How AI Is Changing Daily Life and Enterprise | Utilizing AI Ep. 2

AI will get real when it's able to augment our daily lives through smart devices and personal data. This week on utilizing ai, we're diving into the news about AI assistance, local processing, and the risks when AI gets us wrong. Welcome to utilizing ai, the podcast focused on practical applications of artificial intelligence from the RUM group.

Each episode brings together diverse perspectives to explore news and use cases of the ways in which AI is transforming enterprise IT and the infrastructure it serves. I'm your host, Stephen Foskett, president of Tech Field, a business unit here at the Futurum Group, and I'm joined today by a couple of other folks. Let's meet them.

Hi everybody. My name's Brad Shimmin. I'm VP and practice lead for data Intelligence, analytics and Infrastructure at futurum.

And I'm Olivier Blanchard, research director and practice lead for AI devices at Futurum as well. So each week, uh, we talk about the ways, uh, that AI is, is really getting real in the enterprise, and we're gonna get a little personal this week. Uh, in fact, we're gonna talk about the ways in which all of us are using ai.

This was a big topic at AI Field Day, uh, last week as we talked about with Nick patients on utilizing ai. And it is going to, I think, um, sort of head off some of the concerns people have about AI as well as raise some new ones. So let's kick it off a little bit.

Uh, Olivier, I think people have this vision, uh, generally that AI involves massive data centers somewhere sucking down power building, giant models, slurping up all of human knowledge, all that kind of stuff. But really that's not, uh, where this industry is headed. No, it's not.

Uh, so I mean that it partly is where it's headed, but I think that we have to understand that there are essentially two phases to this, uh, this AI digital transformation that we're going through. One is training the models, right? Which requires a lot of GPUs and a lot of data center resources, uh, in order to be able to just like keep developing the new ones and training them, et cetera.

But the real scale when we achieve it is going to be an inference. An inference is basically when the models, when the assistants and the agents are doing the work that you want them to do, that they've been trained to do. And so you have these two competing sort of, uh, uh, resource rich, uh, models of, of data centers and data centers plus something else.

And there's something else is really the edge. It's intelligent devices, which is my focus. So it's smartphones, it's PCs, it's smart television, smart speakers, smart glasses.

Uh, I'm pointing to my glasses. They're actually dumb glasses. There's no technology in them, but you know what I'm talking about.

Um, and other, other devices that we haven't seen yet. Uh, you know, like even your cars are getting smart. So a lot of this inference, a lot of this processing at the edge, which is, um, much faster and much more secure because the interactions that you have with your device and with the, the, the portion or the layer of the model that's running on your device as opposed to running in the cloud is much closer.

So the interactions are immediate, there's no lag. Uh, and also they're a little bit more secure because the data, your, basically your queries and then the responses are not having to go to travel to a data center somewhere and then travel back to your device. Um, so there's, there's this development of, of AI capabilities through chips and software that is going to come out increasingly out of the cloud, out of the data centers and move to the devices that we're all surrounded by on a daily basis.

And that's going to radically change, first of all, how we allocate resources to ai, but also how we interact with ai. Yeah, I think it's easy to get sucked into this sort of, um, hype around data centers and the fact or the idea that we can't do this without standing up nuclear reactors in Idaho somewhere to, to make this happen. And it does bother me to to see that, because that idea, the idea that Olivia, you just mentioned about this being a huge difference between training and inferencing.

And when we first stumbled upon generative ai, the idea was that, oh, well, we're all gonna be investing in huge GPU farms and it's all gonna be about training. And as we've learned over the last couple of years, it's not about that at all. I mean, it is for a select number of frontier model makers.

And I think it's safe to say that, you know, many of the, uh, innovations that Olivier you're talking about, uh, are built on the shoulders of those frontier models. So they do matter, but it's not that we're all doing it all day. What matters most to most people, whether you're a consumer or in the enterprise, is that you are able to have an, an inferencing experience with a model that you know is secure and reliable and performance.

And, you know, you don't have to do that in the cloud. There, there are some situations, right, where if you, if I'm a small company and I'm trying to secure my data center, am I gonna do a better job of that? Or is Microsoft Azure gonna do a better job of that?

And if you can run everything inside of their VPC, then great, that's gonna be a lot. You're gonna be a lot better off than if you were just trying to do it on-prem yourself. But if you're a large organization, especially one that has to deal with, you know, external controls over security and governance, you know, you're gonna want that control.

And as you just mentioned, certainly if I, if you're a consumer, I, I wanna run this on my device, on my laptop, and I want to know that everything I say to my pet model stays on my machine. That that's critical. Yeah.

And we're already starting to see that. But I, I just wanna like make one more point to you, uh, to your question, Steven. And it's, um, that both the hardware, so the, the chips themselves are becoming much more efficient in, in performing these AI workloads, whether it's training or, or inference, but also the models themselves, the software are becoming more efficient.

And so when we, when we talk about the, the massive expenditures, right? I think it's like over 400 billion this year alone from, uh, Microsoft, Google, uh, Amazon, um, and, uh, the, basically like the, the four cloud players, right? 400 billion to build data centers.

But the reality is that if, if you look five years into the future every year, the models, models that were only, that we could only run in the cloud can now run on devices. And we're seeing this happen year after year after year. And so I, I feel like we're, we're planning for a scale of data center capacity that doesn't really make, that makes sense today, but that doesn't make sense in four years or five years when a lot of these models won't need that many GPUs and that many data center re resources to run, right?

And so it, it's, it, it seems that we're, we're not really having the conversation or we're not addressing the reality of that those efficiency gains year after year, generation over generation, and that we're barreling forward with these massive investments in data centers that we may not need, at least not for, for training for other things, maybe. Um, obviously we're gathering a lot of data, so I think that we have to store a lot of data from users that are interacting with ai. There's, there's a lot of that happening.

Um, but in terms of like building massive data centers to train models, that doesn't make sense in the long term. It just doesn't, well, It, it, it, I think that we needed that to get to where we're going, right? You know, we needed to be able to build those models.

'cause we couldn't have built, you know, uh, llama or Claude or Gemini without that kind of investment. But now that we have it, I think that we're seeing, um, at least in a popular impression is that we're seeing sort of diminishing returns on investment in terms of the quality of the models. Essentially.

The fourth generation models are really, really good. Um, the LLMs are, and, um, and the same is with many of the other models that we're using. I mean, image generation has gotten extremely good.

Uh, video generation is still a little sketchy, but it's getting there, um, you know, and, and, and translation and, um, you know, sort of speech to text and so on, that has just gotten basically good enough that we don't need to worry about it anymore. And as you said, what happens then is that those things get quantized and they are then run on, you know, end user devices. And, and as you said as well, it's important to note that these devices have gotten just exponentially better, and I don't use that word lightly.

Um, you know, if you compare, for example, um, I have a, uh, Nvidia Jetson board over here with a 32 bit cuda from, you know, way back when. Um, if you compare what that board can do with what's in this phone, well, it's not just night and day, it's like a different planet. Um, you know, the, the inferencing in arm chips from Apple, from Google, from, uh, Qualcomm, from many others is getting so much better, so much quicker that we are able to run, uh, locally what we never would've dreamed that we could.

Um, as, as well as some of the other supporting infrastructure, we're seeing a lot more use of large amounts of unified memory. Uh, you know, we, there's a lot of news about the, uh, Nvidia Spark, uh, that just came out, um, which has, um, just a tremendous amount of unified memory, but essentially the architecture is very similar to Apple's system on chip architecture. Um, you know, and, and it's, it's interesting to see the ways that the technology is improving and allowing us to run this stuff locally.

Um, and, and at the same time, to your point, it's not just about the hardware, it's about the software. One of the things that I love is that we've applied many of the lessons of cloud computing to AI in that essentially now we're getting to the point where we don't necessarily know where the model is running. Now, sometimes we do, you know, I installed, you know, Alama on my Mac and I, you know, installed Mac Whisper and I, and I know that those are running locally, but when you look at what companies, especially companies like Google and Apple are doing in terms of making, uh, these models transparent from the device to the cloud, it is really remarkable.

And I, and I wanna dive into that. There's been some rumors this week that Apple is finally, we can say finally, right? Um, rolling out Sure.

The next generation Siri with an LLM behind it, uh, soon. Um, well, I, I put an asterisk after soon. We don't know when, and we don't even know if this is true, but there's been a lot of people suggesting that they're gonna be using a special model built by Google.

Apparently they had a bake off between Google and Anthropic. Uh, they picked the Google solution because Google was able to basically do the things Apple needed to do to build this next generation series. So, um, let's talk about that a little bit.

Olivier, um, uh, those device, those, um, questions. I mean, already when you're using the Apple Intelligence features, you don't know whether it's running on device or in the cloud, but you do know that it's protected and you do know that it's seamless, right? Yeah.

So, um, I think that the most extreme example of this that I've seen, um, was from Samsung earlier in the year when they released the S 25, uh, the Galaxy. Um, and they had a custom chip from Qualcomm. So basically it was the, the Snapdragon, but they, they made one for Galaxy specifically.

And what they did is they created two layers, uh, for their AgTech capabilities. One was essentially sort of the general AI that you would use, which is Android and it's Gemini. Uh, and I'm a power user of Gemini.

So I'm, I'm really thrilled that Gemini might be behind Siri 'cause it means that Siri will finally be good. Um, but, but also they had, um, they, they created this on device inside of the firewall, uh, post encryption security, uh, capability where they can run models and actually do training on the device for hyper-personalized ag agentic experiences. And so this is kind of like a first run, first generation.

I don't have a lot of, like, I don't have high expectations for what it's gonna be able to deliver, but as a concept of creating this ultra secure, ultra hyper-personalized ag agentic capability on device, and then using orchestration, essentially AI on the device to decide, okay, this is stuff that needs to stay on the device, and this is stuff that can go out the Gemini, right? Uh, and, and building that model is sort of, I think the, the, the first step to creating this true multimodal and, um, hybrid AI model that just draws from the cloud when it needs to, and then just stays on device when it can. And the recent discussions I've had with, with Google, um, and they're very, very strong in this.

And they have device ecosystem, they have a software ecosystem. They're in Gmail, they're in, uh, you know, Google Docs. They're everywhere, right?

Um, their, their trajectory, I feel is, is trying to create a space where you don't necessarily know, or if you don't want to, you don't need to know where the inference is happening, if it's in the cloud or, or on device. So long as you get the best features, the best user experience, um, and if you do for security reasons or because you wanna optimize your experience, want to move certain things on device, and it keeps some things in the cloud, they'll give you a sort of a dashboard or ability to do that. Um, so it's really exciting.

Um, and, um, again, as, as a huge fan of Gemini, uh, and having been disappointed in Apple's inability to develop their own, uh, in-house ai, uh, for their own stack, I feel like this time, as they did with OpenAI, I think that this is the right partner for what they're trying to accomplish with their, uh, their device ecosystem. Yeah. And I have to wonder, did did they, were they unable to, or were they simply, you know, unsure if they should, because even though, Steven, you, you mentioned that it's seamless, I found with the first iteration of Apple Intelligence, uh, when they brought in OpenAI that they were very deliberate in separating the two out what was local and what was being sent on to OpenAI.

Well, that, that's, I I liked that. Yeah, go ahead. Yeah, That, that's, uh, interesting.

But yes, they're very deliberate in separating OpenAI from Apple Intelligence, but that's only the OpenAI boundary. The, the, the Apple Intelligence stuff is also AI and is also running seamlessly. Yeah, Agreed.

And okay, think back, uh, just a couple of years, maybe 10 or 15, uh, we, we were all using our swipe ahead swipe typing keyboards on Android phones, and they were getting really good at predicting what the word was supposed to be that I was trying to type out like infrastructure. And they did that through a, a technique that's been around for quite a, some time called Federated learning. And what Federated Learning did was allowed us to basically train a really good model in the cloud without actually having to send on everything I typed on my swiping, swiping head keyboard to do it.

And so it did it securely. There are techniques like that. And, um, I think the other one is differential privacy is another one.

Uh, and we're seeing it starts to evolve now with the smart routing capabilities within a lot of agentic platforms that all focus on this sort of seamless sort of movement to optimize that inferencing so that it's performant secure, and you got the best model for the job at hand. So I love that about the way our market's evolving right now as it's not trying to be, oh, you're all in on the cloud, or you're all in on the edge or device. It's saying, you know, we have the right tools for the right job, let's apply it so everyone's happy.

Yeah. And, and, and that's kind of the point. I think that the, I was very impressed by what Apple announced in 2024 at WW DC for the Apple Intelligence features, simply because what they weren't announcing was a giant do it all copilot, LLM what they were announcing was a suite of features that enhanced the operating system with intelligence, with, with a, with ai and it, and it really is AI as well as integration with OpenAI.

Now, I am not impressed by the fact that Apple completely dropped that ball and apparently internally knew that they were dropping that ball and decided to try to do a rebound, and it's take another year to try to bring it to market. Um, yeah. 'cause they, they really, really missed.

And I think that this is very public, um, you know, egg on their face, timing Is everything. Yeah. Delivering these features that they promised for the I iPhone, um, 17, uh, I don't know, or 16.

Um, but the point is, um, you know, that, that they, they, they didn't deliver what they were promising they did deliver, I think some of the framework, um, you know, and, and, and, and right now, if you're using, whether you're using a Mac or you're using an iPhone, if you do things like, um, there's a summarize, um, text, there's a, you know, key points. And so those are, those are using Apple's own LLMs, and those really do run either locally or in the cloud, depending on whatever iOS decides to do very like what Olivier described from what Samsung is doing. It's just, they're not very good.

And that's a big problem. I, I think for a lot of users. Um, you know, so I find myself actually using, um, you know, Gini and chat GPT for tasks that Apple Intelligence ought to be able to do, simply because Apple Intelligence doesn't do it well.

Um, but I'm hopeful, you know, hope Springs eternal that they'll actually deliver that in, in, not in iOS 18, but in iOS, uh, 26, which is a big chunk, Right? Or is it, is it 27? I dunno.

Yeah. Will it be 27? And the other thing that I, that I was interested in is, um, so Apple announced this private cloud compute concept at the time in 2024, which is a really cool idea, which was that they were gonna have their own ARM-based servers and they would be able to burst your data into a secure enclave running on this ARM-based server.

And, um, it, it would keep it secure to you. It would be just the same as if it was your machine for that couple of instance that it takes to run the, spin up the model with your data and run them run it. But then I noticed actually just a couple of weeks ago, apple announced that they shipped the first, um, homegrown server for, uh, Apple's private cloud compute, which makes me say, wait a second, I thought you did that last year.

Um, so maybe, maybe they didn't, I dunno. Or maybe they're using and delivered. Yeah.

So I don't wanna say I'm losing my faith in Apple because I love their, their idea, but, um, the execution has really left a lot to be desired. Yeah. There's not always a very clear timestamp on what they do.

Um, no, but, so, okay. I'm, I'm an Android user, uh, in a sea of Apple users. Everyone I know has an iPhone except for me, um, in my social circles, green, green Smfs.

Ooh, I'm sorry. I'm sorry. Exactly Right.

Yes. I you're blue, I'm, I'm triggered. Um, thanks for that.

Um, but, but essentially what the, the joke among those of us who are in the Android ecosystem is that whenever Apple, and especially with a, a new iPhone release announced, announces all these, these cool new features, there's always this shock, right? Like, you guys didn't have this yet. Like, we've had this for, you know, three to five years.

And, and I feel like that's kind of like the MO for Apple, right? They used to be a very innovative company, and to some extent they are, their silicon's fantastic, their designs are great, but in terms of features, they really wait until the Android ecosystem, uh, has sort of, you know, tried and tested a feature and it's baked in and it's really easy to replicate or to adapt to the I, uh, to the, to the iOS, uh, ecosystem. And, um, for me it's disappointing because I, I'm old enough to remember when Apple was innovating.

Now they're just kind of following. And so I think that the same, the same kind of internal culture of product and feature development might've been the reason they missed on ai. And they might've thought they had more time than they, they actually did.

So now they're playing catch up the way, Get into iOS and Android fanboy. That is not gonna be the topic of this whole show, uh, or this Series, maybe another show do, another one Said, yes, apple has a fast follower, uh, right fast best follower approach to technology. Yes, Sure.

Well, That's okay. It has served them well because they, yeah, good follower approach. They have, you know, a die hard fan base that still lines up for the orange aluminum, whatever they're shipping at the time.

So it's, it's hard to really fault them for that. And then you can see with some of their innovations, like with iOS and iPad OS and Mac OS 26 lately, that they've been trying to sort of modernize themselves. Yeah.

The same way they did back when we switched from osx, uh, sorry, OS 10, uh, onto, off of next and onto the, the current kernel. Um, but it's, it's not easy to do. If it were, we wouldn't have Windows 11, you know, so I, I can't fault them for that.

Yeah. I really can't. Sure.

So Let's, let's talk, um, I, again, I don't wanna get too nerdy here, but let's talk hardware a little bit because Olivier, you know, we're lucky to have you on here because you know this topic better than almost anyone. Um, the, the emerging, um, platforms for running ai, um, outside the data center. And Apple has done a remarkable job.

Their M five is that which just came out, is absolutely tremendous. It's very, very high performant. They have, you know, an a an integrated neural processing unit that is really first class.

They've got great GPU great CPU course. Um, but Apple is no longer alone in having really high performance low power cores. And you already mentioned, you know, you know, copilot PCs and Snapdragon, um, Broadcom is right there.

Um, we're hearing that ARM is gonna be, or that, um, uh, a MD is developing an ARM processor as well. Um, what's your take on this sort of, um, non Apple, uh, mobile, uh, AI or, or low powered ai? Well, I think, you know, ARM is, uh, is sort of like the, the quiet winner in all of this.

Uh, and, and ARM doesn't make chips, at least as far as I know. They don't make chips yet. They just sell licensing, basically IP and architecture to, uh, to implementers.

And so obviously one of their, their biggest, uh, uh, customers, uh, or implementers is Apple. Uh, another one is Nvidia. So you see where this is going.

Um, the, the ARM architecture is extremely, uh, good for AI applications. The performance per watt characteristics beat X 86. And so companies like Intel and Arm that have been the X 86 companies for decades, Intel and AMD are Yeah.

Yeah. I Made the same mistake five seconds ago. Yeah, Yeah.

Um, have, have, I wouldn't say they've struggled. I think they've done really well, um, in, in keeping up with what, uh, what RM uh, RM based chips are, are performing or doing. But we're definitely seeing a sea change, right?

We're seeing less intel and less a MD in the mix of, uh, in the PC space, for instance. Uh, we're seeing more Windows on arm come forward. So obviously Apple is its own thing.

And there, there are ARM-based processors, uh, but Snapdragon entered the PC space a year and a half ago with, uh, its Snapdragon x, uh, tiers of, of chips that have done extremely well, um, at different price tiers and, and, and have the longest battery life. They have the best performance per wide of all of them. And we've been hearing rumors for the past year, year and a half that Nvidia was also getting into the PC space.

Um, it's happening, it's just, just taking a minute. Yeah. And I think it's more of a software issue than a hardware issue.

There's not an issue with, uh, the processors themselves or the manufacturing the capabilities. It's, it's a software issue they're waiting on, on some of the other pieces of the puzzle to be there before they can launch. But, so now you're going to have where you had Intel and a MD with X 86, sort of like ruling the PC space.

Now you also have Apple, which is arm, you also have Qualcomm with their Snapdragon X Chip, which is arm, and then you also have Nvidia. And I think one of the first examples of this, which you mentioned earlier, was the, uh, Invidia Nvidia, um, uh, spark product, right? Um, which is co-developed by, with Media Tech, by the way.

And it's something that's not super well known when, when you look at a diagram of the processor of basically the board, it's, it's at least 50% media, tech, uh, real estate and Nvidia as well. And we're seeing an a, a growing partnership between media tech, uh, and Nvidia in bringing these, these bits of hardware to market. And I feel like the, the Nvidia PC chip, uh, that will come out probably next year, um, is going to be another example of that.

So DGX Spark is sort of like the, the toe in the water of Nvidia and, uh, more windows on arm and ARM-based chips entering the PC space and potentially moving into other areas as well. And, and I think that that's, uh, what the takeaway for our listeners is really that, um, you know, you hear a lot of about Apple's incredible hardware, and they do make incredible hardware, but they're not alone. And there are many companies making incredible hardware as well.

And there's a lot of really exciting stuff on the, on the horizon that we're, I just can't wait to see them announced and, and add them to this conversation. And then, as you already said, as well on the model side, um, we also have companies, you know, I mean, we've mentioned OpenAI, um, anthropic, uh, Google, uh, of course, you know, Microsoft, meta X, ai. Uh, there are so many companies that are working on these sort of foundational models.

But, um, you know, Google has actually emerged really as an interesting one because so many of us have Google Workspace accounts, and we're able to use Gemini as part of that. I mean, I, I use it not just daily, I use it many times a day. Um, I probably use Gemini more than chat GPT even though I intentionally use chat GPT more than Gemini, um, simply because it's just everywhere.

And, and it looks like we're gonna be getting it, uh, as I said with Apple Siri. But there's another news story that I wanna hit here before we, before we wrap up, and that's sort of, um, what happens, um, if this technology is everywhere, um, is it possible that, uh, the AI models could start entering our political discussions? And, uh, I don't wanna make this too much of a political, uh, but Brad, you brought up this interesting story of Google, uh, removing Gemma from its AI studio after a concern from a senator.

So let's, um, let's think about the implications of that when people are trying to implement ai. Uh, tell us a little bit about that story. Yeah, it's, it's hard, and I think it's another vote in favor of the smaller models, the models that you can control and own.

And ironically, the model that Google removed that we're talking about was talking about its smaller model. So there's a deep irony there. But, um, the point is that, um, yeah, it is, it is the case that if you're investing in AI and you're building solutions on top of it, you are investing in the models and that architecture and the way they work and how they work.

And if the rug can be pulled out from under you on any given day because of some political favor or disfavor, that's a risk that comp many companies would seek to mitigate. And so we're very much seeing, and I think we mentioned it a bit ago, about this ability to do smart routing, uh, within frameworks for agent agentic systems as being an emerging area of interest, that if you can do that and you have sort of a, you, you focus more on the engineering, the context engineering that you do on the prompt engineering, then maybe you can start to remove some, or mitigate some of the risks of being completely dependent upon a given model or a given model family. But I, I think we'll never get away from that.

And I think that that's okay, um, simply because as we've just been talking about, we all have a preference for how we like to work with models in which models we work best with. And companies are no, no different. They're the same as us.

So, you know, you get a nerd in a, in a model family, you get used to it, you build up engineering knowledge about it, you don't wanna jump off of it. So, um, you know, it's, it's one of those areas that companies just need to be, uh, I, I think, aware of and perhaps look to and make investments that me mitigate some of the risk that might be associated with this sort of availability. And, and I would say even more than that, this consistency of the models, I think any of us who have used like OpenAI chat GPT over the last couple of years have lamented sometimes the movement from one model to the next, because we lose something in the translation, if you will.

You know, when we went from four to five for instance, uh, I think a lot of people were very upset because the model was suddenly not very talkative in a friendly manner. And it's, it's, you know, whether you're talking about, uh, an avatar that you chat with every night, or you're doing your rag on your pipeline for, you know, an important business decision, it's no different. It's the same thing.

So it's, it's something to, to think about. Yeah. And for the, uh, no, I agree with all that.

For, for the enterprise specifically, all of the demos I've seen of, um, you know, AI studios, which are essentially kind of like the, the marketplace internal marketplaces and management systems for all of the models that that enterprises are running. I feel like we're already in a place where, um, enterprises are testing different models and running different models for different things. Uh, but I feel like it's just a, it's a diversification play.

If, if nothing else, that additional risk that just got introduced into this market of having the, the rug pulled out from under you, because maybe for political reasons or geopolitical reasons, a model or two have to be pulled or they're no longer available, uh, it's gonna be really important, um, for enterprises to work with their technology partners on the software and the hardware side to sort of diversify their, their AgTech and model portfolios, if you will, to kind of de-risk that, uh, that AI stack that they're trying to build for, uh, for their companies and for their customers. Yep. E exactly.

And, and I think the nice thing is when it comes to hardware, when it comes to platforms, when it comes to models, when it comes to, uh, you know, AI partners, um, we have an embarrassment of riches here. And indeed, you know, the, the, the incredible thing is you can have something like this where something gets basically, it's like a recall in the automotive industry. Uh, oops, uh, definitely shouldn't have done that.

Um, and yet it is not even a speed bump to the industry because essentially there are even, you know, uh, even within the context of, uh, you know, Google and, uh, AI studio and so on, they have, uh, alternatives and we all have alternatives. In fact, maybe we have too many, it's a little confusing sometimes, I would admit to deal with models. I think the, the, the is, uh, and you've, you've heard it here first, but it's, uh, instead of cornucopia, it's a knik cope ai Oh dear.

K Ai. I like That, that, that's a dad pun. Yeah.

There It is. Okay. Dad, whatever you say.

Um, yeah, exactly. And, and I think that that's, uh, gonna be something as well that we'll be talking about here, but we do have to wrap. Um, thank you both for joining me for this episode of utilizing AI podcast from the Futurum Group.

Uh, as we wrap, I wanna get a quick, uh, give you guys a quick moment. Uh, tell us where we can connect with you, where can we continue this conversation? Where can we find your coverage of the AI industry?

Uh, Olivia, Well, obviously, you know, on all of the future and platforms, so look for me there, but also the, the, the best place to find me is usually on X, and that's, uh, OA Blanchard. Uh, super simple. Uh, occasionally you'll see me On tv, uh, very rarely will I be on LinkedIn.

Alright, and I'm, I'm on absolute opposite. I, I, I refuse to go on X and I, I spend most of my time on LinkedIn, so under Brad Shimmin. Uh, but we're both, uh, Olivier and I are on, uh, the future sites, and we, we make, uh, rogue appearances wherever, wherever we're, we're needed, or, uh, asked As rogue as possible.

You'll be making rogue appearances here, uh, as well. Um, and, and, like, like you, uh, Brad, I'm, I'm on LinkedIn a lot. Um, I'm also on the socials, you'll find me on x, uh, AI or sometimes, but, uh, mainly, uh, blue Sky, Mastodon, uh, LinkedIn and so on, as well as of course these, and, uh, every Tuesday, pretty much every Tuesday on the Textron Gang.

So looking forward to that. Uh, we will be back with another episode of utilizing ai, uh, next week. This is gonna be a weekly podcast every Wednesday.

So please do, um, subscribe. You'll find us in your favorite podcast application. Just look for utilizing ai.

Uh, you'll also find these on YouTube. You can subscribe there if you wanna see our pretty faces. Um, we would love to hear from you.

Uh, drop us a line, drop us a rating or review. Uh, this podcast is brought to you, uh, by the analysts and experts at the Futurum Group, uh, where Insights meet ai. Uh, for show notes and more episodes, head over to Text Strong ai, uh, the utilizing AI YouTube channel or the text TV app.

Thanks for listening, and we will catch you next week.