speaker-0: Hey everyone, welcome to Chat B2B, the podcast that gives you a front row seat to AI and B2B marketing. I'm your host, Andrew Rau, co-CEO of Intercept. And on this show, we talk with enterprise marketers who are piloting and scaling AI use cases. We dive into what are they learning and what you can apply in your own organization. Today's guest is Jeanette Kennedy, who leads device marketing at Microsoft across their windows and devices business. She's on the front lines of the AI PC movement. helping partners and organizations understand what is the on-device opportunity for AI. What is a Copilot Plus PC and why does it need to be part of your AI strategy? We also dive into Microsoft's new Copilot Cowork and how it takes AI from giving you answers to executing multi-step workflows for you. Lots of great stuff to dive into. If you like what you hear, don't forget to follow or subscribe wherever you get your podcasts. This is Chat B2B. Let's get into it. Thank you for joining us. How you doing?
speaker-1: I'm great. I'm so excited to be here today. We're finally talking about something where markers don't usually put into the strategy document, which is the device and how important it is for their every day, right? ⁓ At Microsoft, really Windows devices are AI capable. So yes, you're using the word AI PC today, Andrew. I will allow for it. We don't use that term because really we feel like all of our, well, Not that we feel like we know all of our devices are AI capable. It depends on what type of device you have to really elevate that AI capability. So Co-Pilot plus PCs now is really that like ultimate experience and unlocks a whole new world of what AI can be. ⁓ And in full transparency, that term AI PC just oversimplifies it, but I'll embrace it today. We'll talk just for you.
speaker-0: Just for today, we'll make an exception. But I think you're actually hitting on a key point, which is people are referring to you in all different ways, right? AI PCs, AI powered PCs, AI ready PCs, there's Co-pilot plus PC. So maybe can you kind of break that down for us of, you mentioned AI PCs kind of undersell it. Where does the Co-pilot plus PC come in and what is the difference? between the two.
speaker-1: Yeah, so think of AIPC as a laptop with a third brain. ⁓
speaker-0: Okay, I like that.
speaker-1: So you got the CPU, you got the GPU, and now you have the NPU, right? And the NPU comes from all of our silicon partners. So all of them have it, Intel, Qualcomm, and AMD. And it's designed to run AI efficiently on your device. So what it's doing is it's taking some of those heavy workloads off of your GPU and your CPU and bringing it onto your NPU and dedicating that space to AI functionality. Traditionally, when you asked AI PC something, it would send data up to the cloud, it would wait for a response. Well, we have this thing called Copilot Plus PC now, which will actually run AI locally on your device. So it's faster, more efficient. It has a lot of privacy things built into keeping data, sensitive data safe. That's key. It is. And for marketers, you may not think about that. But then if you think about the work you are doing and product launches and some of that sensitivity you're dealing with, you want to make sure that your data and what you're dealing with is protected. also with productivity, we want that instance. We need that moment's response. And waiting for a server to do a round trip is like, sometimes it's like, wow, why is that taking so long? Co-Py Plus PC brings a whole new experience, live translations, AI assistant generated creative, is huge, camera effects. It really becomes your AI experience on the device. And it shows up at the speed of your thoughts with the right device that you have. You got to make sure you got a Co-Py Plus PC.
speaker-0: No, I honestly, think you're just in that sort of explanation. You're just showing like, you're right, AI PCs does undersell it because there's so much you can do now. I think that you're to your point on the privacy and security, like I think that's key because that, mean, I think the devices as they become more capable, they become sort of like a, you know, a bit of a threat vector, like people are going to, you know, they'll look at them as like a key endpoint to get at. But You know, the other thing that comes to mind is that, you know, the cost of to run these AI models, I think is going to go up. that's my perspective. think, you know, you've got like Sam Altman that recently was saying how, look, we sell ⁓ intelligence on demand. And so, I think that we will see companies that the major, that the frontier companies move from selling subscriptions to selling solutions and potentially Staffog, et cetera. So it might not be, hey, it's $20 a month, but $20,000 a month for a set of agent capabilities. And so I think that having this run on the device is probably going to be the most efficient option. I actually see it as, I think this will be the default option. Run it on device. can still push through the cloud to get the crazy frontier model capabilities.
speaker-1: for like
speaker-0: everyday capabilities, I don't know if you need that. Like I think you can process this stuff on the device.
speaker-1: We kind of break it down into two is like run it locally when it's speed critical. Like you need something that will start really fast. If it's sensitive, you want to run it on the device because you you want to keep that confidentiality. And if you need more stability now may not relate for a marketer necessarily, but somebody in the field, we all often have use cases around like insurance people that go to natural disaster area. they need it to run locally because there's probably no internet, let alone electricity in that area. Right. So that's a real use case that we often talk about is like people want to use AI in their workflows in their everyday. Well, if they don't have internet and electricity, how do we enable that and have AI on the device really does enable that person to be able to still function as if they were with electricity and Wi-Fi.
speaker-0: It's a point, because when you're in the field or these kind of edge use cases, you don't have connectivity, you still need that intelligence on demand. Like you get used to working, having access to AI tools, and then it's like maybe you're on a plane and Wi-Fi is down. Your productivity is severely impacted. Yeah, it has happened to me multiple times. And it's like you just almost...
speaker-1: Has that happened?
speaker-0: get a little bit frustrated with yourself because when you're so used to treating AI as a thought partner, it's like when it's taken away, you definitely feel the impact. You made a comment around, again, people thinking about the AI strategy, not necessarily thinking about devices or their copilot plus VCL, I'll say that. So when you think about the on-device AI experience, what is the big unlock that we think people might be missing. You mentioned, by the way, the translations and a bunch of other capabilities. Maybe we can go back over some of those. What's the big unlock here that we're missing?
speaker-1: lot of people think it's all about the apps, right? When they think about unlocking and really with some things we launched recently, which we'll probably talk about today as well. It's about what compute is happening. Like what are you trying to function? What is the workflow you're trying to bring to life? So it's not always just about the apps and a lot of people think, ⁓ I got to create the latest and greatest big app to really make impact. Well, no, take it back a second and let's think about the workflows that you run. and start shifting some of that into the use of AI. marketing is a high volume content engine, briefs, edits, creative development, analytics, meeting recordings, which I love. ⁓ my God, I cannot live without meeting recordings and the notes from.
speaker-0: That's the expectation. It's like, me the transcript. I need the summary and facilitator two and teams is amazing. Yes. I don't know. that's something. A hundred percent. It's like, Hey, you're halfway through the meeting. You've covered this. then anyway, really big fan of that too.
speaker-1: you on track. So yeah, but we're constantly challenging the word of change and local AI just kind of helps us manage those micro moments faster and smoother. And you know, we're really moving at the space, the pace that we need. And it's less about now the, transactional moments. And it's about context setting and having a conversation and empowering the agent that you're working with. to actually work with you, right? Like you would never work with a person, like say you had an assistant, you would never give them tasks without context. You give some details to what you need them to do. Well, that's the same thing with co-pilot is like garbage in, garbage out. Like you want one response, you put a response in, you're gonna get one response out where if you embrace it and start to work with AI, you know, it's just, it's really powerful, but I think a lot of people just aren't there yet. They're, they're just starting to uncover what that means. And then it's not to be annoyed by all the questions that Copa is asking you understand it's trying to set the context.
speaker-0: I actually encourage any AI model that I'm working with to ask me questions. I actually think that's the best way for it to understand your intention and the goal and the parameters and all that. Cause you made a really great point, which is would you treat basically a person this way of like, do this for me and give it no context, no examples, not of like, Hey, this is what I don't want you to do. And we assume that AI is just this super intelligence that will just get it. And because it's super smart, we just think it knows what we want, but it doesn't. I always say that. And how well do husbands do that? So it's like, you know, we've got to, like I always say is like, let's say you've got a new agent, how are you onboarding them? How are you training them? How are you giving it feedback? ⁓
speaker-1: should be reading our minds, right?
speaker-0: And so I think that we need to think about it in the same way. So you talked a little bit about some announcements and one of the ones hopefully we can get to is ⁓ Copilot Cowork. Yeah. That's a huge one. So for those, and I can just kind of set the context for those who haven't heard about it yet. So I believe this was built in partnership with Anthropic, right? So they've got Claude Cowork. You're able to run multi-step tasks across your M365 suite. Right. So all like your Outlook and Teams, Excel, all the apps that you're in all the time. To me, I think this is a huge breakthrough from telling you stuff to doing stuff for you. But maybe you can give a broad overview and then we can just talk about how big of a leap this way is.
speaker-1: Yeah, it is a huge leap. It's going from transaction of answering questions to now executing and being that assistant for you. So we describe it as turning your your intent now into a plan and progressing with that plan with checkpoints to make sure that you are in control and that the final output is what you have. thought in your mind. Let me go back to reading your mind, right? Because we all have a thought of what we want. We just are trying to figure out how to work with Copilot to get it to that point. Well, this actually does help us now. Cowork will take you, like you said, work through all of the applications too. So it's not just you're doing one prompt in Word to execute that piece of the puzzle. You now are using Cowork to execute a workflow across multiple applications where you actually don't need to tell it. where it needs to go. It will actually try to figure that out based on how best your context is that you're setting and answering those questions. for marketers, it's like, it's, used to do singular things. It was help me draft this launch email for my campaign. Now it's help me run the campaign workflow end to end. Let's plan the prep meeting. Let's do some focus time. Let's assemble those inputs and what are those outputs and those drafts and what are those followups? Like it's a entire workflow that you're trying to build. And you don't have to know how to build agents anymore. You don't need to know the back end of that, right? Some people were really scared. ⁓ my God, coding, like, what am I doing? You don't need to know that. You just need to be able to be very detailed in setting that context, treating it like an intern checklist of like, these are the things I need it to do. And really providing that purpose and the approvals and the steering of those response. questions that you're getting back and forth. And don't get annoyed by all the 20 questions because I know we simply get annoyed by 20 questions all the time. I have a toddler, I know what it means when you get 20 questions. But again, it's to make you more successful and ensure that that workflow is functioning the way that you expect it to function. So it's prompt and response has been helpful and great. you know, my dad has figured out how to use it now. So it's fantastic. It's reached all age demographics. ⁓ And now it's the next evolution. It's the plan and execute. And it's going to be transformational for sure.
speaker-0: I'm so excited for it. actually, so I haven't had the chance to get hands on myself, but I have heard from folks that it is incredible. That it is really, really, really good product. Like, I think this is where AI really becomes useful. Like, obviously the OpenClaw stuff is blowing up right now where you can have agents running on your computer. It's really the same concept except for, I mean, let's face it, as fun as OpenClaw is and it is a transformational technology. It is just not secure, right? Like we just can't run any client workflows, any of our workflows on OpenClaw. Like, okay, it's a personal project, even still, you just assume that, you know, let's just say it's breached. That's kind of how you need to assume with OpenClaw, but cowork is completely different because it's obviously connected into your Microsoft tenant. But one of the things I was going to say is with cowork, think, sorry, with AI in general, There's been a lot of studies that have showed like the HPR, the Harvard Business Review study. There's one that looked at a US tech company and I think they tracked it over eight months. And they basically said AI allowed people to work faster, take on more tasks, but it was sort of extending work into personal hours. it actually, what they said is it intensifies work. Now, I think that with cowork, it can actually do stuff for you. I think that gets us to this, we get to live a more balanced life potentially, because it's not just prompt and response and then you got to take it and move it over here. Like it's actually doing some work for you. So that's why I'm so excited about it. And for the folks that are like, well, I don't know how it works. It's kind of like, do you know how your car works? Not really. I like you probably get the fundamentals, but you, if you probably don't know how your car works, but do you drive it every single day? Do you your kids in it? Do you like, do you trust it? Yeah. So I don't know. That's kind of a parallel. Okay. Marketers, look at, ⁓ copilot plus PCs. What are the top use cases today that, that, that they can be doing? Like, how can it change their work today? What's, what's your thoughts?
speaker-1: Yeah, so creative creation and you know, we use the word creative. Everyone thinks like visuals, but think content as well and messaging. you know, faster ways to do first drafts. So maybe you need a first draft before going to an agency to get buy-in internally on an idea, right? You don't need to be that creative expert. You can bring somewhat of the idea to life using AI and then using different variations, prompting to get it to where you need it to be, know, listening to the CEO or the LT, bring feedback in a meeting and then take that feedback quickly and adjust and then you can get it to a point where you get that approval. So, ⁓ you know, we're all about in the marketing world testing and quickly adapting and AI can do that. It could be a couple words, it can be the whole creative ⁓ and you can do that more in-house to be able to have those abilities to quickly adapt. You may not always have budget to do things externally with agencies. You might need to do some of those small things in-house. When it comes to meetings, we mentioned earlier how we rely on it, but turning calls into those briefs. I've done that before because I'm very good about setting context verbally and talking about something and getting excited about and feeding off the room. then I'm like, damn, what did I just say? I got to put that in a brief now. Well, recording that meeting, takes all that and then you can adjust it using prompting to get it into a brief format and not lose that momentum of that fantastic meeting you just had and moving forward. Video, we got it today. Better camera, better captioning, better accessibility. And for me in a global team role, when I go and present to Asia as an example, they have. the captioning on, they are making sure, yeah, they're making sure they can understand what I'm landing. And ⁓ I don't have to necessarily worry about it. They're using the technology so that it's adaptable for them and they understand what I'm trying to convey and land. And I think things are landing a whole lot better now, which is fantastic. ⁓ And then that sensitive work, we have product launches all the time and we talk about local. AI experience and having our launch plans and our strategies and being able to bring it onto your device ⁓ and just function a little bit differently without, you know, the multiple share points with multiple access points and that stuff would walk down. So it's just, and this is faster without sacrificing quality really. They're having that copilot plus PC and really everything I just talked about to some degree. You can use that across any AI PC, but we're at CoPilot plus PC. There is an elevated experience across all four of those types of use cases. And we built a great tool so you can help understand what a CoPilot plus PC is because the... It is tough sometimes to really know what are those additional experiences that you can light up on your device. So we're doing some great work at try to bring that more...
speaker-0: Differences, yep.
speaker-1: more into the use case scenario across multiple roles within a business.
speaker-0: Yeah, not to oversimplify, but I think that if, as we are trying to get more richer and more powerful on the Visee experiences, you're going to need more power. You're going to be, and again, I don't want to get into the whole tops thing, but you're going to need more performance. And I think ultimately that is the thing that Copilot plus PC is unlocking, right? Is allowing you to be able to do a lot of these. Exactly. Your third, exactly your third brain.
speaker-1: third brain. ⁓
speaker-0: And you know, the translations, it's funny because some people are like, well, know, AI power translations are not quite there yet. It's like, look, this is the worst it's ever going to be. And it's already pretty, pretty damn good. It's really, it's again, month over month, we're seeing improvements, right? Like drastic improvements. So that's the other thing too, is like what you may have experienced, let's say three months ago, four months ago is not the same, right? ⁓
speaker-1: It's getting faster too. we, our timelines are going shorter and shorter on being able to deliver localized assets to our fields because the workflow of the localization process now has AI within different touch points of that process to speed things up. So we've seen remarkable change, quality timelines and different efficiencies all around.
speaker-0: And now, like in your example of taking the transcript and creating the brief, now even using co-work to update that in the exact template that you want, update that file, do it locally on the PC. think it's just going to be ⁓ a game changer. for us to intercept ⁓ OS, it's an AI-powered platform that we use to run a lot of the agency solutions. But the vision is, and talking to our CTO, Francis, about How do you put this powerful capability into the hands of smart people, right? ⁓ Of the team. How you distribute this though, what we're thinking about is how do you run it locally on device? To be able to have the cloud access in some cases, but a lot of the everyday stuff, being able to run it locally. And I think that's, you I know that you mentioned Cyverns, what's the app? What's the killer app, right? I mean, it's not only about apps, but go build it. You know, like it's never been easier to actually do that. So that's the other use case that we're seeing is, is we build these, this operating system. How do we actually install that locally on the device? And this is where Copilot plus PCs has really been like a key consideration for us. but yeah, so lots of, lots of, lots of use cases. think, ⁓ the other thing I was picking up on is again, you went back to sort of like the questions and making sure that the, you know, the model understands what you want to do. I think a lot of marketers are not really thinking about AI as a thought partner. Like it's, you know, it's that 10 % stuff of like, okay, maybe write my email or whatever, but AI has been trained on the equivalent of like 200 million books worth of data and it can recall it 100 % in seconds. So get it to spar with you, get it to ask you the questions and like even clarify your own thoughts, like you said, it's easier to sometimes talk it out. So even turning on voice mode and being able to just chat it out. Do you ever do that with yours? just curious. is it just me? have. Okay. Yeah.
speaker-1: No, it's a nice little companion to have some time to time and throw ideas at it. But I have done that where you do the talk. It's great because it actually does transcribe it as well. It's not like you lose, you're just making it easier for yourself because you're actually just talking out loud versus having to type it all right.
speaker-0: Yeah, yeah, 100%. So lots of use cases that we can run on device. What does that mean for cloud? mean, how does a marketer think about what do I run on the device versus what I run in the cloud? Is there a framework or how should people think about that? Because now everything just run locally? I don't know. I think there's also a place for cloud too. What you think?
speaker-1: Yeah. So we talked a little bit of the locally already. It's like if you're doing stuff that's speed critical, you want to do that locally sensitive stability. Like if you are doing repeated tasks, you want that stability so that the job can get done. For running on the cloud, think of supersized and you need huge reasoning or big generation, heavy multi-modeled ⁓ execution. ⁓ Also, cloud is still really important for sharing and being able to share what you're doing with your team and using data that is part of your organization. So there is a connectivity you need to bring in some of that intelligence of that data because you probably don't have it locally on your device. And if you do, I hope it's protected because you need to make sure that data is not just out there on any device that. could get stolen, and you never know where that data goes, right? ⁓ But then if you need massive compute bursts, you don't really want that on your device. It's going to slow things down. You need the cloud to help you with that. So there is some balance to it, but you'll figure it out. It's really just managing it in your day to day to see what you can bring locally onto your device. ⁓ and playing around with it. So, reserving the cloud really for those heavier lifts. So, local is for speed and sensitivity, cloud is for like scale and complexity is my little sound bite for you today.
speaker-0: I actually, I'm going back to the whole car analogy and just thinking about it. It's a bit like a hybrid car, right? Like if you want more power, then that's where gas kicks in. Maybe that's the equivalent of, I don't know. And then it's like the, you know, the electric version is if you're going slow speeds and stop and go traffic, that might be on device. Yeah, stability, right? ⁓ But you know, and it's funny because go back to the whole like security and privacy thing. I mean, I know that the AI providers say that they don't trade on your data and I do trust that mostly. think that you do see articles that come up like in routers that talked about, I think it was Meta where they were training their image recognition model and they had like a couple hundred people from like Y Pro. I think it was a team in India that was actually reviewing all of your pictures to train the model. So it's kind of like, whoa, whoa, whoa, whoa. There is a, I guess anything really that you put in the cloud. mean, obviously there's security protections and all that, but there still is, you have to, it's not as safe as obviously having it your device. And I think that that just security and privacy is just going to be a really important thing. So.
speaker-1: Yeah, because think of two things, device loss, like if you lose the device and data sits on that. But then security foundations is the starting of your AI journey. So if you back it up a little bit, we consult a lot of businesses on building up their AI frontiers, the way that we like to use these days. The first thing they do is to check about security and data compliance. And if you have a garbage foundation, you're not going to be able to build upon it. our consultants are constantly working with businesses to really make sure your data infrastructure set up, or making sure your security foundational things are set up before you even adopt what AI means and go into this wonderful world of AI. AI doesn't change the need for good security hygiene. It makes it more important.
speaker-0: Yes. So that's actually a really good point. So in terms of security considerations that we need to be thinking of, you mentioned sort of hygiene, infrastructure, making sure that we're set up to do this securely. And any other considerations we should just think about as we think about on-device AI? Like, how do we just make sure it's protected?
speaker-1: Well, hopefully your business has implemented some of those things of like Windows Hello, right, that you can't just walk up, anyone can walk up to your device and plug in a password, like you want that visual identity, right? We have a great video coming out about that, by the way, it's fantastic. ⁓ And then really about data, what you can store on your device. And there's some documentation you may not be able to store on your device, and it needs to be locked up in a SharePoint in the cloud. it's managing the definitions of that documentation. You know, what's highly confidential, confidential, general, public. And we use data labeling quite religiously across all of our content because like even if I wanted to go change a document security label, I have to get justification as to why I'm changing it. And in most cases, it's because I identified it wrongly from the beginning. So it's like oops, mistake, but they do want justification as to why things are changing. But that labeling just brings thoughtfulness to how you're managing that document that you're working with. simple things like that that may not necessarily be a standard in a business practice and in your workflows, and it could save you a lot. So it's the little things here and there that really do matter.
speaker-0: I find that ⁓ talking to a lot of IT folks is that they have a lot of security capabilities that they have access to that they just don't use or they haven't turned on for some reason. And it's like, it baffles me because it's like you've, you've kind of already paid for it. So why aren't you, why aren't you turning them on?
speaker-1: It's the training, honestly, being able to train your staff, your teams about how to use data labels as an example, right? Like, what is the right context for each of those settings? How to apply it? How does it restrict your usage or sharing, right? ⁓ So it is really about setting up that great foundation of training everybody on how security should be top of mind for them as much as it's top of mind for an ITDM.
speaker-0: Yeah, makes sense. So clear message here, which is as much as we all want to get to the, you know, the novel, sexy, fun stuff, you got to make sure that you're ready for it. Right. Like, make sure you're set up. You're not going to blow anything out. ⁓ again, I go back to this concept of a killer app. I always hear this of the next generation of apps are going to take advantage of the the NPU, right? That will be able to run or dedicated to run a lot of the AI processes. But a lot of the apps are not there yet. And so I guess the question comes up of like, invest today versus maybe waiting until we see what some of these killer apps might look like? What's your thought on that? What's your response to that?
speaker-1: Well, AI is here now and the platform shift has already started. So you can fall behind very quickly and marketers who adopt technology early on will win, right? And you don't want to fall behind. And I'm not putting pressure on anyone. Like I feel it sometimes too where I'm like, ⁓ I really wish I knew how to build an agent. I really wish I knew how to, and I actually had this conversation with somebody recently like, ⁓ we want to build something in six months from now. Can we start sculpting it now? And I said, absolutely not, because I don't know what the technology will be in six months from now. Like we, we might be able to build something totally different because we don't know what that technology is. Or, you know, cowork as an example just kind of came, well, I'm sure there was a long road to here, but we only learned about it recently, which changes potentially the need for agents and how you might be able to use coworker. for those workflows. So now my mind has shifted to do I really need to learn how to build an agent or do I really need to learn how to manage my workflows within cowork, right? So when, these capabilities start to standardize themselves as in like what's old versus what's new, know, the muscles have to change. You have to adapt quickly on what is current, what is most useful to you. experiment, play around with it. And I think that's where people get a little scared. It's like, I don't know how to experiment. I don't know how to play around with it. Like, am I gonna break something? And I think often we're in the waiting game and you can't anymore. Like you gotta try, try. I say it to my three year old all the time. You gotta try. You gotta explore. you know, commonly we hear don't have time for it, not sure what I'm doing. ⁓ You gotta make time for it. You gotta figure out, carve out time in your calendar, just kind of play with it. What I do now is if I need to start something new, I go to Co-Pilot first and I say whatever I need to say. They may not get it right or they may not wanna do it, might do it the right way that I wanna do it, but at least I tried it and I have a starting point. And maybe I figured out something new I could do with Copilot while I experimented on this other task. just always going AI first with anything I'm doing has opened some doors of exploration.
speaker-0: Yeah, I mean, there's a few really good points here. think one in which is you can't afford to wait anymore because things are moving so damn fast. But ⁓ also, you could always ⁓ make a case of why not do it, right? Like there's always a reason why to wait because it reminds me of the Matt Dixon research that was what was this book called? Jolt Effect, where he talked about I think it was like up to 60 % of qualified pipeline, you will lose not to a competitor but to indecision. Because he basically says it is safer to make no decision than to make the wrong decision. And so it's this concept of he's like, it's not fear of missing out, it's fear of messing up. So this not FOMO, it's FOMO type of thing. But it's true though, I think that people are generally unsure. And so it's safer to just, you know, not act, but I think that you will get left behind, right? Like it's not meant to instill fear, but if you want to be, you know, relevant, even just in your own personal career, you should be advocating to have access to copilot plus PCs and like the on-device AI, because ⁓ things are moving so quickly and that you want to stay relevant, right? So. And again, the laptops as well, you're not buying them every year as an organization, you're buying them in a cycle. So if your cycle is now, I don't know why you wouldn't.
speaker-1: Yeah, you're buying for the next five years and imagine we're going to be with IEI in five years. Like you need a device that can manage that.
speaker-0: Right and so otherwise you're just kind of buying it twice like you're buying something that can be fine for now Then you're like, ⁓ my god, I need to buy something else ⁓ Now I wanted to also go back to your point on apps or agents and whether you know to know how to build the agents and and things like that ⁓ If marketers are thinking about in that position of do I build my own app instead of waiting for the whole the killer app that's coming What's your guidance on this? Is it like hey? look at what coworker is doing is like that's kind of an orchestrator or your chief of staff in a way. It's an agent and you can teach it to what to do or how can they build their own stuff? What's your perspective of that?
speaker-1: It can be daunting because you're like, where do I start? ⁓ So my recommendation is looking at your workflows, looking at what you do today. Let's not create something that's supporting something you want to do in the future. Let's look at things you do today, end to end. Pick a pain point. Is that briefing? Is that the content strategy? Is that an event follow up? Like pick a moment and start with that workflow. Don't think about the tool. Don't think about what the out. where the final thing is that can fix your problem. Think about the problem itself first. Describe what that problem is in great details. Defining what is those inputs that you need to make that workflow successful? What are those standard prompts and templates that you use to support that workflow? ⁓ What is that automation you're potentially wanting to get out of it? And then what is that impact that you want to see this workflow? potentially take away from the rest of your team. Like maybe it's hours, maybe it's time, maybe it's just better detail. And managing that workflow, defining it, and then trying to figure out first and foremost, how can Co-Pod, how can coworker help you with that? So experimenting with that one workflow. Let's try not to dump everything into it. Because that can be overwhelming. You want to perfect one thing. And then staying with that workflow to build upon that and making sure it's doing what you want. Don't be like, ⁓ we get so excited that you got one thing kind of functioning using AI and then start to go off to the next thing and trying to figure out the next one. Perfect that one. Make it perfect. Make it your use case internally, right? To really showcase the value of AI because people can poke holes easily and you want to try to show the best in class. is how we're using AI in our org, our marketing org. ⁓ and then get that buy-in and support and trust amongst your organization to then go and build more things. Because sometimes it might cost some money to build some of these AI tools, depending on what you're trying to accomplish, where maybe your first couple, you can get away with doing them amongst your team and your organization. But you're building that trust that you guys, as a marketing organization, trust AI, use AI, are experts in AI.
speaker-0: I think it's a great point on really understanding what the problem is you're trying to solve. Like you actually said, like, don't even think about AI, like actually abstract it from the thought process. What is the problem you're trying to solve? And I think that that is where you're going to drive impact. feel that AI in some ways can be a bit of a checkbox in that everyone's like, well, I got to do something AI. You know? And so I think it leads to kind of like AI theater. a little bit of like, ⁓ yeah, we're doing this AI thing here versus like actually trying to drive impact. And I think the other point that you made around, you can probably do some like onesie twosie things on your own, but as you think about scale, ⁓ you need a framework. And so I know, again, Francis on the technology side, you know, he's a huge GitHub fan. And so working very closely with the GitHub team in building an environment, that they've developed that allows us, who are the non-technical folks, to build application solutions. Like we can basically vibe code different things as a line of business. And we do it within the environment, within the GitHub environment so that it's ⁓ safe and that, again, you're not going to create any security misconfigurations or some back doors that basically you get out. Exactly, right? So I think that you too need a framework because the vibe coding stuff is very empowering. But if you don't govern it, it can also be dangerous, you know? Yep. So I agree. You need like proper frameworks, know, and foundations to get it right.
speaker-1: And we've started an intake process because everyone's been wanting to build their own things, their own agents, their own stuff, make it easy for people to use it internally. Then people are wanting to make things now stuff for our partners to use. it's like, okay, how much stuff do we have? Are we having now? Like it's like, so we have an intake process now where these types of things are, are, the questions they're asking so that we are making the right tools. We are, replicating the right workflows, or it's like, I have a great idea of what I want to do, my coworker has a great idea. There's actually tools or processes or workflows that we could probably bring these two things together. And we just don't, we're just not talking and bringing those ideas together. We're an intake process and a team that evaluates what are those AI tools or workflows that we're going to empower our internal teams or our partners to use. they have a larger vision on that and can it help us make those decisions and start to bring things together where it makes sense.
speaker-0: Yeah, yeah, it makes sense. And we have a similar intake process. That way you are not building the same things. And that like, right, it happens. But then also it's like people have these really great ideas that when you combine them together, it's actually one main thing, you're saying, right? So that's actually a good segue, because one of the things I was going to ask is that, is you think about where AI PCs go and it becomes, I was reading some IDC research that I think it's by 2028.
speaker-1: It happens.
speaker-0: 60 % of total PC shipments will basically be, you know, AI-ready in some capacity. What do marketers need to do? Like if it's coming anyway, and you'll likely get access to it, ⁓ how do we just be well positioned to actually take advantage of this stuff? You mentioned security before getting infrastructure ready. Like what other things can we do so that we're ready for the moment? Because I feel like it's going fast.
speaker-1: It's already here. Well, at least on how we define AI PC, it's here. ⁓ Everybody can get an AI capable device. It's a question of like, do you have a copilot plus PC? Which yes, that roadmap is looking like more and more all of our devices will be copilot plus PCs as all of our silicon partners catch up with the components. But yeah, you do need to be ready as an individual, but also an organization. ⁓
speaker-0: Yeah, it's here.
speaker-1: It's, think about operating system and like talking about devices. So we think about our device, but think about your operating system as like your day to day. So what are your templates? What did you, what is the module messaging that you always use? What is the reusable building blocks that you have? And really bringing that all together as your template and your foundation because AI works and is super powerful on. repetition and scaling. So if there's some things that you can standardize, absolutely do it. We don't want to limit differentiation and creative, but there is going to be an ability that you can work a whole lot faster if some things are just set as the core. ⁓ And then what are those guardrails? And I've done some great work with your team too is like,
speaker-0: Yes.
speaker-1: What's allowed? What's sensitive? What must stay local? What can go to the cloud? Defining the dos and the don'ts of what your AI engine will do for you is going to be super, super important. And then training, thinking about those scenarios. What are those workflows? What is it the campaign planning? Is it the event execution? ⁓ my god, that takes a whole engine to do. That is crazy. Is it the partner co-marketing? Like I'm already trying to think about some of the things I can do to really enable our partners because our partners, there is a vast range of different capabilities when it comes to AI. I have some partner marketers who are just figuring out what co-pilot is and what prompting is. So I'm creating simple prompt guides to help them. Right. And then we got other ones that
speaker-0: meeting is where they're at, right?
speaker-1: Yeah, they have access to co-work and great. Now we can give them the building block assets versus giving them fully baked campaigns because they have the workflow abilities to use AI to build the assets that they need. ⁓ And then also thinking about not just all the pre as a marketer, how to use AI for the posts, the stuff that not all of us love, which is the reporting and the metrics and the data and. Right. Using AI to help you. So building out all those different scenarios of your entire team, of the workflows that they all have, and just writing things down and having that documentation so that when AI comes in and is kind of your everyday moving forward, you have some of these foundational things as an understanding of across your team, across the leadership teams, across the ITDMs, so that you are set up to kind of start fast.
speaker-0: You know, it's, it's, ⁓ what comes to mind is that it's not a device refresh moment. This is actually like a cultural refresh. It's like a department refresh. Like if you really want people thinking AI first, I would say use this as a moment to do that. Right. It's like, look, we got device refresh here, the new things that we're going to do. Here's the intake process of how we're going to think about building new, new agents or Gentoo capabilities. Right. I think it's a really, it's a good opportunity for us to like, you know, like create a, yeah, to create, to create that, that, that reset because there's just new capabilities here that we we need to take advantage of.
speaker-1: in the go, like always trying to get the next campaign out or get the next creative idea out. Yes. There needs to be a moment in time if you're going to adopt AI that you do that reset and you properly onboard what AI means to your team. It's not just plopping it in. Yeah. It's thinking about it. You can't just all with it, right?
speaker-0: And I think the other thing we're seeing too is like really intentional how you message it to your team. Because I think that, you you talk about meeting partners where they're at. I think it's also meeting team members where they're at too. Like where there's still a lot of this whole, ⁓ there's still fear. There's this human reaction to be scared of change. And so it's like, ⁓ well, what does that mean if let's say I'm a partner? Partner marketing manager. And so if partners can do a lot of themselves, what does it mean for me and all this kind of stuff. I think like messaging that I think is really important too. ⁓ I wanted to ask you, I mean, as we start to like think ahead, like where stuff goes. And so you've got, you know, the large language models that are advancing all the time. And then you've got the small language models. Where do you see this kind of playing out over the next couple of years? Where do you see sort of the trajectory of these things?
speaker-1: It's really getting intelligent around what is the best model for what you're trying to do. Getting the right model for the job at hand. So if you're not familiar, large models are really for those broad reasoning, complex, ⁓ multi-layer executions that you're trying to do. Small models are really good for like focused task, repeatable, scalable, potentially could run locally depending on how. Your device is set up, right? So the future for marketing is those small models really power your everyday workflows. What are those things that you have to do every day? May sit more in the reporting mechanisms, More sit in those repeatable functions that your team is doing. Where those bigger models, deep understanding, heavy creative lifts, that's probably where a lot of those larger models will sit. big models, big questions, small models. content or constant momentum and change and adjusting. So I like that. Yeah, it's really making it fit for the job that you're trying to accomplish. So I think we go back to like the question we asked earlier is like, how do you set yourself up for success? What are the things again documenting your workflow is documenting what your team does. So then you can find the right models that fit for each of those different functions and workflows that you're trying to accomplish.
speaker-0: You know, that's actually something else I want to touch on is the, is the, you know, the SOPs, the standard operating procedures. A lot of people want to automate stuff. And one of the first things that we, we, we look at is like, well, what are the SOPs? Cause AI is really good about, you know, following a set of instructions and workflows. But if you don't have a sense of how it works today, if you don't have it documented, we know it's going to change, right? With AI in the loop, it's going to change. But if you don't have that documented, you don't really have a good start point. So exactly. and, to the thought on the, the small language models, like when I see that basically the models every year they get about, they drop about 10 times the cost, right? So it's like, it's 10 X cheaper every single year. And so when I think about that for a year out or two years out today's best models, frontier models are probably
speaker-1: You're understanding, yeah.
speaker-0: already there in terms of being able to support everyday workflows. And if that can be so efficient to be put on the device, this is why I'm actually very bullish. And I was mentioning this to you earlier. I'm very bullish on devices because I just think we will be able to compact these into smaller language models ⁓ and be able to do very specific things that's much more economical. then also just like we don't have the energy infrastructure. to shoot everything out to these frontier models anyway. But anyway, small language models. I like the way you put that large language models, like big problems to solve, small language problems or small problems on small language models. repeatable. Okay, well, last question I have for you. So you're B2B marketer, you're sort of building your AI roadmap, any guidance you have for them.
speaker-1: anything about So make devices, I gotta be biased, make devices part of your strategy conversation with your IT departments, know, help them understand why you need that device or devices across your team. It's not just, you know, the perfect app or the perfect software you need. It's actually the machine to help you power it and be successful. And if you don't have that capability at the edge, you're not going to be successful. So advocate for yourself, I guess. And we've done some great work on how do we help position that across different roles of how a copilot plus PC can be used. So there's a lot of great information out there. Use copilot to help you build a justification on why copilot plus PC is needed for your team and your role. Build that repeatable set of workflows. I think we've really touched upon this one a lot, but pick three workflows. start looking at what are those touch points throughout the entire week that are repeatable, start to think about, know, where can AI bring me efficiencies and get your team buy-in on that, bring them in, help them also give input on where do they feel like their workflow is something that is repetitive and that we can look at. Because when you bring them in and they're part of the process of changing something from a task they do every day into a workflow that's automated through AI, you'll have better buy-in. There's less of that scared of, my God, my job, is it going to disappear? No, what you're doing is you're empowering them to be innovative. And now they're learning why it's important to move that reputable task over. And they're learning how to do it too. So they're totally bought in, right? ⁓ And then prepare for how co-work can come to life ⁓ across. because most businesses and I was actually on a podcast, I was listening to a podcast. I wasn't speaking on that one. I was listening and a lot of the chat in the podcast, was around like, but my business doesn't allow that or I don't, I can't get that app or I can't get, it's not secure enough. Use Call Pilot. It's part of the M365 system. Most businesses, large corporations give you that ability. So start using the tools that are actually available to you instead of trying to go and build justification as why you need something else. Figure out what the things are that are available to you first and try to try to explore the how far you can go with those tools first. So, ⁓ co-work is opening up the door for marketers and I'm really excited to see what it can bring and how how each team has adopted it differently. So it's going to be it's going to be exciting.
speaker-0: Yeah, well said. mean, I think we have access to a lot of these tools. It's a matter of are we making time to learn them, experiment with them, right? And you don't get to the big innovative gains unless you hands-on exploring, you're learning. It's all iterative until you see the patterns and then you can really create these light bulb moments for you and your team. So, ⁓ but Janette, just want to say thank you so much for joining us on the pod. This was great. This was a lot of fun. It was super fun. And thank you for joining us today.
speaker-1: and done.
speaker-1: I'm great. I'm so excited to be here today. We're finally talking about something where markers don't usually put into the strategy document, which is the device and how important it is for their every day, right? ⁓ At Microsoft, really Windows devices are AI capable. So yes, you're using the word AI PC today, Andrew. I will allow for it. We don't use that term because really we feel like all of our, well, Not that we feel like we know all of our devices are AI capable. It depends on what type of device you have to really elevate that AI capability. So Co-Pilot plus PCs now is really that like ultimate experience and unlocks a whole new world of what AI can be. ⁓ And in full transparency, that term AI PC just oversimplifies it, but I'll embrace it today. We'll talk just for you.
speaker-0: Just for today, we'll make an exception. But I think you're actually hitting on a key point, which is people are referring to you in all different ways, right? AI PCs, AI powered PCs, AI ready PCs, there's Co-pilot plus PC. So maybe can you kind of break that down for us of, you mentioned AI PCs kind of undersell it. Where does the Co-pilot plus PC come in and what is the difference? between the two.
speaker-1: Yeah, so think of AIPC as a laptop with a third brain. ⁓
speaker-0: Okay, I like that.
speaker-1: So you got the CPU, you got the GPU, and now you have the NPU, right? And the NPU comes from all of our silicon partners. So all of them have it, Intel, Qualcomm, and AMD. And it's designed to run AI efficiently on your device. So what it's doing is it's taking some of those heavy workloads off of your GPU and your CPU and bringing it onto your NPU and dedicating that space to AI functionality. Traditionally, when you asked AI PC something, it would send data up to the cloud, it would wait for a response. Well, we have this thing called Copilot Plus PC now, which will actually run AI locally on your device. So it's faster, more efficient. It has a lot of privacy things built into keeping data, sensitive data safe. That's key. It is. And for marketers, you may not think about that. But then if you think about the work you are doing and product launches and some of that sensitivity you're dealing with, you want to make sure that your data and what you're dealing with is protected. also with productivity, we want that instance. We need that moment's response. And waiting for a server to do a round trip is like, sometimes it's like, wow, why is that taking so long? Co-Py Plus PC brings a whole new experience, live translations, AI assistant generated creative, is huge, camera effects. It really becomes your AI experience on the device. And it shows up at the speed of your thoughts with the right device that you have. You got to make sure you got a Co-Py Plus PC.
speaker-0: No, I honestly, think you're just in that sort of explanation. You're just showing like, you're right, AI PCs does undersell it because there's so much you can do now. I think that you're to your point on the privacy and security, like I think that's key because that, mean, I think the devices as they become more capable, they become sort of like a, you know, a bit of a threat vector, like people are going to, you know, they'll look at them as like a key endpoint to get at. But You know, the other thing that comes to mind is that, you know, the cost of to run these AI models, I think is going to go up. that's my perspective. think, you know, you've got like Sam Altman that recently was saying how, look, we sell ⁓ intelligence on demand. And so, I think that we will see companies that the major, that the frontier companies move from selling subscriptions to selling solutions and potentially Staffog, et cetera. So it might not be, hey, it's $20 a month, but $20,000 a month for a set of agent capabilities. And so I think that having this run on the device is probably going to be the most efficient option. I actually see it as, I think this will be the default option. Run it on device. can still push through the cloud to get the crazy frontier model capabilities.
speaker-1: for like
speaker-0: everyday capabilities, I don't know if you need that. Like I think you can process this stuff on the device.
speaker-1: We kind of break it down into two is like run it locally when it's speed critical. Like you need something that will start really fast. If it's sensitive, you want to run it on the device because you you want to keep that confidentiality. And if you need more stability now may not relate for a marketer necessarily, but somebody in the field, we all often have use cases around like insurance people that go to natural disaster area. they need it to run locally because there's probably no internet, let alone electricity in that area. Right. So that's a real use case that we often talk about is like people want to use AI in their workflows in their everyday. Well, if they don't have internet and electricity, how do we enable that and have AI on the device really does enable that person to be able to still function as if they were with electricity and Wi-Fi.
speaker-0: It's a point, because when you're in the field or these kind of edge use cases, you don't have connectivity, you still need that intelligence on demand. Like you get used to working, having access to AI tools, and then it's like maybe you're on a plane and Wi-Fi is down. Your productivity is severely impacted. Yeah, it has happened to me multiple times. And it's like you just almost...
speaker-1: Has that happened?
speaker-0: get a little bit frustrated with yourself because when you're so used to treating AI as a thought partner, it's like when it's taken away, you definitely feel the impact. You made a comment around, again, people thinking about the AI strategy, not necessarily thinking about devices or their copilot plus VCL, I'll say that. So when you think about the on-device AI experience, what is the big unlock that we think people might be missing. You mentioned, by the way, the translations and a bunch of other capabilities. Maybe we can go back over some of those. What's the big unlock here that we're missing?
speaker-1: lot of people think it's all about the apps, right? When they think about unlocking and really with some things we launched recently, which we'll probably talk about today as well. It's about what compute is happening. Like what are you trying to function? What is the workflow you're trying to bring to life? So it's not always just about the apps and a lot of people think, ⁓ I got to create the latest and greatest big app to really make impact. Well, no, take it back a second and let's think about the workflows that you run. and start shifting some of that into the use of AI. marketing is a high volume content engine, briefs, edits, creative development, analytics, meeting recordings, which I love. ⁓ my God, I cannot live without meeting recordings and the notes from.
speaker-0: That's the expectation. It's like, me the transcript. I need the summary and facilitator two and teams is amazing. Yes. I don't know. that's something. A hundred percent. It's like, Hey, you're halfway through the meeting. You've covered this. then anyway, really big fan of that too.
speaker-1: you on track. So yeah, but we're constantly challenging the word of change and local AI just kind of helps us manage those micro moments faster and smoother. And you know, we're really moving at the space, the pace that we need. And it's less about now the, transactional moments. And it's about context setting and having a conversation and empowering the agent that you're working with. to actually work with you, right? Like you would never work with a person, like say you had an assistant, you would never give them tasks without context. You give some details to what you need them to do. Well, that's the same thing with co-pilot is like garbage in, garbage out. Like you want one response, you put a response in, you're gonna get one response out where if you embrace it and start to work with AI, you know, it's just, it's really powerful, but I think a lot of people just aren't there yet. They're, they're just starting to uncover what that means. And then it's not to be annoyed by all the questions that Copa is asking you understand it's trying to set the context.
speaker-0: I actually encourage any AI model that I'm working with to ask me questions. I actually think that's the best way for it to understand your intention and the goal and the parameters and all that. Cause you made a really great point, which is would you treat basically a person this way of like, do this for me and give it no context, no examples, not of like, Hey, this is what I don't want you to do. And we assume that AI is just this super intelligence that will just get it. And because it's super smart, we just think it knows what we want, but it doesn't. I always say that. And how well do husbands do that? So it's like, you know, we've got to, like I always say is like, let's say you've got a new agent, how are you onboarding them? How are you training them? How are you giving it feedback? ⁓
speaker-1: should be reading our minds, right?
speaker-0: And so I think that we need to think about it in the same way. So you talked a little bit about some announcements and one of the ones hopefully we can get to is ⁓ Copilot Cowork. Yeah. That's a huge one. So for those, and I can just kind of set the context for those who haven't heard about it yet. So I believe this was built in partnership with Anthropic, right? So they've got Claude Cowork. You're able to run multi-step tasks across your M365 suite. Right. So all like your Outlook and Teams, Excel, all the apps that you're in all the time. To me, I think this is a huge breakthrough from telling you stuff to doing stuff for you. But maybe you can give a broad overview and then we can just talk about how big of a leap this way is.
speaker-1: Yeah, it is a huge leap. It's going from transaction of answering questions to now executing and being that assistant for you. So we describe it as turning your your intent now into a plan and progressing with that plan with checkpoints to make sure that you are in control and that the final output is what you have. thought in your mind. Let me go back to reading your mind, right? Because we all have a thought of what we want. We just are trying to figure out how to work with Copilot to get it to that point. Well, this actually does help us now. Cowork will take you, like you said, work through all of the applications too. So it's not just you're doing one prompt in Word to execute that piece of the puzzle. You now are using Cowork to execute a workflow across multiple applications where you actually don't need to tell it. where it needs to go. It will actually try to figure that out based on how best your context is that you're setting and answering those questions. for marketers, it's like, it's, used to do singular things. It was help me draft this launch email for my campaign. Now it's help me run the campaign workflow end to end. Let's plan the prep meeting. Let's do some focus time. Let's assemble those inputs and what are those outputs and those drafts and what are those followups? Like it's a entire workflow that you're trying to build. And you don't have to know how to build agents anymore. You don't need to know the back end of that, right? Some people were really scared. ⁓ my God, coding, like, what am I doing? You don't need to know that. You just need to be able to be very detailed in setting that context, treating it like an intern checklist of like, these are the things I need it to do. And really providing that purpose and the approvals and the steering of those response. questions that you're getting back and forth. And don't get annoyed by all the 20 questions because I know we simply get annoyed by 20 questions all the time. I have a toddler, I know what it means when you get 20 questions. But again, it's to make you more successful and ensure that that workflow is functioning the way that you expect it to function. So it's prompt and response has been helpful and great. you know, my dad has figured out how to use it now. So it's fantastic. It's reached all age demographics. ⁓ And now it's the next evolution. It's the plan and execute. And it's going to be transformational for sure.
speaker-0: I'm so excited for it. actually, so I haven't had the chance to get hands on myself, but I have heard from folks that it is incredible. That it is really, really, really good product. Like, I think this is where AI really becomes useful. Like, obviously the OpenClaw stuff is blowing up right now where you can have agents running on your computer. It's really the same concept except for, I mean, let's face it, as fun as OpenClaw is and it is a transformational technology. It is just not secure, right? Like we just can't run any client workflows, any of our workflows on OpenClaw. Like, okay, it's a personal project, even still, you just assume that, you know, let's just say it's breached. That's kind of how you need to assume with OpenClaw, but cowork is completely different because it's obviously connected into your Microsoft tenant. But one of the things I was going to say is with cowork, think, sorry, with AI in general, There's been a lot of studies that have showed like the HPR, the Harvard Business Review study. There's one that looked at a US tech company and I think they tracked it over eight months. And they basically said AI allowed people to work faster, take on more tasks, but it was sort of extending work into personal hours. it actually, what they said is it intensifies work. Now, I think that with cowork, it can actually do stuff for you. I think that gets us to this, we get to live a more balanced life potentially, because it's not just prompt and response and then you got to take it and move it over here. Like it's actually doing some work for you. So that's why I'm so excited about it. And for the folks that are like, well, I don't know how it works. It's kind of like, do you know how your car works? Not really. I like you probably get the fundamentals, but you, if you probably don't know how your car works, but do you drive it every single day? Do you your kids in it? Do you like, do you trust it? Yeah. So I don't know. That's kind of a parallel. Okay. Marketers, look at, ⁓ copilot plus PCs. What are the top use cases today that, that, that they can be doing? Like, how can it change their work today? What's, what's your thoughts?
speaker-1: Yeah, so creative creation and you know, we use the word creative. Everyone thinks like visuals, but think content as well and messaging. you know, faster ways to do first drafts. So maybe you need a first draft before going to an agency to get buy-in internally on an idea, right? You don't need to be that creative expert. You can bring somewhat of the idea to life using AI and then using different variations, prompting to get it to where you need it to be, know, listening to the CEO or the LT, bring feedback in a meeting and then take that feedback quickly and adjust and then you can get it to a point where you get that approval. So, ⁓ you know, we're all about in the marketing world testing and quickly adapting and AI can do that. It could be a couple words, it can be the whole creative ⁓ and you can do that more in-house to be able to have those abilities to quickly adapt. You may not always have budget to do things externally with agencies. You might need to do some of those small things in-house. When it comes to meetings, we mentioned earlier how we rely on it, but turning calls into those briefs. I've done that before because I'm very good about setting context verbally and talking about something and getting excited about and feeding off the room. then I'm like, damn, what did I just say? I got to put that in a brief now. Well, recording that meeting, takes all that and then you can adjust it using prompting to get it into a brief format and not lose that momentum of that fantastic meeting you just had and moving forward. Video, we got it today. Better camera, better captioning, better accessibility. And for me in a global team role, when I go and present to Asia as an example, they have. the captioning on, they are making sure, yeah, they're making sure they can understand what I'm landing. And ⁓ I don't have to necessarily worry about it. They're using the technology so that it's adaptable for them and they understand what I'm trying to convey and land. And I think things are landing a whole lot better now, which is fantastic. ⁓ And then that sensitive work, we have product launches all the time and we talk about local. AI experience and having our launch plans and our strategies and being able to bring it onto your device ⁓ and just function a little bit differently without, you know, the multiple share points with multiple access points and that stuff would walk down. So it's just, and this is faster without sacrificing quality really. They're having that copilot plus PC and really everything I just talked about to some degree. You can use that across any AI PC, but we're at CoPilot plus PC. There is an elevated experience across all four of those types of use cases. And we built a great tool so you can help understand what a CoPilot plus PC is because the... It is tough sometimes to really know what are those additional experiences that you can light up on your device. So we're doing some great work at try to bring that more...
speaker-0: Differences, yep.
speaker-1: more into the use case scenario across multiple roles within a business.
speaker-0: Yeah, not to oversimplify, but I think that if, as we are trying to get more richer and more powerful on the Visee experiences, you're going to need more power. You're going to be, and again, I don't want to get into the whole tops thing, but you're going to need more performance. And I think ultimately that is the thing that Copilot plus PC is unlocking, right? Is allowing you to be able to do a lot of these. Exactly. Your third, exactly your third brain.
speaker-1: third brain. ⁓
speaker-0: And you know, the translations, it's funny because some people are like, well, know, AI power translations are not quite there yet. It's like, look, this is the worst it's ever going to be. And it's already pretty, pretty damn good. It's really, it's again, month over month, we're seeing improvements, right? Like drastic improvements. So that's the other thing too, is like what you may have experienced, let's say three months ago, four months ago is not the same, right? ⁓
speaker-1: It's getting faster too. we, our timelines are going shorter and shorter on being able to deliver localized assets to our fields because the workflow of the localization process now has AI within different touch points of that process to speed things up. So we've seen remarkable change, quality timelines and different efficiencies all around.
speaker-0: And now, like in your example of taking the transcript and creating the brief, now even using co-work to update that in the exact template that you want, update that file, do it locally on the PC. think it's just going to be ⁓ a game changer. for us to intercept ⁓ OS, it's an AI-powered platform that we use to run a lot of the agency solutions. But the vision is, and talking to our CTO, Francis, about How do you put this powerful capability into the hands of smart people, right? ⁓ Of the team. How you distribute this though, what we're thinking about is how do you run it locally on device? To be able to have the cloud access in some cases, but a lot of the everyday stuff, being able to run it locally. And I think that's, you I know that you mentioned Cyverns, what's the app? What's the killer app, right? I mean, it's not only about apps, but go build it. You know, like it's never been easier to actually do that. So that's the other use case that we're seeing is, is we build these, this operating system. How do we actually install that locally on the device? And this is where Copilot plus PCs has really been like a key consideration for us. but yeah, so lots of, lots of, lots of use cases. think, ⁓ the other thing I was picking up on is again, you went back to sort of like the questions and making sure that the, you know, the model understands what you want to do. I think a lot of marketers are not really thinking about AI as a thought partner. Like it's, you know, it's that 10 % stuff of like, okay, maybe write my email or whatever, but AI has been trained on the equivalent of like 200 million books worth of data and it can recall it 100 % in seconds. So get it to spar with you, get it to ask you the questions and like even clarify your own thoughts, like you said, it's easier to sometimes talk it out. So even turning on voice mode and being able to just chat it out. Do you ever do that with yours? just curious. is it just me? have. Okay. Yeah.
speaker-1: No, it's a nice little companion to have some time to time and throw ideas at it. But I have done that where you do the talk. It's great because it actually does transcribe it as well. It's not like you lose, you're just making it easier for yourself because you're actually just talking out loud versus having to type it all right.
speaker-0: Yeah, yeah, 100%. So lots of use cases that we can run on device. What does that mean for cloud? mean, how does a marketer think about what do I run on the device versus what I run in the cloud? Is there a framework or how should people think about that? Because now everything just run locally? I don't know. I think there's also a place for cloud too. What you think?
speaker-1: Yeah. So we talked a little bit of the locally already. It's like if you're doing stuff that's speed critical, you want to do that locally sensitive stability. Like if you are doing repeated tasks, you want that stability so that the job can get done. For running on the cloud, think of supersized and you need huge reasoning or big generation, heavy multi-modeled ⁓ execution. ⁓ Also, cloud is still really important for sharing and being able to share what you're doing with your team and using data that is part of your organization. So there is a connectivity you need to bring in some of that intelligence of that data because you probably don't have it locally on your device. And if you do, I hope it's protected because you need to make sure that data is not just out there on any device that. could get stolen, and you never know where that data goes, right? ⁓ But then if you need massive compute bursts, you don't really want that on your device. It's going to slow things down. You need the cloud to help you with that. So there is some balance to it, but you'll figure it out. It's really just managing it in your day to day to see what you can bring locally onto your device. ⁓ and playing around with it. So, reserving the cloud really for those heavier lifts. So, local is for speed and sensitivity, cloud is for like scale and complexity is my little sound bite for you today.
speaker-0: I actually, I'm going back to the whole car analogy and just thinking about it. It's a bit like a hybrid car, right? Like if you want more power, then that's where gas kicks in. Maybe that's the equivalent of, I don't know. And then it's like the, you know, the electric version is if you're going slow speeds and stop and go traffic, that might be on device. Yeah, stability, right? ⁓ But you know, and it's funny because go back to the whole like security and privacy thing. I mean, I know that the AI providers say that they don't trade on your data and I do trust that mostly. think that you do see articles that come up like in routers that talked about, I think it was Meta where they were training their image recognition model and they had like a couple hundred people from like Y Pro. I think it was a team in India that was actually reviewing all of your pictures to train the model. So it's kind of like, whoa, whoa, whoa, whoa. There is a, I guess anything really that you put in the cloud. mean, obviously there's security protections and all that, but there still is, you have to, it's not as safe as obviously having it your device. And I think that that just security and privacy is just going to be a really important thing. So.
speaker-1: Yeah, because think of two things, device loss, like if you lose the device and data sits on that. But then security foundations is the starting of your AI journey. So if you back it up a little bit, we consult a lot of businesses on building up their AI frontiers, the way that we like to use these days. The first thing they do is to check about security and data compliance. And if you have a garbage foundation, you're not going to be able to build upon it. our consultants are constantly working with businesses to really make sure your data infrastructure set up, or making sure your security foundational things are set up before you even adopt what AI means and go into this wonderful world of AI. AI doesn't change the need for good security hygiene. It makes it more important.
speaker-0: Yes. So that's actually a really good point. So in terms of security considerations that we need to be thinking of, you mentioned sort of hygiene, infrastructure, making sure that we're set up to do this securely. And any other considerations we should just think about as we think about on-device AI? Like, how do we just make sure it's protected?
speaker-1: Well, hopefully your business has implemented some of those things of like Windows Hello, right, that you can't just walk up, anyone can walk up to your device and plug in a password, like you want that visual identity, right? We have a great video coming out about that, by the way, it's fantastic. ⁓ And then really about data, what you can store on your device. And there's some documentation you may not be able to store on your device, and it needs to be locked up in a SharePoint in the cloud. it's managing the definitions of that documentation. You know, what's highly confidential, confidential, general, public. And we use data labeling quite religiously across all of our content because like even if I wanted to go change a document security label, I have to get justification as to why I'm changing it. And in most cases, it's because I identified it wrongly from the beginning. So it's like oops, mistake, but they do want justification as to why things are changing. But that labeling just brings thoughtfulness to how you're managing that document that you're working with. simple things like that that may not necessarily be a standard in a business practice and in your workflows, and it could save you a lot. So it's the little things here and there that really do matter.
speaker-0: I find that ⁓ talking to a lot of IT folks is that they have a lot of security capabilities that they have access to that they just don't use or they haven't turned on for some reason. And it's like, it baffles me because it's like you've, you've kind of already paid for it. So why aren't you, why aren't you turning them on?
speaker-1: It's the training, honestly, being able to train your staff, your teams about how to use data labels as an example, right? Like, what is the right context for each of those settings? How to apply it? How does it restrict your usage or sharing, right? ⁓ So it is really about setting up that great foundation of training everybody on how security should be top of mind for them as much as it's top of mind for an ITDM.
speaker-0: Yeah, makes sense. So clear message here, which is as much as we all want to get to the, you know, the novel, sexy, fun stuff, you got to make sure that you're ready for it. Right. Like, make sure you're set up. You're not going to blow anything out. ⁓ again, I go back to this concept of a killer app. I always hear this of the next generation of apps are going to take advantage of the the NPU, right? That will be able to run or dedicated to run a lot of the AI processes. But a lot of the apps are not there yet. And so I guess the question comes up of like, invest today versus maybe waiting until we see what some of these killer apps might look like? What's your thought on that? What's your response to that?
speaker-1: Well, AI is here now and the platform shift has already started. So you can fall behind very quickly and marketers who adopt technology early on will win, right? And you don't want to fall behind. And I'm not putting pressure on anyone. Like I feel it sometimes too where I'm like, ⁓ I really wish I knew how to build an agent. I really wish I knew how to, and I actually had this conversation with somebody recently like, ⁓ we want to build something in six months from now. Can we start sculpting it now? And I said, absolutely not, because I don't know what the technology will be in six months from now. Like we, we might be able to build something totally different because we don't know what that technology is. Or, you know, cowork as an example just kind of came, well, I'm sure there was a long road to here, but we only learned about it recently, which changes potentially the need for agents and how you might be able to use coworker. for those workflows. So now my mind has shifted to do I really need to learn how to build an agent or do I really need to learn how to manage my workflows within cowork, right? So when, these capabilities start to standardize themselves as in like what's old versus what's new, know, the muscles have to change. You have to adapt quickly on what is current, what is most useful to you. experiment, play around with it. And I think that's where people get a little scared. It's like, I don't know how to experiment. I don't know how to play around with it. Like, am I gonna break something? And I think often we're in the waiting game and you can't anymore. Like you gotta try, try. I say it to my three year old all the time. You gotta try. You gotta explore. you know, commonly we hear don't have time for it, not sure what I'm doing. ⁓ You gotta make time for it. You gotta figure out, carve out time in your calendar, just kind of play with it. What I do now is if I need to start something new, I go to Co-Pilot first and I say whatever I need to say. They may not get it right or they may not wanna do it, might do it the right way that I wanna do it, but at least I tried it and I have a starting point. And maybe I figured out something new I could do with Copilot while I experimented on this other task. just always going AI first with anything I'm doing has opened some doors of exploration.
speaker-0: Yeah, I mean, there's a few really good points here. think one in which is you can't afford to wait anymore because things are moving so damn fast. But ⁓ also, you could always ⁓ make a case of why not do it, right? Like there's always a reason why to wait because it reminds me of the Matt Dixon research that was what was this book called? Jolt Effect, where he talked about I think it was like up to 60 % of qualified pipeline, you will lose not to a competitor but to indecision. Because he basically says it is safer to make no decision than to make the wrong decision. And so it's this concept of he's like, it's not fear of missing out, it's fear of messing up. So this not FOMO, it's FOMO type of thing. But it's true though, I think that people are generally unsure. And so it's safer to just, you know, not act, but I think that you will get left behind, right? Like it's not meant to instill fear, but if you want to be, you know, relevant, even just in your own personal career, you should be advocating to have access to copilot plus PCs and like the on-device AI, because ⁓ things are moving so quickly and that you want to stay relevant, right? So. And again, the laptops as well, you're not buying them every year as an organization, you're buying them in a cycle. So if your cycle is now, I don't know why you wouldn't.
speaker-1: Yeah, you're buying for the next five years and imagine we're going to be with IEI in five years. Like you need a device that can manage that.
speaker-0: Right and so otherwise you're just kind of buying it twice like you're buying something that can be fine for now Then you're like, ⁓ my god, I need to buy something else ⁓ Now I wanted to also go back to your point on apps or agents and whether you know to know how to build the agents and and things like that ⁓ If marketers are thinking about in that position of do I build my own app instead of waiting for the whole the killer app that's coming What's your guidance on this? Is it like hey? look at what coworker is doing is like that's kind of an orchestrator or your chief of staff in a way. It's an agent and you can teach it to what to do or how can they build their own stuff? What's your perspective of that?
speaker-1: It can be daunting because you're like, where do I start? ⁓ So my recommendation is looking at your workflows, looking at what you do today. Let's not create something that's supporting something you want to do in the future. Let's look at things you do today, end to end. Pick a pain point. Is that briefing? Is that the content strategy? Is that an event follow up? Like pick a moment and start with that workflow. Don't think about the tool. Don't think about what the out. where the final thing is that can fix your problem. Think about the problem itself first. Describe what that problem is in great details. Defining what is those inputs that you need to make that workflow successful? What are those standard prompts and templates that you use to support that workflow? ⁓ What is that automation you're potentially wanting to get out of it? And then what is that impact that you want to see this workflow? potentially take away from the rest of your team. Like maybe it's hours, maybe it's time, maybe it's just better detail. And managing that workflow, defining it, and then trying to figure out first and foremost, how can Co-Pod, how can coworker help you with that? So experimenting with that one workflow. Let's try not to dump everything into it. Because that can be overwhelming. You want to perfect one thing. And then staying with that workflow to build upon that and making sure it's doing what you want. Don't be like, ⁓ we get so excited that you got one thing kind of functioning using AI and then start to go off to the next thing and trying to figure out the next one. Perfect that one. Make it perfect. Make it your use case internally, right? To really showcase the value of AI because people can poke holes easily and you want to try to show the best in class. is how we're using AI in our org, our marketing org. ⁓ and then get that buy-in and support and trust amongst your organization to then go and build more things. Because sometimes it might cost some money to build some of these AI tools, depending on what you're trying to accomplish, where maybe your first couple, you can get away with doing them amongst your team and your organization. But you're building that trust that you guys, as a marketing organization, trust AI, use AI, are experts in AI.
speaker-0: I think it's a great point on really understanding what the problem is you're trying to solve. Like you actually said, like, don't even think about AI, like actually abstract it from the thought process. What is the problem you're trying to solve? And I think that that is where you're going to drive impact. feel that AI in some ways can be a bit of a checkbox in that everyone's like, well, I got to do something AI. You know? And so I think it leads to kind of like AI theater. a little bit of like, ⁓ yeah, we're doing this AI thing here versus like actually trying to drive impact. And I think the other point that you made around, you can probably do some like onesie twosie things on your own, but as you think about scale, ⁓ you need a framework. And so I know, again, Francis on the technology side, you know, he's a huge GitHub fan. And so working very closely with the GitHub team in building an environment, that they've developed that allows us, who are the non-technical folks, to build application solutions. Like we can basically vibe code different things as a line of business. And we do it within the environment, within the GitHub environment so that it's ⁓ safe and that, again, you're not going to create any security misconfigurations or some back doors that basically you get out. Exactly, right? So I think that you too need a framework because the vibe coding stuff is very empowering. But if you don't govern it, it can also be dangerous, you know? Yep. So I agree. You need like proper frameworks, know, and foundations to get it right.
speaker-1: And we've started an intake process because everyone's been wanting to build their own things, their own agents, their own stuff, make it easy for people to use it internally. Then people are wanting to make things now stuff for our partners to use. it's like, okay, how much stuff do we have? Are we having now? Like it's like, so we have an intake process now where these types of things are, are, the questions they're asking so that we are making the right tools. We are, replicating the right workflows, or it's like, I have a great idea of what I want to do, my coworker has a great idea. There's actually tools or processes or workflows that we could probably bring these two things together. And we just don't, we're just not talking and bringing those ideas together. We're an intake process and a team that evaluates what are those AI tools or workflows that we're going to empower our internal teams or our partners to use. they have a larger vision on that and can it help us make those decisions and start to bring things together where it makes sense.
speaker-0: Yeah, yeah, it makes sense. And we have a similar intake process. That way you are not building the same things. And that like, right, it happens. But then also it's like people have these really great ideas that when you combine them together, it's actually one main thing, you're saying, right? So that's actually a good segue, because one of the things I was going to ask is that, is you think about where AI PCs go and it becomes, I was reading some IDC research that I think it's by 2028.
speaker-1: It happens.
speaker-0: 60 % of total PC shipments will basically be, you know, AI-ready in some capacity. What do marketers need to do? Like if it's coming anyway, and you'll likely get access to it, ⁓ how do we just be well positioned to actually take advantage of this stuff? You mentioned security before getting infrastructure ready. Like what other things can we do so that we're ready for the moment? Because I feel like it's going fast.
speaker-1: It's already here. Well, at least on how we define AI PC, it's here. ⁓ Everybody can get an AI capable device. It's a question of like, do you have a copilot plus PC? Which yes, that roadmap is looking like more and more all of our devices will be copilot plus PCs as all of our silicon partners catch up with the components. But yeah, you do need to be ready as an individual, but also an organization. ⁓
speaker-0: Yeah, it's here.
speaker-1: It's, think about operating system and like talking about devices. So we think about our device, but think about your operating system as like your day to day. So what are your templates? What did you, what is the module messaging that you always use? What is the reusable building blocks that you have? And really bringing that all together as your template and your foundation because AI works and is super powerful on. repetition and scaling. So if there's some things that you can standardize, absolutely do it. We don't want to limit differentiation and creative, but there is going to be an ability that you can work a whole lot faster if some things are just set as the core. ⁓ And then what are those guardrails? And I've done some great work with your team too is like,
speaker-0: Yes.
speaker-1: What's allowed? What's sensitive? What must stay local? What can go to the cloud? Defining the dos and the don'ts of what your AI engine will do for you is going to be super, super important. And then training, thinking about those scenarios. What are those workflows? What is it the campaign planning? Is it the event execution? ⁓ my god, that takes a whole engine to do. That is crazy. Is it the partner co-marketing? Like I'm already trying to think about some of the things I can do to really enable our partners because our partners, there is a vast range of different capabilities when it comes to AI. I have some partner marketers who are just figuring out what co-pilot is and what prompting is. So I'm creating simple prompt guides to help them. Right. And then we got other ones that
speaker-0: meeting is where they're at, right?
speaker-1: Yeah, they have access to co-work and great. Now we can give them the building block assets versus giving them fully baked campaigns because they have the workflow abilities to use AI to build the assets that they need. ⁓ And then also thinking about not just all the pre as a marketer, how to use AI for the posts, the stuff that not all of us love, which is the reporting and the metrics and the data and. Right. Using AI to help you. So building out all those different scenarios of your entire team, of the workflows that they all have, and just writing things down and having that documentation so that when AI comes in and is kind of your everyday moving forward, you have some of these foundational things as an understanding of across your team, across the leadership teams, across the ITDMs, so that you are set up to kind of start fast.
speaker-0: You know, it's, it's, ⁓ what comes to mind is that it's not a device refresh moment. This is actually like a cultural refresh. It's like a department refresh. Like if you really want people thinking AI first, I would say use this as a moment to do that. Right. It's like, look, we got device refresh here, the new things that we're going to do. Here's the intake process of how we're going to think about building new, new agents or Gentoo capabilities. Right. I think it's a really, it's a good opportunity for us to like, you know, like create a, yeah, to create, to create that, that, that reset because there's just new capabilities here that we we need to take advantage of.
speaker-1: in the go, like always trying to get the next campaign out or get the next creative idea out. Yes. There needs to be a moment in time if you're going to adopt AI that you do that reset and you properly onboard what AI means to your team. It's not just plopping it in. Yeah. It's thinking about it. You can't just all with it, right?
speaker-0: And I think the other thing we're seeing too is like really intentional how you message it to your team. Because I think that, you you talk about meeting partners where they're at. I think it's also meeting team members where they're at too. Like where there's still a lot of this whole, ⁓ there's still fear. There's this human reaction to be scared of change. And so it's like, ⁓ well, what does that mean if let's say I'm a partner? Partner marketing manager. And so if partners can do a lot of themselves, what does it mean for me and all this kind of stuff. I think like messaging that I think is really important too. ⁓ I wanted to ask you, I mean, as we start to like think ahead, like where stuff goes. And so you've got, you know, the large language models that are advancing all the time. And then you've got the small language models. Where do you see this kind of playing out over the next couple of years? Where do you see sort of the trajectory of these things?
speaker-1: It's really getting intelligent around what is the best model for what you're trying to do. Getting the right model for the job at hand. So if you're not familiar, large models are really for those broad reasoning, complex, ⁓ multi-layer executions that you're trying to do. Small models are really good for like focused task, repeatable, scalable, potentially could run locally depending on how. Your device is set up, right? So the future for marketing is those small models really power your everyday workflows. What are those things that you have to do every day? May sit more in the reporting mechanisms, More sit in those repeatable functions that your team is doing. Where those bigger models, deep understanding, heavy creative lifts, that's probably where a lot of those larger models will sit. big models, big questions, small models. content or constant momentum and change and adjusting. So I like that. Yeah, it's really making it fit for the job that you're trying to accomplish. So I think we go back to like the question we asked earlier is like, how do you set yourself up for success? What are the things again documenting your workflow is documenting what your team does. So then you can find the right models that fit for each of those different functions and workflows that you're trying to accomplish.
speaker-0: You know, that's actually something else I want to touch on is the, is the, you know, the SOPs, the standard operating procedures. A lot of people want to automate stuff. And one of the first things that we, we, we look at is like, well, what are the SOPs? Cause AI is really good about, you know, following a set of instructions and workflows. But if you don't have a sense of how it works today, if you don't have it documented, we know it's going to change, right? With AI in the loop, it's going to change. But if you don't have that documented, you don't really have a good start point. So exactly. and, to the thought on the, the small language models, like when I see that basically the models every year they get about, they drop about 10 times the cost, right? So it's like, it's 10 X cheaper every single year. And so when I think about that for a year out or two years out today's best models, frontier models are probably
speaker-1: You're understanding, yeah.
speaker-0: already there in terms of being able to support everyday workflows. And if that can be so efficient to be put on the device, this is why I'm actually very bullish. And I was mentioning this to you earlier. I'm very bullish on devices because I just think we will be able to compact these into smaller language models ⁓ and be able to do very specific things that's much more economical. then also just like we don't have the energy infrastructure. to shoot everything out to these frontier models anyway. But anyway, small language models. I like the way you put that large language models, like big problems to solve, small language problems or small problems on small language models. repeatable. Okay, well, last question I have for you. So you're B2B marketer, you're sort of building your AI roadmap, any guidance you have for them.
speaker-1: anything about So make devices, I gotta be biased, make devices part of your strategy conversation with your IT departments, know, help them understand why you need that device or devices across your team. It's not just, you know, the perfect app or the perfect software you need. It's actually the machine to help you power it and be successful. And if you don't have that capability at the edge, you're not going to be successful. So advocate for yourself, I guess. And we've done some great work on how do we help position that across different roles of how a copilot plus PC can be used. So there's a lot of great information out there. Use copilot to help you build a justification on why copilot plus PC is needed for your team and your role. Build that repeatable set of workflows. I think we've really touched upon this one a lot, but pick three workflows. start looking at what are those touch points throughout the entire week that are repeatable, start to think about, know, where can AI bring me efficiencies and get your team buy-in on that, bring them in, help them also give input on where do they feel like their workflow is something that is repetitive and that we can look at. Because when you bring them in and they're part of the process of changing something from a task they do every day into a workflow that's automated through AI, you'll have better buy-in. There's less of that scared of, my God, my job, is it going to disappear? No, what you're doing is you're empowering them to be innovative. And now they're learning why it's important to move that reputable task over. And they're learning how to do it too. So they're totally bought in, right? ⁓ And then prepare for how co-work can come to life ⁓ across. because most businesses and I was actually on a podcast, I was listening to a podcast. I wasn't speaking on that one. I was listening and a lot of the chat in the podcast, was around like, but my business doesn't allow that or I don't, I can't get that app or I can't get, it's not secure enough. Use Call Pilot. It's part of the M365 system. Most businesses, large corporations give you that ability. So start using the tools that are actually available to you instead of trying to go and build justification as why you need something else. Figure out what the things are that are available to you first and try to try to explore the how far you can go with those tools first. So, ⁓ co-work is opening up the door for marketers and I'm really excited to see what it can bring and how how each team has adopted it differently. So it's going to be it's going to be exciting.
speaker-0: Yeah, well said. mean, I think we have access to a lot of these tools. It's a matter of are we making time to learn them, experiment with them, right? And you don't get to the big innovative gains unless you hands-on exploring, you're learning. It's all iterative until you see the patterns and then you can really create these light bulb moments for you and your team. So, ⁓ but Janette, just want to say thank you so much for joining us on the pod. This was great. This was a lot of fun. It was super fun. And thank you for joining us today.
speaker-1: and done.
