Empowering Businesses with Microsoft’s Copilot Studio: Navigating Autonomous Agents

Mark Smith: 0:14
Welcome to the Copilot Show, where I interview Microsoft staff innovating with AI. I hope you will find this podcast educational and inspire you to do more with this great technology. Now let's get on with the show. Today's guest is from National Tennessee in the US. He works at Microsoft as a Principal Program Manager for Copilot Studio and Conversational AI. He helps people leverage Copilot Studio to build amazing copilots. You can find links to his bio and socials in the show note for this episode. He was last on the show for episode 497, and we talked about Power Virtual Agents, so that was a wee while ago, but welcome back, dwayne.

Dewain Robinson: 0:58
Thanks, I'm glad to be here, Mark. It's awesome to get to talk to you again. We just were kind of talking before that. It's been so long and I couldn't even believe it was. Yeah, it's hard for me to believe it's even been that long. It just seems like yesterday you and I were chatting about this.

Mark Smith: 1:13
Things are changing so, so rapidly. If you look at you know we're recording this at the start of 2025. I feel we've come off an epic year where corporations are really starting to go. What does AI look for us? And then the last three months I feel we were heavily there was heavy marketing, not just from Microsoft, but all the major players in the world in the generative AI space LLMs, et cetera. Around that, 2025 is going to be the year of agents. So, with that backdrop before us, when you look at this year ahead, what excites you right now the most?

Dewain Robinson: 1:53
I think the thing that is going to be very interesting in the Copilot Studio world, right, is, I would say, that there's this perception of AI and what it can do, and then there's the reality of what it could do and how easy it is to do that with that AI back last year. Right, you could really build some really cool stuff With that AI back last year. Right, you could really build some really cool stuff. And I think what's happening is the commoditization, the roughing of the edges that were a little bit in there. And so I think that in itself or what I would call like a synchronous or a, you know, kind of like a synchronous co-pilot, where you talk to it and it talks back to you is going to be one of the things that is going to be one of the things that is going to be the major innovation. It's not like oh, we're going to see something that we never thought of. It's just going to be the things that you are wanting to do will become very easy and commoditized, and the integrations to be able to tie that in with Azure and other pieces are just going to become easier and easier, and you could do it before, but now it's going to be like oh, there's a button and it's an easy button to do it.

Dewain Robinson: 3:04
And then I think the other thing I would say, mark, is that as we walk into the autonomous agent space, that's going to really change the perception. It's beyond the concepts of what you can do with AI by itself. It's now AI working in conjunction with business process automation to be able to do something you could never do before. And I think the combination of these two things are really going to be a year. That's just insane. And, as you're aware, I mean the innovation on the AI side and the AI model side is just going to keep going too.

Dewain Robinson: 3:42
So, when we talk again and you know, because we're not going to wait so long, right, mark so then when we talk again next time, I would anticipate there's probably going to be some innovation in the model space that you can do and stuff like that, because we already started seeing some of that. Like you can now stream audio in and you're not having to do text to speech. Yeah, like that's a big deal, right. And so I think there's going to be a lot of interesting things when we start going beyond just chatware. Yeah, I think chat stuff is going to get super easy to do, but then these other things are just going to really take it to a level that people really didn't think about, and I think that's the beauty of it. And this is the year that I think it's not just about open AI. It's about open AI's integrate, how Microsoft's going to take open AI and combine it with the Microsoft asset pool, and then you're going to get something that's completely different, that you couldn't do before, which I think is really awesome.

Mark Smith: 4:37
Have you seen, I suppose in 2024, looking at the year in reverse, without revealing any customer details or anything did you see anything that was quite you're like? You know that was cool, that's a cool use case, or that was a cool way to use a technology? Because you mentioned voice there and I think voice is a massively underexplored territory in how we go forward. I just had a customer the other day say listen, when I fill out my CRM system, I really just want to be able to open the record and say the first name is Bob, the second name is Frank and it just types it in just with me talking and you know, auto filling out, that type of functionality which we get that degree coming in in the Dynamics suite of products. And but it was interesting. This is an old guy, but his way of how you know he would like to use AI combined with what he was doing and but was there anything that really kind of stood out top of mind story that was highlights for you last year?

Dewain Robinson: 5:36
I had a couple of things, that kind of top of mind, and I will say there was this thing that when I think back where I got value that I didn't expect. So it's not necessarily like the use case in this scenario. Well, I'll say this use case was fairly interesting because and everyone's heard about the use case, because it's the Holland America one that we talk about publicly, right but the thing that was about that I didn't expect. You know, we had done business value assessments and all this stuff about what we expected, the return and how this thing was going to perform and everything. And we were expecting to get like a return from like over in this area, you know on the left and what also, but we didn't expect this value area on the right. We didn't even know that there would be impact there.

Dewain Robinson: 6:29
And and what it came down to is that if you think about it, think about I have an API or I have something that you want to do, and in the case of this it's like cruises I want to find a cruise. So when we were doing this, the customer had an API that had here are all the different cruises that you might want, and so we needed to go search for a cruise that would be the right cruise for you cruise that would be the right cruise for you. And what we found and this became similar because it wouldn't matter that it was finding cruises it could be like you wanted to search ServiceNow's knowledge base right For knowledge articles. What we didn't expect at the time and it makes logical sense now that this would be the case is what we were doing is using an AI model to look at the context of the whole conversation and generate a le really improved search results, coming back from an existing search service that then could be consumed and consumed by the AI with the context as well, but knowing that the search quality was going to go up and the search results were going to be better off of your API. That was just a standard lexical search.

Dewain Robinson: 7:59
I didn't expect that. I didn't think that that would be something that was the case. But we saw that not only with the Holland America example, but we also were working with another company and they were trying to do something with ServiceNow knowledge articles and we saw the same thing and we went oh man, we didn't think about that. And so there's actually some you know features and stuff that we started putting into Copilot Studio because we realized after we saw that that that's going to mean that a lot of people are going to want to be able to do that and we want to make that easier to do and all of that. And so I think that was kind of like a really aha moment for me when I was looking at that. And then the second thing that was like really cool is I did a video that showed the new orchestration engine.

Mark Smith: 8:49
I was just going to ask you about it.

Dewain Robinson: 8:50
Yeah, yeah, and I did one where I do one with the PO, like I'm creating a PO, where I do one with the PO, like I'm creating a PO. And what a lot of people don't know is that I'm building this thing. And I get on the phone and I call up Gary Pretty, which is the lead over it, and I said, hey, I need you to tell me what are all the things that I need to show to demo this thing in the best light. And so I built this thing with no intention, right? I was just wanting to check this box, check this box, check this box. And I went through it all market and I built all this stuff with all the different pieces to do it.

Dewain Robinson: 9:25
And then, out of a fluke, I just went, wait a minute, if it has all that, if I say, create me two POs with this one with this and this one with that and everything. And it just did it. It literally figured out all the assets and did it and dynamically assembled it and did it not just once, but did it for two different things, summarize it all out and spit it out. And that demo and that video that you saw from that stems from me building that as a way to demonstrate it on a stage, to show people this new feature.

Dewain Robinson: 9:59
But then I'm trying to do something real world and then it occurs to me that wait a minute, this thing can do more than what I really thought about when I was building it, which showcases the fact that if you do these things right and you modularize it right and your model descriptions are good, it'll actually do stuff that you didn't even intend it to do, because it's flexible enough to figure it out. And that's kind of a really big aha moment. We even had one of our guys that was demoing on a stage and he was showing something and we had negative feedback on him because he said on the stage he went oh wow, get a feedback on him. Because he said on the stage he went oh wow, that worked like I didn't expect, and so, like those things were really impressive to me as I was going through. So those are the standout ones for me right now.

Mark Smith: 10:48
I like it, I like it. And now, in 2025, and we see this concept and what I'd like you to do actually start with your description of what agents are what is semi-autonomous, what's autonomous and, let's say, what's fully autonomous or really you know, set and forget type agent scenarios. I think orchestration plays into that mix, because you're going to have these agents that have deeper expertise, but you're going to need an orchestrator to know which agent to call or which one to fire in a scenario. Let's start with how do you explain the concept of agents that we're hearing about now?

Dewain Robinson: 11:53
Yeah, so there was a little bit of a rebrand. That happened, and Microsoft were great at marketing and rebranding things without being clear that we did it. So the word co-pilot let's start there. Co-pilot was being used like everywhere. Yes, right. And so we started to get a problem, which was nobody understood what the word co-pilot meant, and then we started using co-pilot and the word agent, and then we started getting into apilot and the word agent, and then we started getting into a new one which was going to be a function, an extension and a plugin. And all of these words started becoming a problem because people were using them interchangeably and I literally had two sales guys fighting in front of a customer on what co-pilot did, and they didn't even realize they were both right, they were talking about different products. And so the thing that if we start out what is a Copilot now by branded by Microsoft, you're going to see that it's going to have like a big C in the front, and the reason behind that is Copilot is a product from Microsoft, it is a product that you configure, and sure you could say it's an agent that's prepackaged and available to you, right? And so that's why you're going to see this word co-pilot. Now, an agent is going to be different and you will see an agent with a small A, not with a big A, and the reason behind that and this is, by the way, I've had to have a long talk with marketing to get this, so just trust me, it's not like this is common nature.

Dewain Robinson: 13:29
The word agent is meant to say it is a conversational AI implementation that uses generative AI that you're in control of that you build and that you are in control of the orchestration. When you go over into, like Copilot for M365, you don't have full control of the orchestration engine. Now you can throw something in and you can extend it and you can do all that, but you can't make it not do something that it's programmed to do, whereas an agent means you are in control of it, and so if you think the word co-pilot agent big C, little A means that's what an extension is of a co-pilot product is a co-pilot agent. And so if you're saying agent, then we're just talking about something you built. Now how you build that thing, it can be whatever. And so, yeah, we always thought it was funny and you'll get a kick out of this Mark is that we changed the name from Power Virtual Agents to Copilot Studio. But now, copilot Studio doesn't build copilots, it builds agents.

Mark Smith: 14:34
Yes.

Dewain Robinson: 14:35
So I had fun with marketing on that one, yes, but the thing is, is so in Copilot Studio, what are you going to do? You're going to build agents. Right Now, can you build agents in different ways? Absolutely, you can, but fundamentally understand that an agent that is a real intelligent agent, that is something bigger than just a model, is going to need to have an orchestration engine. And that is what Copilot Studio is. It's an out of the box, it is a SaaS-based conversational orchestration engine.

Dewain Robinson: 15:09
Right, and we saw the evolution of that to incorporate using not a traditional language model, and for those that don't know what I mean by that, it used to be that you would have to train an NLU model. You'd say here's six different ways of saying this thing, and that way I can say this is the intent. And then you could do the same thing with entities, which, if you want to understand the difference between intents and entities, an intent is if I said I want to order a large pepperoni pizza. The intent is that I want to order something, then the entities are large pepperoni pizza. The entities are large pepperoni pizza, and so, being able to do that, you had to train that into a natural language understanding implementation and traditionally, this is where large language models came along, because you don't train one Not typically you don't train one and, matter of fact, to train a large language model from the ground up takes like six months to do with a massive data center. It's worth the GPUs to be able to get it done. So, when we're looking at these, you don't do that anymore. How you use these is you had the innovation that happened where I don't need to do intent routing or determining what tool to use. I can use a orchestration engine that's not powered by a traditional NLU. I can do it with a large language model, and this is what you heard us talk about generative actions and everything, and you're going to see like you'll hear us refer to this as generative orchestration now, because we realize like actions, which is the wrapper around an API, we needed to separate that from the orchestration engine capability, and I'll explain why in just a second. So, once we've done this, this is where it really gets into what you're talking about.

Dewain Robinson: 16:52
What is an agent? Okay, so now we can start to classify them. Well, and there's different agent types. Now you have what I refer to as a synchronous agent. We just call it an agent, which is what you traditionally think of as like a chat, gpt orbased orchestrator that could understand the context of the conversation, and or the Power Automate flow, to go create a PO, and here's the one to go get an approval, and oh, here's one to go get information about our HR policy and all that. So what you're seeing is it's basically acting as like a router, a conversational router, and that, by the way, is what you know. Conversational orchestration is yes.

Dewain Robinson: 18:01
So that's when we started to see that you're going to start to see these other agent types, and I think this leads into a thing that I know many people get confused about, which is what is the difference between what you're hearing us talk about autonomous agents. And why is that? Because people think about it and they go oh well, smart, business process automation or something like that and they go well, wait a minute, how is that different than Power Automate? And because now we're starting to see these classifications where it's not you're talking to it and it's talking back. Now it can be that something else triggers it and then it does something, and a lot of people go well, that's Power Automate. So Power Automate can have something that triggers it and then it does something and a lot of people go well, that's power automate. So power automate can have something that triggers it and then you can follow a business process and everything. So here's the difference, mark, and this is why it was so important to land a conversational orchestrator based off a large language model.

Dewain Robinson: 18:57
Because now imagine back up to Copilot Studio for a second. Imagine if I was going to hire someone. I'm going to write a job description. I take that and I say here's what your job is, here are the rules you should follow, here are the tools at your disposal to be able to do your job. And this is what your job is. And when should you do your job? Well, when you receive an email, when someone puts a record in a CRM, when someone fills out a form like whatever, right. So there's this thing. That's not that someone's talking to you, but guess what? Those autonomous agents can talk to you, but they don't have to, because they can go and look at it and imagine I have six different power automate flows. I explain what they all do. I come over here and I say, oh, here's some knowledge with like information on products and information about business processes and how those things should, what those policies are and things of that nature. And then I have, like, all these different tools, and even a topic could be a tool, yeah, so now I'm triggering those things and it can do whatever I needed to go do based upon something going on.

Dewain Robinson: 20:10
So one of the demos that we did I mean I didn't show this one publicly. I can't remember if I made the video for this one or not. I have to go back and look what I did. I think this is the video I'm working on right now. But the problem, marcus, it's getting harder to build the crazy complexity that's going to make everybody go wow, because I know I can do it, but I have to go build it all.

Dewain Robinson: 20:29
But what I did at that particular thing at the CIO Summit is we took and made a form that you could go ask it about what are cruise specials? What are cruise specials? And it would go look you up based upon your email address and figure out what you had cruised in the past, make a product recommendation, look at what the specials are, figure out what you like, go in and create a CRM and implement you know record for it, update the CRM record, then make you an email and email it to the person and giving them the cruise specials. And then someone could email back to it and be like, hey, you gave me these three options, but I was thinking about I wanted to cruise Greece this year Do you have any specials for that? And it could then converse with them and then, and all the time when it's doing this, it's updating the CRM system for people to know what was going on and it's helping with the pipeline.

Dewain Robinson: 21:18
So the thing is, is that that's where you get into autonomous agents?

Dewain Robinson: 21:24
I think everyone gets confused because they make it too simple.

Dewain Robinson: 21:26
They think it's like just business process automation or that there's a step within the business process automation where I want to look at the model and figure out what to do next.

Dewain Robinson: 21:36
But it is literally like the brain of the orchestration engine being able to be triggered and then it understand what its job is and the tools it has, so that it can just figure the brain of the orchestration engine being able to be triggered and then it understand what its job is and tools it has, so that it can just figure out what's the right thing to go do to be able to complete it.

Dewain Robinson: 21:51
And right now, I would say we're in the infancies of what it can do, so you might hit some limitations and stuff, because it was only in preview, but the intent is that it works like that, and a lot of people then freak out when I talk about this, mark, and they go oh my God, it's going to replace me for my job and it's like no, it's going to take the thing that you have to do over and over and over again and make you not have to do that so you can work on the things that you need to provide the higher quality, and so I really think this is the thing that's really going to be amazing in 2025.

Mark Smith: 22:26
If we look at that example that you've just given and we say it's a job description, right, if you look at a job description for a role, it'll fit on one side of a sheet of paper maybe. Then you start doing the job and you realize it's really 300 sheets of paper because we don't tell you that you're going to have to be in these meetings and you're going to have to, you know, reply to an aggregated email. We don't put all those nuances of the job in the job description. But you start to pick it up and when I think of agents and the element that I would love to see is the concept of memory, because with memory then we can learn.

Mark Smith: 23:06
And in that example of the cruise scenario and like, I don't see memory being something that can inherently sit in an LLM because, as you said, an LLM takes six months to train and it's ready to go. So somewhere outside of that, we need some concept of being able to remember past activity, et cetera, and also to learn from error, right, right, how does a rocket get to the moon? It makes lots of errors all the way that it self-corrects on to get there. Do you think that's something that's going to sit in the camp of Copilot Studio particularly, or is that going to be something that's going to be more on the foundry side of the game as we move forward and that ability to store more long-term memory?

Dewain Robinson: 23:52
So a combination. It's going to be like a combination, so think of it that you already have memory. So Copilot Studio already has memory in two places, but there's a third place that's coming right. Copilot Studio already has memory in two places, but there's a third place that's coming right. So the first place is there has always been something that, if you go all the way back into bot framework, we used to refer to this as the conversation state. So the conversation state and your ability to write something that you're talking about into memory is ability to do that, and that's where variables come along, right. So, if you think about it, this goes back to okay, everybody's like oh, I'm going to go and use AI Studio or I'm going to go use OpenAI's implementation and I'm just going to. Everything's just in this one model. Well, that's where you don't have a conversational orchestration engine or a dialogue management system in a sense, right, and that means you don't have a conversation state, so you can't store something that I don't want to take into consideration in the next turn, but later I want to use it right, and so that's something that Copilot Studio has today and that's why our variables come in, and there's been a ton of work on variables that nobody even paid attention to, like being able to store a record in a variable and be able to update a record within a variable, so that you can take a whole JSON payload and drop it into a variable and then, just whenever you wanted to use it, you could just concatenate that into the query set and then provide that into the large language model for context. And so then the second part is going to be in the prompt, so the prompt itself is going to give you the ability to pass this in, and everyone kind of understands that one probably if they're following this conversation to this level. And then we've got the new part right, which is when OpenAI started introducing the concept of memory itself and having a way to provide memory to the model. That is more beyond the prompt, and that is something that you'll see Copilot Studio embrace Now, when you talk about foundry and you start talking about this, this is just you're going to build something to do this, like to pass it in and to be able to deal with it, and so that's kind of the value of the whole Copilot Studio thing is that it's not that it's low code type of thing I think a lot of people get over the moon about low code, it's the fact that it is a SaaS platform, so that you don't have to go build all this stuff.

Dewain Robinson: 26:24
So, whether you really want low code or pro code or whatever you want to do, that's really not a separation in Copilot Studio. So much anymore. Can you use Copilot Studio and do low code Absolutely. Can you do pro code Absolutely? So the question really gets down to do you want to SaaS or do you want to build it yourself? And then some of that's going to be I want to build this one little piece by myself, and that's going to be where you're going to see, like Copilot SDK and all of that. So you're going to see an evolution of the bot framework SDK happening which is going to make it where it can handle memory and it can do all of those things, and you're going to see the same thing in foundation and all that, and so all these things in our work, it's just what level of control and how much depth do you want?

Dewain Robinson: 27:16
Copilot Studio? It's not low code, pro code, it's SAS or no SAS. Do you want SAS or do you want PaaS? That's really better the way to think about it and I could have my SaaS call for this particular thing. I need to be in more control, so I'm going to build a service in PaaS that I'm going to call from my SaaS, and that's actually the most common configuration that we see. If you just build everything in PaaS, then your ROI is going to suck pretty bad and your maintenance is going to be very high. So that's why I think it's a combination of these things and I think you'll see this get a little clearer to people, because at Microsoft we're learning, just like you guys are learning right, and you find the mistakes like we've already made a few mistakes Like look at M365 Copilot. Originally there was like this little dropdown box that you had where you're going to build all these plugins, and that thing's being deprecated. It's coming out Right and now you're seeing a Copilot agent replace it, and it's because we realized people wanted to be able to at mention a Copilot agent so that they can direct it to what they wanted to talk to, and I also think you'll probably see in this space that'll help you on this mark as well.

Dewain Robinson: 28:29
When it comes to memory, is I think we might with the combination of memory, the combination of a platform that has a conversation state. That can be because we're in a SaaS. We can standardize the conversation state to make it look standardized so that you don't have to worry. I can build things that can use it because it's standard the way that it works, where, if I let you have an SDK, I can't make everybody follow the same standard for the way we're going to store variables. So now, with those things happening and memory coming in and everything, I think this might actually open the door for something that I think you're kind of alluding to, which is imagine an agent that can talk to another agent or understand what another agent can do and being able to orchestrate between agents in a way that doesn't make the user really irritated or cause a massive nightmare. So I think there's a possibility that that might be viable with this whole memory thing coming out.

Mark Smith: 29:38
So I like it.

Dewain Robinson: 29:40
Long-winded answer, but there you go.

Mark Smith: 29:42
It's good. It's's good we're at our final question. We've run out of time. I have the ui ai. Co-pilot is the ui for ai, and I see it as a dimension probably beyond what microsoft are currently talking about it right, which is it's a great catchphrase, whoever the marketer you came up with it.

Mark Smith: 30:02
Fantastic what I've noticed in the enterprise customers that are implementing Copilot. Or actually, let's back up, Any enterprise customer has a lot of software let's say, 200 to 300 different applications that do something in that organization, from the big vendors like SAP Pega ServiceNow, these type of things Microsoft, of course, like SAP Pega ServiceNow, these type of things Microsoft, of course. And people have fatigue inside organizations around application switching. I need to do something around inventory, so I'm going to jump into my SAP system. I'm going to do something around my holiday leave, so I'm going to jump into my HCM system. Blah, blah, blah, blah Right, blah, blah, blah, blah, right.

Mark Smith: 30:40
And each of these product manufacturers have now come up with their AI interface to their software. And what I see happening is now people are going okay, which AI do I use and how does that AI differ in how I prompt that one to how I prompt that one? And so when Microsoft came out with this UI for AI, I'm just like awesome. We don't care where the data sits in your organization that we're going to call as part of this AI process.

Mark Smith: 31:10
If I have a single interface, I can build out my skill at working with Copilot, the UI for AI and, like you're saying, add the agents in as long as I've started developing. Hey, there's an agent that can do this and I can add them in, but I don't have to keep going. Am I in the right AI interface to do what I need to do? And I see the UI for AI as a compelling story by lighting up the agent experience that you can do with Copilot Studio, by bringing in the ServiceNow data, but I don't need to use the ServiceNow AI. I don't need to use the SAP AI if I basically can consume it through APIs. And this is where orchestration comes back into play, because when I'm talking about purchase orders or order numbers, we're talking about a connection to SAP here. I'm going to go and retrieve from that as part of what I do.

Dewain Robinson: 32:02
So I think it's close, but I think where this is going to this is a guess, right? So, just to be clear, this is how I'm thinking. This is based upon what I'm seeing, I think how we get there. It's not that I need to do it Like, let's say, servicenow has got this really cool AI that they come out with that does a really great job over ServiceNow data, because they know their data structure, they know how to search it, they know how to query it, and so they can probably write a really compelling AI for their self. But what they can't do is also bring in Salesforce. No, salesforce is the expert AI over there, right? And so the problem we ran into is I've said this many times I said the answer isn't one AI. It's not to build Iron Man's AI. Why can't I remember Tony Stark's AI?

Dewain Robinson: 32:55
So forgive me about my inabilities to be geeky here, but the idea is you're not going to have that, because humans speak with context. That's not spoken so, like if I say what is the finance rate? If I'm talking to an HR person, I mean something different than if I'm talking to someone in the finance department for the organization. One's corporate tax rate and the other is personal tax rate, and if I don't know the context, then it doesn't work. This is why co-pilot agents had to be incorporated with, where you can do the app and talk to a co copilot agent so that you can say I'm asking this to the HR thing, I'm asking this to that. So that's how we're handling the unspoken context, right.

Dewain Robinson: 33:49
But if you think about it, these sub-AIs there's no reason that an AI can't ask another AI a question and then use that as how to respond. So imagine if an AI under its function was to and it goes back to orchestration again, yeah, imagine that we have an AI that is a master orchestration engine that can also have good connections to call the API to another AI to ask it a question. Yes, okay, not to access the data, but to just ask it a question. I love it.

Dewain Robinson: 34:27
I love it. Now I get a response from the AI who really understands its data and its stuff, and it comes back with it and imagine, then I ask the question in order for me to answer that question, I actually have to go to three systems, three different other AIs, to be able that it has to go there to go figure it out. So what I'm think where we are right now is we're in the thousands of AIs, right, but the reality is we don't need one because the human context, unspoken context, problem. But what we might need is a few that that's the HR one, that's the corporate one or whatever.

Dewain Robinson: 35:07
So imagine, now that Copilot Studio starts to slot itself into a place where it is the conversational orchestrator for Microsoft that is configurable, is the conversational orchestrator for Microsoft that is configurable. And think about what I said about the explanation of all the different tools, and the AI has just become yet another tool, and now you can say I could actually create this. The reason I think it's going to work this way, mark, is I think this is what I told you at the very beginning if we do a callback to the beginning, the thing that makes me think this is the thing that was the unexpected value prop, which was even a lexical search, got better when I was able to have an AI build the query with the context. So imagine now how good an AI could build a question that it needs to ask another AI to get what it needs from the response. And so I think, if you think about it, that's where Copilot Studio is going to become like the superpower, because it will be the AI of AIs.

Dewain Robinson: 36:12
Yeah, it will be the thing that you explain how to go get all these things and how they all come together, and you're not going to program that in C sharp. You can program that in English or Dutch or whatever your spoken language is. This is why a lot of people need to be thinking about prompt engineering. They need to be thinking about that new skillset, which is programming in natural language. Being like a lawyer, how do I write an explanation that's very crisp and take out all the loopholes so that AI behaves how I want it to behave, because in that orchestration world, it's going to be super important. So this is the part that's actually going to become very interesting, and it needs to be a SaaS, because you need a runtime engine that's going to execute consistently, and so this is Copilot Studio.

Mark Smith: 37:08
I like it, dwayne, this has been awesome. I'm excited about this year. It's going to be an epic year.

Dewain Robinson: 37:14
Well, mark, I think everybody's going to think we pre-planned that question and I just want to go on the record and say, guys, actually not, this is actually just a really good riff with you, because it's interesting how all this starts just coming together, you know.

Mark Smith: 37:27
Yeah, I love it. Thank you, sir. I hope to see you in, probably March MVP Summit. If you're around, I'll see you there.

Dewain Robinson: 37:34
I will be there and I am looking forward to getting to see all my wonderful MVPs, like yourself and you know. If there's anybody that's wanting to kind of follow me, feel free to follow me on LinkedIn or, I think, twitter. I'm like Dwayne 76 and I spelled Dwayne really weird. So look at the show notes and if all else fails, you can go to CopilotStudioDudecom, and that's where you'll find the YouTube channel that we've talked about, and so there's a lot of stuff I'm going to be putting on that. I'm going to kind of double down on that this year mark, so I'm definitely interested in seeing stuff, and you need to give me feedback. By the way, you and all your community need to get youtube.

Mark Smith: 38:13
Stuff is amazing. Right, it's the co-pilot source to go to and to as your videos drop. It's epic. Thank you, sir.

Dewain Robinson: 38:21
Well, I appreciate it. You are too kind, sir. You are too kind, yeah Well, like I said, exciting 2025, and we can't make it this long again Between the next time I talk to you. Agreed, there you go.

Mark Smith: 38:35
Awesome, I love it, I love it, thank you. There you go. Awesome, I love it, I love it. Thank you, sir. Bye-bye. Hey, thanks for listening. I'm your host, Mark Smith, otherwise known as the nz365 guy. Is there a guest you would like to see on the show from Microsoft? Please message me on Linkedin and I'll see what I can do. Final question for you how will you create with copilot today?