View Recording: Harness AI in 2025

OK. Good morning or good afternoon. Depending upon where you’re coming from. We’re looking forward to having a nice conversation today about AI in 2025 and where we go from here. I am Nathan znozki. I’m concurrency’s chief technology officer. I have run about 75 exhibiting sessions with enterprise organization, leadership teams and a big question that comes to me is where are we going from here? What? ‘S What is the state of AI? What is happening within organizations? What’s the kind of progress that organizations are making and more and more like how our organizations choosing to invest? So the intent of this session today is to introduce you to some of the latest data that organizations are communicating about their priorities. To give you some perspective on the different modalities of how AI is being implemented within organizations today. And to give you the ability to think about that in the context of your own organization. If I would love for you to get a lot of the content that I produce, so I have a LinkedIn profile that is available right here. Happy for you to scan that QR code. What you’ll find is that goes to my LinkedIn. I have a weekly newsletter that talks about AI trends and innovation topics. I think you’ll get a lot of value from that. The last one was about Satya Nadella’s latest statement around SAS is dead, which I’ll talk a little bit about here, and then one before that was about 5 habits that you can build in your organization around about executing AI and helping them to be able to le. AI in their everyday work among about 20 other topics. So I think you find that really useful. The intent today is for you to get your questions answered as well as get some perspective, so I’d love for you to liberally use the chat. And or provide some some questions as we go along. So feel free to drop those in the chat. I’ll be monitoring that as we have a session today and happy to dig into different topics as we as we have the conversation. So thank you so much for being part of it. I’m looking forward to to presenting some content to you, OK? So what are we? What are we doing here? What I’ll be talking about is what is happening in the market. What should we be aware of? I want you to know why it matters. And then how to take action in a practical way within your organization? Also, all these slides will be available to you, so if it’s something you’d like us to send to you after the session is done, I would really love for that to be something that you can take advantage of in your organization. So make sure you let us know. And then before you leave, there’ll be a a survey. That will let you ask for those slides. As well as be an opportunity for you to. Partner with us in different ways to help make AI real in your business, and that’s that’s something we would love to do. That’s really what we live and breathe every day. OK. So the first, the first thing that you can see on the screen is a report coming out of the World Economic Forum and some information that relates to where priorities are with varying businesses across different types of industries. So on the. Bottom here you can see different industries and then on the. Sort of horizontal. You can see the different priorities that exist from a tech impact for those industries, and some of these are very emerging. So an example in emerging technology would be quantum and encryption. The upcoming impact of quantum isn’t really hitting anyone for real yet a because it’s just emerging. B. Because it’s an enabling technology that would take a lot of other things for it to be made real, but it’s on the horizon, whereas other things are less on the horizon. So what is the most significant activity across all organizations and especially across all industries? It’s the impact of AI. So what you’re seeing here is that. Very significant percentage of employers are telling the World Economic Forum. That that technology is going to be impacting their functional vertical in a significant way before enduring between 2025 and 2020 thirty. And that’s where a lot of their investment will be going. I sort of wonder like who’s saying no to that? In some of those boxes, but what you can certainly see is that there’s a consensus across many of those industries, or really all of those industries that this is going to be a area of big impact. What you can also see similarly is that robots and autonomous systems is right behind it now. I think that really combines 2 themes. The first theme is this idea of autonomous systems, which could be a digital autonomous system which is an area of very immediate impact from the context of AI, but also the idea of robotics being is something that is substantially influenced as a result of artificial intelligence in a. Way that it wasn’t possible before. Four. So you’re seeing so human equivalent types of robots that are starting to do work that were possible for robots to do before, such as working at Amazon Warehouse, being able to function in in sorting boxes and moving things and packing boxes that a robot wouldn’t really been. Able to do before, but now is really capable of doing so. You’re seeing both of those technologies. Being use cases that employers are paying attention to. So I think the first question that we all have to ask ourselves is what are my pains interested? Am I paying attention to that in my organization? You’re here. So I presume you are, but it’s certainly something that I think is really relevant. Now take that and move that into how organizations are thinking about the roles and jobs and skills that exist in the context of their organization and. With this report also coming from the World Economic Forum is presenting to us. Is that there’s a high percentage of companies across all industries that seek re skilling and upskilling being a critical a critical priority within their workforce. To to be able to respond to the needs of AI and the opportunity for AI to be able to have real, substantive change within their organization. So what they’re saying is that 77% of employers are communicating that reskilling and upskilling existing workforce. Is important to them in the context of AI right below that is this context of hiring new people with skills to design AI tools or pick out new tools that provide enhancements around the organization. That’s a really close, fast follow 69% right behind the 77%. So really those are two big themes 1 theme is. They understand that every employee has a need to be able to leverage AI. To be able to be the best version of themselves, and they also realize that the very nature of the business is going to use AI to be able to execute on its capabilities. And that might mean designing new tools or buying tools that enable you to be able. To make that real. And then right behind that is this question around, well, if we’re if we’re struggle to upskill our existing work forces, what are we going to do? And it goes to this point that we will probably need to hire new people with the skills to work better alongside AAI, and that 62% is really the companies that are either a reconciling that we’re going to struggle to do, that upskilling, which is probably every. Will have some of those struggles. Or their their understanding that there they may need a lot of that hiring because of growth or because of a reality around what they think that upskilling. It’s gonna look like in their business either way. We know that it’s a combination of three things. It is reskilling existing workforce. It’s hiring new people and it’s designing new tools that we’re going to use in our business or buying those tools. So all three of those things are really, really critical. And then of course, you can see some other kind of supporting scenarios that have a little bit lower level on the survey, but also really important in the context of the what the World Economic Forum is telling about. Upcoming jobs. This is very similar to what you’re seeing from other you know responsible or well, well put together surveys of it or tech, or even business leadership around the impact of artificial intelligence center organizations. So one example, certainly supported by the folks of of Gartner or Forrester or or what have you, other organizations that have similar similar kinds of perspective. So that I think is something we can all take as an asset. So what? I’m going to do now is is talk through each of those themes. Three things. One employee AI adoption scale. What has changed? What’s happening in that context? The next is around product engineering. So what are we? What are people doing when they’re building AI tools? And the third is how are they using data and preparing data to make both of those things true? So those are the three following topics that we’re going to spend the rest of the time together talking about. So the first of these is. I want to take you down a walk down memory lane for a minute. So who remembers? Who remembers this when? Steve Ballmer said there is no chance that the iPhone is going to get any significant market share. No chance. No chance. I remember him saying that. I remember that interview. But then I remember this one even more when he said it doesn’t appeal to business consumers customers because it doesn’t have a keyboard. Remember that when he was like he was kind of making fun of it. He was like, oh, nobody’s gonna use this thing. I kind of remember the reality of that where like everyone had a BlackBerry, everyone was was just type, type, type, type you way and there was all this concern that I actually thought it was so cool when I had a one of those pocket P Cs that had. The the keyboard on it and stuff and I could do stuff. I thought that was so cool. But I remember that because at the time it felt so artificial. To interact on knife like without that keyboard. And I remember my my first friend that had this iPhone we were sitting in a restaurant and he was showing to me and I thought it was just magical. I thought it was magical but what? It really dropped home was this idea of Universal App Store. It dropped home this idea of the universality of that like this was the new place we were going to get information. And see Ballmer, in this context, was what other way he’s trying to be a good sales guy. Or if you just frankly like just didn’t get it either way, he was wrong, right? He was in a position where, like, he was about to be disrupted, that whole market was about to be disrupted, but he couldn’t see the force for the trees in that same spot. We have some similar conversations going on today, tongue in cheek, OK. But Bennyhoff and Satya been having sort of this tiff, right? And this tiff has been around like. Copilot General AI is it the second coming of Clippy? Should we be focusing on just individual high first principles use cases or should we be building enabling technologies? The answer is kind of both, right? Like bennyhoff’s not wrong. It’s not wrong that like we should build principal AI systems. Yes, absolutely. But he is wrong in the context that General AI has a major place in the way that we all are building a capability to be able to be more productive. And I’m talking about that over the next couple minutes as we talk about employee AI adoption. So one of the most important things to reconcile as we talked about this employee AI adoption is the transition in skills and i’t this because going back to that World Economic Forum report, it’s the realization that this is a skills gap that we need to fulfill in our. Organizations. We need to be a very intentional about the transition from where we are now to where we’re going to be and that means that we’re moving from a current job that possesses a high degree of creative, repetitive work to a next job that starts a transition that Rep. Work to creative work, but it would be a mistake to even think that creative work is an area that we’re also not leveraging. Artificial intelligence, certainly that repetitive work is. Is fertile ground to go after. For automation and optimization, delegating the AI agents to perform activity that creative work is just as fertile. It just means that there’s a lot of it. So when we look at that future job, all of us are gonna really be very effective delegators. That’s our goal is to get much more effective at delegating to AI agents that can perform activities for us in gaining as the AI agent space gets more and more capable. The things we can delegate to them continue to increase. And that’s happened over last year has happened continually. We’ll talk about some of those things. So what are those kinds of activities that people are having to learn? Well, there’s five of them primarily. The first is having an AI agent. Research and prepare information or an asset. That’s sort of step one, right? So imagine I have an intern that intern can get things ready for me. I asked the intern to go get information for my meeting with my. Customer, they prepare that information, it substantially decreases the amount of time necessary for me to be prepped and ready for that meeting. And now I’m in a position where I can have an effective conversation with that that customer of mine. Sometimes we run out of time, sometimes I don’t have time to get that ready, so I have a less effective interaction or I have a person that does it all for me, or I have time and I have much more effective interaction. Our goal is to enable an AI agent to be able to. Give us the information we need to be effective. That’s a very like baseline goal of an AI agent. A second goal of that AI agent is to receive delegated management of post meeting action items. Or you could just call it like action items in general. Like it might not even be meeting. It might be action items that need to be taken based upon an interaction with somebody. I’m going to talk through some examples and I’m practically using today. In this context. The next is helping individuals in your organization to ask AI agents to take regular action based on a reoccurring situation. Something that happens a lot, I get this question everyday. Can this AI agent answer this question for this person who brings this question to my table? This is a process I run every day. Can I? Can I take action based on that process? Can I prepare this information every time every Monday I meet with this customer? Can I have this information ready for that meeting? I don’t even have to ask you. Just get it ready. Which then moves us into this idea of running a process autonomously. This idea of something that happens recurringly within my organization, then taking steps to take action on that autonomous. Activity. That activity happens autonomously on a regular basis without me even having to ask it, but automating something that I would have previously had to do some really impressive AI enabled RPA capability. Are becoming available that when you think delegating tasks to AI agents really can fit into that box and that’s this last task is getting this point of I wanna delegate not just a simple task but a set of complex tasks that all relate to each other and this. Is where we get to this idea of multi agent structures where I have an agent that does one thing. I have an agent that does another thing. So I have a we have a company. We’re working with right now. That is taking their legal sort of like the the when I’m entering into a contractual relationship with a company. Legal documents that I’m evaluating, they have a set of AI chatbots, AI agents that will inspect those documents for certain types of criteria. My net terms my this set this type of relationship, this type of approval maybe this this term that has to exist that that the fines are indemnity clause or whatever it is, right? So we want to evaluate the document for all these things. We do this for like writing a statement of work. I’m writing a statement of work for customer. I want to make sure the app met like we have. The check right, we have the math add up. Does this happen? Does this exist? Does this exist? Do I have any other customers names in the document? Like all these things, we need to look for an AI agent can perform those tasks and we can delegate that activity to them and prevent embarrassing situations where I sent a statement of work that has a different customer’s name in it. Or maybe the math didn net up and that looks stupid. I can avoid those things by applying an AI agent to it. So these are kinds of, but these are skills people need to learn these skills in order to be able to be effective. So I’ll ask this question. Can this be a moment where we enable every person to be the best version of themselves? And I mean that very authentically? When I think about this term, I think is this an opportunity for us to take the mundane activities that? Are either things we can’t do on time for things we don’t prioritize, things that I need done but aren’t a great use of my skills, and delegate those in the AI engine. And then use that information to maximize the potential of every person. And that’s really what enabling AI within an organization’s goal is. And some of that is accomplished and some of it is not based upon what AI can do today, but there’s a possibility of where this can go around many of those capabilities. And there’s a reality of. What it can do now, which is really impressive that we can enable every person to be able to accomplish so in order to kind of talk about that a little bit more. I’d like to talk about. Things that we do in the context of meetings. So these are these are six things that if you currently have copilot and. You’re in a meeting today. I will actively use copilot to perform for me, and it’s to a certain extent, like once I’ve started doing these things, I I couldn’t imagine not having copilots to perform those activities. There’s many others. There’s many, many others that you might be delegating, but these are ones that I do. On a regular basis. So the first one is. Did you join late? If so, ’cause that happens to me like one thing I will join. One thing will go long. Another thing will start late. Ask what happened already in the meeting, so I’ll say like, tell me what’s happened so far and it will give me a little update on what has happened so far. Under underappreciated but really valuable capability in the context of copilot. So that’s our sort of starting thing. Get into the meeting and join late. You need to know what happens somewhere. The next thing that happens is sometimes I’m not exactly sure what someone said. Maybe I didn’t remember it ’cause earlier in the meeting. Or maybe I just didn’t hear it well and I just made me had a thick accent. Or it’s just I wasn’t. I wasn’t listening well enough, ’cause. I was multitasking and I wanna know what they said. So to avoid the embarrassment of me asking that person to repeat what they said I asked. Copilot, they seem to have a better ear than me because I can usually tell me what a person had said. So good example of me being able to get someone’s answers of questions real time. But then maybe I need to also be creative. And I’m I’m I consider myself a pretty good question asker. But if you sometimes run out of questions but you, you know that there’s some out there you just can’t remember what they are. Maybe you’re asking what are questions to this meeting that haven’t been asked based upon the tone or based upon the relationship between the people or based upon the perceived outcomes that we’re trying to achieve. What hasn’t been asked? What should I be asking? Then moving this into summarizing the meeting in tuning the action. Items getting it to a point where we can really take action on it. And that being something that either is just like auto results or a scenario where I’m asking it for those action items and I’m tuning it based upon like that looks good to add this and I’m I’m adding as it goes and then I’m autonomously asking it to follow. Up on all my meetings with those action items. Send out the action items to the people that are added. In have a follow up with them. In X number of days, put a meeting on the calendar for us to follow up all those things that could have happened from that meeting. And then this is the one that I just discovered. OK, this last one I just discovered this. I was working on a project that was probably five or six different stakeholders across 10 workshops. We had lots of conversations and I was working on the road map for it, and there’s a question that I knew. We answered in one of those workshops, but I wasn’t sure where and I didn’t wanna have to like e-mail them because I knew they had asked answered it already. So what I had was the transcripts of all of my meetings with them. So I asked copilot. Generically, I didn’t even point it at the transcripts. I said go look at the meaning transcripts for this project and this customer and tell me the answer of this question because I need it for my road map and it was able to find it across all his transcripts and. I just find that so fascinating. And valuable that like the aggregate asset of the interactions that we’ve had is now an asset that I can go back to much more effectively because really like who listens to to meeting recordings like I’ve recorded a whole lot of meetings in my life. But it wasn’t until. I had transcription and the ability to ask have copilot ask questions of those transcripts that I ever listened to, a meeting recording, and now it’s like. Super valuable asset. So this idea of taking action for meetings, this is just like an example of one of the areas that you could use copilot for and really moving beyond the idea of copilot like that. You can use a delegated AI agent to perform. Amy Cousland 23:31 Nate, did you see the question about if copilot is listening to our meetings? How do I know? Nathan Lasnoski 23:32 So. Amy Cousland 23:36 Sensitive company information. Is now outside our network and company control. Nathan Lasnoski 23:43 A great question. Thank you for asking that, John. So the the most important thing to understand about copilot is that the only people that have access to that information in the transcript and the data is the people that were in the meeting. So, like, let’s say that like you had a meeting as a legal team and brought a certain topic, will that information is only available to people that were in that meeting or people that were then granted access to that? Transcript or meeting recording. After the meeting, so like if someone else went and tried to look that up. It’s not. Used like no one else would be able to see it, with the exception of the people that have access to it specifically. So that’s I think that’s one of the most important things to understand is private to your organization only, no one else’s. Let’s see the second part of the question is however, a legal team is having issues with Copa understanding a person correctly capturing what they said exactly as it did. We found it makes errors at times. And now that’s the record of hand for what someone said. This is an issue according OK, I get that one thing that I think you can lean back on is you’re also retaining the actual recording itself, right? So like is this a? Is the transcription official legal record? Probably not, right? The official legal record is actually the. The recording itself, what I found is that you know, despite where there may be like a disparity here and there, the value that I’m getting from that transcription is higher than like if I wasn’t doing it. So it’s a good thing to note. I think that’s continually getting better. It’s certainly better than most people’s memories. But if you really want to go back to it, you you just retain the recording and that’s you do have the ability to control that in your Office 365 tenant. Where you can keep time out. That’s exactly true. You do have a lot of capacity in your Office 365 tenant where we’ve moved to keeping them for a year. So if you want a a note on that, we move to keeping them for year. So if you want talk more about like retention and another. Thing that you we sometimes have done is we if we don’t want to keep every recording but we want to keep some as we built automations to take the recordings and drop them in the transcripts and drop them into. The the team for that project, so like it’ll automatically move it in the team for the project and keep it there. So it’s maybe another thing to think about. Great question. Keep on coming. That was. Thank you, Amy for for stopping me. OK, keeping going down the line though. So something else that I’m starting to get really impressed with is that, you know kind of when we started with like a ChatGPT is just it was just textual, right? Like and it wasn’t even like. Referencing grounded content. It was content that was just like, whatever, whatever was in the model. Now I tend not to use as many AI tools that don’t have grounded content. That’s up to date. So whether that means like information that can be returned from the Internet or information from the business system. It the the the model itself is sort of less important to me than the relationship between the model and its ability to relate to a knowledge graph of information. And. What we’re seeing here is the relationship between taking the the copilot interface or an AI agent of any type, and grounding that in data. In this case, it’s a really simple example where it’s grounding in an Excel file. But I’ll show you an example later where we grounded it in a business system where it’s getting data straight from a business system and turning that not in a text format, but in a graphical format. So here what we’re doing is we’re asking a question about information from the Excel file and then returning that information in a visual format that allows us to be able to take action upon it and gain new insights. And I think that’s really where things are continuing to. Move is. Not just having a text interchange. But also it getting closer to what I really wanted to to see and understand from that data. So that’s a really nice step forward that I think we’re continuing to see. And then I mentioned earlier this idea around taking action on recurring things. So something happens. You know what’s going to happen. You want activity to happen if you ever play with like. If this then that it was a a sort of web-based taking action tool but wasn’t AI based. It was very static like it had to be exactly this. And then you do exactly that. I’m excited about where AI is taking this repetitive task activity that I would have I would already. Performing. But now I wanted to take action based upon something that I’d already been thinking about summarizing communications or communicating with people on something, or taking action on something. So I think this is a really nice step forward. This is already rolling out. You should already see it in your tenant. If you don’t, it should be coming soon. Which then moves us into where I think there’s some really amazing capability, which is sort of the AI enabled RPA. So what I’ve done here is have a recording. Of recording or recording. Essentially, if you had a process that you were gonna perform on a regular basis and you wanted to delegate it to a person in your team, you would, like, get that person in your team and you sit next to them and you like, walk. Them through what you do, and then maybe they like write it down or like record the session so they can do it later. But you wouldn’t just like click on things and that’s your only recording, right? You would actually. Record. It walks through. You would describe it as you go. You would describe the intent and what this new AI enabled RPA capabilities are doing is it’s enabling all that to get brought together into a model. So when you build a recurring process that happens, that recurring process is built upon. And the description you’re giving it as well as the actual like following the mouse and clicking on stuff. And I think that’s really gonna take us from like the RPA scenarios we were able to solve before to RPA scenarios that can be solved by aligning AI with that intent and then allowing it to have more ambiguity. That’s part of the process that it follows to allow it to have more. Resilience in in performing automations? So I’m really excited with the where the future holds with this because it takes a lot of those scenarios where like you might have gone to a fully custom space because you couldn’t, you couldn’t build it yourself and like you’d have to build it to a point now. Where like because I can record the kind of activity I’d be performing to to delegate it to an AI agent as a power user rather than a than a developer. So lot that’s happening here, but this is really exciting for you to take a look at. I would highly recommend that everybody start. Play around with the new capabilities around recording your desktop flows. OK. So last thing in well actually not less thing in the context of employee AI was something really important that I think is is worth double doubting on is one of the struggles people have had is like capturing value not that value isn’t happening but like how do I. Know like how do I record the value? How do I understand that that’s actually happening in the environment? And so Microsoft’s come out with something called copilot analytics at issue. That should be rolling out. In your tenants as well. And what copilot Analytics does is it breaks down all the copilot activities into a whole bunch of like subsets of those activities. So you can start to capture like who’s using what for what, and the result of that is to be able to then cross reference that information against types of types of outcomes. So if a person is summarizing meetings and drafting e-mail responses and using copilot chats. And that person is in sales. I might want to know like, OK, what’s the relationship between people are actively doing these things and the the close rate in their time to turn a deal around their time to quote or the number of customer calls are on in a day. I want to understand the. First principle goal of maybe win rate against their usage of technology to be able to make that happen. In this context, we’re showing that. The high usage individuals are able to achieve more deals. One more value in deals 1A higher win rate. And that’s a type of information you’d want to understand in your environment and then work toward those first principles being true, or understand that like these people aren’t using copilot actively, or even if they are using copilot activity, maybe not using it well. And we need to get. Them more capabilities so they can achieve higher business outcomes. And but we wanna know what those business outcomes are. So we can hold our own selves accountable to making sure that this is actually providing value to the business. I think that’s so powerful. Just being able to to do that mapping. So this is a lot of capabilities you’re gonna be seeing roll out and something I think is really necessary is for us to be able to do value driven analysis of of what we’re what we’re getting. Oh, thank you for asking a question. Maybe I missed it, but how does copilot know if the deal is closed? Did I miss that connection? Great question. So what we would do is you would use your, you know your sales management system like CRM or Salesforce or something along those lines and you’d be able to correlate that person’s deal performance against their activity that they’re performing with. In this case, copilot could be, you know, other tools too. But like in this case copilot right. So if they’re a high cop, high percentage copilus user in these categories. I can then say, well, what does CRM tell me about your deal activity over that period and maybe even what I know about the the deals that you’re participating in with by using AI enabled tools, the right times and I can perform that correlation. So I have to use the business system to know that information but it gives me the ability to cross reference those. That’s really the end game, right? Like can I use my what? I’m measuring the person on which would be their deal performance from their sales management system. Against the tools they’re using. To be able to enact those deals. That’s, that’s where that would come from. OK. So in the the last topic in the employee area is this idea around training every employee to build delegated AI agents and then and you probably already saw this roll out in the context of your copilot interface, how you can build agents and build them via a self. Service sort of wizard driven experience. The intent of that is to allow them to. Build those. Build those agents themselves for many purposes. Assuming they can describe it right and it’s getting easier and easier to be able to do that kind of activity and then they might get a wall where they sort of run out of room and they need some help. And that’s where you might be building some integr. Or helping them with low code interaction with things. But if we wanna get them started in this idea that not only can you delegate to an AI agent, but you can think about the kinds of things that will be helpful to you and then delegate those activities to. An AI agent. So like perhaps I need a scheduling agent. That when I get a request from a customer about five times that I could meet with them, I simply ask my scheduling agent to give me those times I don’t have to go looking around on my my calendar for those I I’ve already given the scheduling agent the. Criteria that I would use like only within these timeframes avoid after 4:30 if I’ve accepted other meetings or if I. If I haven’t accepted a meeting, please consider it accepted. So it’s not something you double book me on. You know, things like that. So it’s like having an intelligent assistant that’s able to do that activity for you. That’s really where we’re trying to get to the AI agents. So that’s something that there’s a wizard driven interface you may have already seen this where you start to ask questions about or you start to describe what that agent is and it builds and what the knowledge is that it’s supporting. And that may be a Word document or. SharePoint site or Excel file or whatever it is that your business system that you’re connecting to? So the intention here is that you start to create these agents, lead qualifying next best action expense. Approving whatever it is that is an activity that you perform regularly within your organization and enabling that to be able to power better outcomes within the organization and it’s incremental like that whole legal use case I talked about earlier, like when you’re evaluating contracts like we started with. One thing you’re looking for, like what’s one thing? That, like every single time you write a document, you’re looking for this thing, or every single time that I’m writing a statement of work. Make sure the math adds up. You know, just do this one thing in this document and then start adding on other things that it’s going to perform in the organization. You incrementally get to a point where, like, wow, this is doing like 30 minutes of work for me every single day. And if we. Can get to that point. Then you’re truly able to see the real returns. I wouldn’t be overly onerous. About as you’re starting to think about like you’re just employee level copilot or other AI tool roll out. Don’t get overly onerous on like, my gosh, I got cost. Justify every single part of this to the business, like it will become self-evident. But then you need to measure it so you can justify it later. And that’s where that sort of copilot analytics comes in. OK. So in this context, ultimately what we’re trying to do is keep moving up the the value stream of what AI systems are. So like you start here. I’m I’m not using. I’m not using AI on the left hand side and I want to move to a position where AI is minimally a tool for me, right? It’s a tool where I can. Get information back I can. Like, well, one thing I will do like I’ll I was doing this this interaction building this this plan for a company. I said what like I want to know when I have my upcoming QBR with them, what their business performance has been for the last three months and. What the outlook is? For this upcoming calendar year and it gave me a series of references as to where this information is coming from and a good output of what their business performance look like. That didn’t make me tone deaf as I was having that meeting right ’cause I knew a lot about the business performance and where they were right now, so I wasn’t rolling in and this company had some struggles, so I wasn’t rolling in. Being like cool, you’re growing and. Everything’s awesome. It was like, hey, I know you’re you’re having some business performance issues on this area. Let’s talk about how we can. And optimize how you’re handling inventory to be able to make that better. So AI is a tool. AI is a consultant might be around how it starts to invoke suggestions around where you would take that information. So like, I’m not just asking for that, but maybe it actually provides some perspective around like here’s where I would go from here. I’m asking that next question right. Maybe it’s because I asked it to be a consultant. Or maybe it just gave it to me already. What would you do with this? What? What about this interaction? Can could I do better? Which then gets this idea of collaborator where sort of like 5050 it’s it’s an active participant in the context of moving my relationship or activity forward. And many of us will have a long way to go before AI systems get to that point. But it’s a possible future that is very realistic for many of the systems that we’re building, which gets us into these ideas of AI as an expert or as an autonomous system, which in some cases is very realistic. We’re building autonomous AI to perform very simple tasks. Or we might be performed building autonomous AI to do really complicated tasks that requires a lot of investment. Understanding that anytime you build something autonomously, you have a lot of sort of backing activity that needs to happen in order for that to be successful. Another question in the chat so copilot can read that sequel database to make the connect correlation. Thank you for asking that follow up question. Copilot can connect with Dataverse or other business systems to make that business data part of the resource graph. So yes, it could return. We do that all the time, for example like return information from this business system. To answer this question either at scale like across like multiple employees. Like we’re talking about the the correlation of business outcome and copyle usage, but also can be like any question that I might be asking like I want to return information from the business system. So I can make an intelligent decision on it too. So both of those use cases, it’s a. It’s a scenario that we can we can take advantage of. OK. Last thing around, employee productivity, I think I’ve said that a couple Times Now, but I I kind of forgot I added this piece into there. I think this is a really interesting possible future for AI is what if offshore development was replaced by delegated AI agents? What if? Offshore development was replaced by delegated AI agents. This is something where it’s almost a precursor. The the productivity that we’re seeing from developers using AI as part of. Get a copilot is a precursor to the same data that we’ll see proven with everyday employee activities. What we’re seeing with GitHub copilot is. Users who have access to AI tools in development consistently say that they are more productive, consistently say that they are more fulfilled consistently are able to accomplish tasks more quickly, and you know very quick example of this is a independent breakdown of developers split between even tasks and. Then evaluating the average time to complete those tasks shows that people have access to AI as part of their development tooling. Are able to accomplish work substantially faster, especially if. There are more junior level developer but also getting into like creative work or research work that a senior level developer senior level architect might be performing. Why is this so meaningful? Well, it’s meaningful because it it is continuing. We’re continuing to see the movement from AI as a like a tool to just provide answers to evaluating code, to looking for security flaws to even developing huge chunks of code, and getting people started. To now even developing as though I was giving requirements to someone and asking it to get started based upon the standards that I have in my organization, cuz I’ve grounded the AI agent on my on my code base. So we’re seeing quite a bit of ability to keep shifting our work into to an AI agent to be able to perform and get us started, get us moving more quickly. And to be able to eliminate a lot of like repetitive work. So like maybe I. Have a need to evaluate my code base for a certain type of security flaw. That’s something that something like a get up copilot could perform rather than me having to have an individual evaluate thousands and thousands and thousands and thousands of lines of code looking for that particular security fly can use get a copilot to be able to evaluate for that I. Also have found recently that the model diversity. In things like, you know, copilot, the ability to choose what model is the model that they’re using. To be able to has been a big advantage. Many of my our developers prefer one model over another, oddly enough. So it’s an opportunity for them to be able to take and gain for productivity from that. I think this is a precursor to everything that’s happening on the employee productivity side. OK. So the next section of this is around product engineering and I think product engineering is also sort of a example of this in the context of our forward movement. So what does product engineering product engineering is about enabling productivity, enabling us to build systems. So many of the things that we do are not just about. Getting employee productivity but about building systems that provide real impact that are unique to our organization. And that could be by leveraging data. It could be by leveraging ground up systems that were constructed. So what is what? What? Like how? What’s the precursors to doing this? It’s about actualizing the mission of your business. Any kind of successful AI project is about that. Our goal is to actualize that mission and enable us to be able to be more as a result of of that being true. So when we think about this, our goal is to create an innovation hub that enables us to put energy toward those goals. So when we’re building capability to construct or create AI systems. That are especially beyond low code, no code. Our goal should always be to put time toward experiments, to prove that those experiments are the right place for energy to be placed and then to put more energy into the ones that work and less energy in the ones that don’t, right? This is sort of obvious, right? But like building that muscle memory of taking ideas, moving those ideas forward. Into rapid pilots and then building and creating solutions. Is A is a skill that organizations may not have either. A cause you don’t have a development or build capability to begin with or B because you just have never really built a competency around experimentation. Even your developers have been oriented toward really well constructed, well, well, like defined projects, and you haven’t built that innovation experimentation muscle within. Your organization to be able to react to ideas. And that’s what we’re seeing. Organization, specially software digital platforms organization. Being able to take those first steps. So where does that come from? It comes from this idea of using strategic foresight in envisioning and thinking about what are the possible futures that my organization can be placed in really thinking about it. In that context, many organizations, when they enter into envisioning, they think primarily about the current business model, they and. They so running these envisioning sessions with executives. Immediately they go to like problem solving, right? Like, what are problems that I can solve with AI and many of them are sustained engineering, sustained innovation kinds of areas. And that’s there’s nothing wrong with that. Nothing wrong with like taking down problems. But the challenge with just looking at things from a problems perspective is sometimes you miss the the future business opportunity or the threat that’s the result of that or somebody else disrupting you so. What I encourage organizations to do is we have those sessions is don’t just think about like making a list of problems that I can solve with AI. Certainly those exist. Like I’ll be meeting with the company and they’ll say. What if I optimize my inventory? Awesome. Like perfect. Great use case. Do it like that is huge. Low hanging fruit. It’s there. There’s a lot that you can follow up, but what they might be missing is what if nobody wants your inventory in the 1st place because you got out competed in the market? As a result of your inability to respond quickly to your customers because someone else has changed the way that they interact with those customers, we have to think about what is the disruptive motion that might be happy in the in the market. That will be a threat to. My business and then I need to be able to respond to. So innovation is really all about enabling that to be true, enabling us to be able to respond to those activities. So the companies that are doing this well. Are the ones that I’m seeing creating the muscle around what the right ideas are? To put into their pilots 2 they’re building the muscle around how and how they’re building at scale. How do I create the engine? And then they’re pumping ideas through that engine to be able to achieve real value on the backside. So then what they combine with that is the ability to know that based upon the phase they’re in, they need to bring confidence to the the capabilities they need ready at each stage. So like when I’m in POC. I’m just trying to validate it works OK like I’m building. Is this possible? Can I get it to work? With them, by the time you get to production, you need all these other things. So you need to have the ability to reliably deploy it. You need to be able to do ethical AI evaluation. You need to be able to deploy old version of the future, the one you just deployed breaks something many organization. They start POC, they push it into production and they’re like cool. I’m done and they’re not. They’re not. At that point. So this is the big evolution that we’re seeing in many organizations, especially software, digital platforms, businesses that are needing to inject AI into all their the parts of their business. So this is the last topic that I’m gonna. I’m gonna really focus on today is this idea of that Satya Nadella just talked about and Satya Nadella hit on this topic. And I want you to carry this with you. As we as we sort of leave this session today. Topic, he said, is SAS is dead and. That interview, it was on the BG squared. BD Squared sorry podcast. The whole podcast was great, like he was talking about, like how we got into being the CEO and like the politics around all that stuff and like his ideas, his leadership perspective and it was just really insightful all the way across that whole, that whole conversation. But what landed was this idea around like Sass is dead and what do you kind of talked about? Was that all software applications we know today are just fancy interfaces sitting on databases? Forms over data essentially is what he’s getting at, but what he was truly getting at is this idea that we interact with these applications by inputting information into them and then getting insights back via dashboards and reports. And visualizations. But what’s happening is we’re moving into these agent structures where the way the data’s updated, the way that I will interact with these applications is going to be transformed by AI agents performing a high degree of those activities. So imagine for example, you have an application and it exists to provide you with with visibility into a particular business process. And in order to do so, you go into the application you type around and you add things. What if you never went into that interface? Like what? What if I just want the answer? I don’t actually need to know how to use the app, I just want the answer. Well, that’s really the future. Future is I need an answer to something so I can make a decision and maybe that answer is analytics. Maybe that answer is a prediction. Maybe the answer is a prescription based upon that prediction, and I want that information reliably back to me. What I might do is interact with an AI agent that simply goes to those various systems and provides that information back. So what he’s getting at is this idea that like we all on our own iPhones have like hundreds of applications, right? But what if I never even have to go into those applications anymore? What if I’m interacting with an AI agent that provides that information back? Or does that autonomous updating of a task or activity based upon information? So an example of this might be I was in the doctor’s office. And in that doctor’s office, this is the eye doctor. And he went to a digital system. So in this digital system, he would he would like talk to me, and then he swiveled chair to the digital system and type stuff in. And he come back and talk to me, and he swiveled chair to digital system and type stuff in. And I said, how do? You like that? Like how is this working for you? He’s like it’s not. It’s not working like cuz I am constantly having to swivel over to this digital system to input stuff and I can’t focus on my patient. So the next time I came in, he had. This assistant person, like another person in the room, and they were doing all the typing. And he was talking to me. And I’m like, what? This is like insanity, right? This is like the digital system’s actually causing the double the number of people he has in the room. But like what? What solve? What could solve that problem? Imagine you had an AI digital assistant that was listening to the entire doctor’s visit and doing that work for the person for the doctor. The doctors focusing on the interaction with the patient and the digital assistant is capturing all the nodes, updating the medical record. Communicating that back, maybe even giving suggestions and then all the doctor has to do is go and validate that that was true after that call. Now that may be terrifying to you, or it may be like really exciting. But think about the possibility of how an AI agent interfacing into that removes the need for the whole CRUD interface to begin with, right? Like I have an AI agent doing that task. And the goal is to transform the nature of work that requires a data state. It requires the application to be able to be interactable with the AI agent, but at the end of the day it’s changing the interface. It’s changing the way the work gets done. That’s what product engineering is really about. It’s about thinking about the way we do work today, thinking about the way I serve my customer today and pivoting and that’s why Satya gotten on this topic of SAS is dead. He got on the topic of SAS is dead because. He was like seeing this, this, this wave of change that’s going to happen. Around the applications that we work with on a daily basis. So all of that say, think about how this is gonna be true in your organization. Think about how your organization will react to these kinds of scenarios and position not only the applications, but the data that supports them for the mission of your business. So agents can be a a mechanism to be able to make this true in the way that you exe. For your business needs. So what I’m gonna do is is trail it off here. There is a whole bunch of stuff that we didn’t get today are on data that I’d love to talk about more. Maybe we’ll make that a follow up session is to talk more about data, but I’ll leave it in the deck. So I’ll I’ll just quickly burn through. These things, there’s a whole bunch of stuff on data, OK, you can get that in the deck. I love for you to take advantage of it. It’s really about like how does data support these goals? Think you’ll find it really useful, and what I’d love you to do is think about these ways of us taking next steps together. The first is if any of this content would be useful to being presented inside your company to some of your colleagues, we would love to bring the event to you. So this kind of content is things that we can customize, organize, we love to bring it directly to you. We also like to deep dive into any of the topics, so if you’re saying like, hey, man like, how do I take ground in employee AI within my organization? We want to help you take that ground. We know that there’s a tremendous amount of organizations that have not been able to gain that progress. We want to help you gain that progress and we also want you to understand what your Microsoft funding opportunities are. All these have the opportunity to be able to take advantage of Microsoft funding for AI adoption for building AI systems, for building data on those AI systems, and there’s funding available that can help. To reduce the cost of of that, increase the increase. The effectiveness of your of your roll up. So we wanna explore any of these things. So as you leave, please fill out the form AI. Wanna know how you liked the content? Was it useful to you? B deep dive into any topic and see. Let us know if you wanna explore any particular funding opportunities that might be on the on the stack. Any other questions? I don’t see any. I don’t see any if there’s one. I’ve got a couple minutes here. So if there were any other questions, I’m happy to answer them. Otherwise, please fill out the survey and will love to follow up with you.