27:43.162
I mean, I like making cat quilts.
00:27:43.162 --> 00:27:44.500
That's something that I do now.
00:27:44.500 --> 00:27:47.403
I kind of yeah, I've gone down that route of making.
00:27:47.596 --> 00:27:50.565
So what you kill cats, you get the skin.
00:27:50.776 --> 00:27:52.884
Don't say that Don't say stuff like that.
00:27:52.884 --> 00:27:54.840
You said a cat quilt.
00:27:55.597 --> 00:27:58.522
I was like I'm so mad You're using cat fur to make.
00:27:58.522 --> 00:28:00.541
No, oh no.
00:28:03.260 --> 00:28:05.285
I'm giving Mark so many evils right now?
00:28:05.285 --> 00:28:07.500
No, no, I'm giving evils.
00:28:07.500 --> 00:28:09.942
They are super happy.
00:28:09.942 --> 00:28:17.500
I've got good customer feedback from my cat quilts, so that's something that sometimes it's nice to just get away from the laptop, you know, and actually do stuff.
00:28:17.500 --> 00:28:18.744
Like you know this with your garden, right.
00:28:18.744 --> 00:28:30.443
But from a career perspective, like I am super passionate about, like digital literacy, right, like it's and also career planning right, like it's something that we just do not talk about and I don't know why.
00:28:30.443 --> 00:28:39.282
It's one of those things that I'm just like I have no idea why people just are so afraid to talk about their career with, whether they're manager, their peers.
00:28:39.282 --> 00:28:59.644
I've got lots of theories of why, right, but I'm hoping, by talking about it and like I've got the sub stack on it, I've just started, like Motivation Mondays, to try to give people a little bit of, like healthy dose of optimism on a Monday, monday morning, right, to hopefully normalize the fact that people should be having career conversations and being open and given them the tools to do so.
00:29:00.377 --> 00:29:00.801
I like it.
00:29:00.801 --> 00:29:02.275
For all of those in the Southern Hemisphere.
00:29:02.275 --> 00:29:03.573
She means Tuesday morning.
00:29:03.573 --> 00:29:06.938
She just yeah, ok.
00:29:06.938 --> 00:29:10.115
And so I love the stuff you're doing around resume, cvs and stuff.
00:29:10.557 --> 00:29:33.661
I just was on a podcast recently, ecosystem podcast, which will sometime be published on the show, and I am shocked with the number of CVs that come across my desk as I am One of the people that decide on people's employment inside my organization at the lack of substance, the lack of personal development, the lack of direction.
00:29:33.661 --> 00:29:41.864
I'm shocked that most people CVs represent they were asked on their last project to do something, so they learned about that something and did it.
00:29:41.864 --> 00:29:43.135
It was not so.
00:29:43.135 --> 00:29:53.864
In other words, anything that's on their CV is because somebody else, you were doing somebody to meet somebody else's agenda, not your own agenda, not your life goals, not where you're wanting to go, not this.
00:29:54.085 --> 00:29:56.690
You know, and it concerns me, and I use an example.
00:29:56.690 --> 00:30:02.694
You know I don't want my, if I had brain surgery, I don't want my surgeon to have looked up on YouTube 20 minutes ago how to do it.
00:30:02.694 --> 00:30:14.433
All right, yet in our industry, where we're touching phenomenal amounts of data and implications of that, I feel there are so many consultants that are literally three YouTube's ahead of the customer videos.
00:30:14.433 --> 00:30:17.094
Right, and that's concerning because it's not a holistic view.
00:30:17.094 --> 00:30:23.942
Right, it's a very much one person's lens in that YouTube and I think YouTube's a great micro learning type of platform and experience.
00:30:23.942 --> 00:30:30.244
But If your career is in the space, act like it Like learn it all.
00:30:30.244 --> 00:30:34.060
I love that from Satya Be learn it all.
00:30:34.060 --> 00:30:41.435
People Like be the best in your category, not just the best at the last item on the project you did.
00:30:41.827 --> 00:30:43.104
And what a time to do it, right, mark?
00:30:43.104 --> 00:30:45.785
Like kind of, we have such an abundance.
00:30:45.785 --> 00:30:51.085
Going back to what we were just chatting about, right, the interoperability between this platform is amazing.
00:30:51.085 --> 00:30:52.410
Do you have an interest in AI?
00:30:52.410 --> 00:31:04.305
Go, take a look at AI, right, the amount of materials is just phenomenal, sometimes overwhelming, for sure, but overwhelm is better than nothing, and you can get the tools to navigate overwhelm right.
00:31:04.305 --> 00:31:10.213
You've got a place to begin, rather than the you know it being kind of like, you know, behind locking key.
00:31:10.768 --> 00:31:13.164
So it makes me sad, I'll be honest with you.
00:31:13.164 --> 00:31:32.932
It makes me sad when I see a poor CV, because I almost feel like one of two things have happened Because that person could actually have passion right, they could actually have passion, they could actually have experience and they've done a poorer job of showing you the person who can potentially give them employment of that passion, right.
00:31:32.932 --> 00:31:41.231
Or you know, this is the second thing, which is what you said, which is they've let someone else really give them the route to where they're at today, like they've been directed to.
00:31:41.231 --> 00:31:42.825
Oh, you're on this project, go learn it.
00:31:42.825 --> 00:31:44.025
You're on this project, go learn it.
00:31:44.025 --> 00:31:50.025
And you find this kind of haphazard piece and it is truly heartbreaking as an interviewer when you ask, oh, why did you do that?
00:31:50.025 --> 00:31:51.605
And I'll be like, oh, because it was on my projects.
00:31:51.786 --> 00:31:54.505
And it's like, so you're not actually passionate about it.
00:31:54.505 --> 00:32:03.270
I mean the challenge there is, the resource inside of the house can be, you know, especially in consulting and professional services.
00:32:03.270 --> 00:32:08.316
It does direct, where employment is a lot of the time and projects for sure.
00:32:08.316 --> 00:32:16.000
But I think people you almost see that as the fact that, oh, I don't have autonomy, I don't have conscious choice to go.
00:32:16.000 --> 00:32:23.125
Well, okay, well, I'm on this project, but I actually have a passion in data, so I'm going to go learn this side of things and that actually opens up so many doors.
00:32:23.125 --> 00:32:38.270
I feel like, for those listening, if you have a passion like that, you might not think it'll open doors, but it actually does, because you'll just get chatting to a random person one day about that topic or an interviewer and they'll be like, wow, this person has taken the initiative to go do something on their own and that's worth a lot.
00:32:38.986 --> 00:32:40.270
Yeah, totally agree.
00:32:40.270 --> 00:32:46.453
Final question I like to ask my guests more and more these days how are you personally using AI?
00:32:46.453 --> 00:32:49.865
So forget about Microsoft, the tech, all that type of stuff.
00:32:49.865 --> 00:32:51.191
How do you use it in your life?
00:32:52.644 --> 00:32:54.732
Yeah, so I use it a lot on LinkedIn.
00:32:54.732 --> 00:33:00.835
So, because I don't want to be putting my own hashtags into my LinkedIn post, I've got bad things to do.
00:33:01.085 --> 00:33:03.634
That's something that I think is super, super good.
00:33:03.634 --> 00:33:19.278
The other thing that I've noticed is that AI has been embedded in existing apps today, so you'll be using note taken apps or whatever Microsoft aside just any other application there's AI being used.
00:33:19.278 --> 00:33:24.491
You'll see a little AI logo or whatever it is, and you'll be like, oh wow, you've embedded that.
00:33:24.491 --> 00:33:30.633
Like I've been learning French, so G-Wo Lingo is a really good app to do that and they embedded AI in the app.
00:33:30.633 --> 00:33:31.709
I mean, it's a different SKU.
00:33:31.709 --> 00:33:36.472
You have to pay my money for stuff like that sometimes, but I think that that's kind of cool.
00:33:36.472 --> 00:33:40.856
So I almost feel like it's almost not a conscious decision mark anymore either.
00:33:40.856 --> 00:33:48.652
Is it like some people may not even realize they're using AI in some of these applications because existing applications have been updated to use it?
00:33:48.785 --> 00:33:50.526
Yeah, agreed, Sarah.
00:33:50.526 --> 00:33:51.490
Thank you for coming on the show.
00:33:51.810 --> 00:33:53.146
Thanks so much, it was amazing.
00:33:54.325 --> 00:33:55.250
Hey, thanks for listening.
00:33:55.250 --> 00:34:00.071
I'm your host business application MVP, mark Smith, otherwise known as the NZ365 guy.
00:34:00.071 --> 00:34:05.069
If there's a guest you'd like to see on the show from Microsoft, please message me on LinkedIn.
00:34:05.069 --> 00:34:08.952
If you want to be a supporter of the show, please check out buymeacoffeecom.
00:34:08.952 --> 00:34:09.847
Point slash.
00:34:09.847 --> 00:34:14.713
Nz365 guy, how will you create on the Power Platform today?
00:34:14.713 --> 00:34:15.746
Ciao.