Hello and welcome to the let's Talk. Azure podcast with your host Sam Foote and Alan Armstrong.
If you're new here, we're a pair of Azure and Microsoft 365 focused IT security professionals. It's episode 42 of season five. Alan and I recently finished our season this year with our season finale. Here are a few things that we covered. The best New Tech of 2024 Worst Fail of 2024 Our favorite episodes, high level stats. Our predictions and plans for 2025. We've noticed that a large number of you aren't subscribed. If you do enjoy our podcast, please do consider subscribing. It would mean a lot to us for you to show your support to the show. It's a really great episode. So let's get started. Hey Alan, how are you doing this week?
Hey Sam, not doing too bad, pretty busy. How about you? Yeah, I would say I'm getting pretty excited. Two things I think the holiday season AK for us. Christmas is upon us. We're getting very close. I don't know if that's exciting or scary. I think that's a bit of both that we're already midway through December as we record this and this is our literally our finale of season five, which is I think mental to say considering giving our year and what we've, what we've recorded.
Yeah, it's insane. Yeah, I think, I think you're right. The, the season, the holidays coming closer and closer. It's definitely great and scary that it's so, so soon with any, what, week and a half left of work probably and so much today. There's always that rush. Isn't that, there's that September to Christmas rush. It might be budgets that need to be used in year goals that need to be achieved by customers X, Y and Z. But yeah, it is just.
Yeah, yeah, always busy whilst, whilst navigating change freezes and things like that.
Yeah. Oh, we need to do all these things. Oh but by the way, we, we're on lockdown from, from the middle of December so you best get cracking basically. Yeah, yeah, definitely. I'm starting to get, not out of offices yet. I think that's going to be next week. I think I'm gonna start to get out of offices but I'm getting a lot of like, you know, from the end of the week I'm going to be out. This is, this is basically the plan. Good luck. Bye. Basically. So, so I, yeah, my, my gut is, is that it's just going to generally get busier and then just going to be a cliff of out of offices and. Yeah. People not being about. Around.
Yeah, 100%. Yeah, definitely. Anything. Anything since our last episode that piqued your interest, Alan? Anything that we've. Yeah. That we've seen? Don't think so. Have not really been snowed under in work to be fair. Found some stuff. This is. Well, I just found some new controls in Power Apps. Oh yeah, Power App. At the moment they look a lot nicer, some modernized sort of controls and things like that. What Canvas app controls, you mean? Yeah, nice.
Yeah, make it easy and nice to look at. I've been doing that over the last couple of days, but yeah, let's get my head back round into that. Great fun but also terrifying at the same time. Trying to remember everything that you. I used to do with that. Yeah, you know, it's like another like, like pull the disc out, put the new, you know, the Power apps disc into your mind and then start ingesting it again.
Yeah, that. I mean that, that's it, isn't it? It's kind of scary when you get an ask, you know, something that you've maybe done previously for a long time or been close to and then you have a lot of time away from it and then you get asked to do it again. You're sort of pulling from your real long term memory. It's like you never forget to ride a bike but you're a little bit rusty. It just slowly comes back to you. Right. You know it's in there somewhere. Yeah, yeah, exactly.
Yeah, no, that's. No, that's. That's really good. Yeah, well, it's our. I can't believe it's our last episode of this, of this, of this season. So. Yeah, for, for people that have tuned in since the end of last season, which we have a season every year essentially now. We do. Anyway, yeah, I was gonna say this. This is our first year.
Oh, is it our first year? Okay, fine. Yeah, so yeah, so we have. It's a season per year and then we typically take a chunk of time off, I would say around the holiday season. So at the end of the year. So this will be our last episode until an episode in early January. And yeah, we, we now work on a. It's three per month, I believe. Now we used to do one per week, but it was just a bit too. It was a bit too much basically. So, yeah, we do three per month now. So one news episode a month and then two content topics. So yeah, and that, that keeps it a Bit more sustainable, doesn't it I would say for us because we do have things going on especially as Alan's likes to jet set around the world quite a lot. It can be jet into a hotel room, a conference to a hotel room and back again basically. Yeah. So. So, yeah. So it just, it just means that Alan and I create one piece of content each per calendar month and we also keep up to date with any new technology that's. That's coming out in these finale episodes. We just like to sort of shoot the breeze really. There's no sort of technical chat really and we yeah just talk about how the previous years sort of gone not just in terms of the podcast but other tech as well. So yeah. Should we get started? Alan, what's our first. What's our first thing that we're going to talk about? Oh yeah. Best new tech of 2024. What, what year? What your options here? Alan, what you thinking? Best new tech of 2024.
Difficult. I mean I would probably. Was it. Has it been this year that Gen AI properly launched itself? Was it last year? Probably ignite last year? I suppose that it probably started to kick off but it's this year where we've actually been a chance to use it. Yeah. Because there's a limit. Wasn't there do you think it's been operationalized this year? I would say Generative AI. That's what I feel like that's you know it's been layered onto everything.
Yeah. And it's. It's probably going to say the last four months of like at least co pilot. It's got really good. Really like you said, operationalized. Yeah. In the day to day there's so much now that I do use it for.
Yeah. And over the past few weeks I have been, I've been sort of putting my hand in my pocket and trying others as well. Right. So I bought, bought some credit on OpenAI. I bought X Premium to get access to Grok and the models we now have access to are incredible. Like it is scary I would say. You know for me I think what has been really eye opening around Generative AI is the ability for you to collate and index and search your own content and I think that's what Copilot shows off really, really, really well. Right. Because as good as something like a chat GPT is whichever model you're using. I'm just talking generically that is pre trained web data, isn't it? Right. And I think, I think a lot of what not the generative side of it but if you're looking for information, Google and blog posts have got you covered pretty well today, haven't they? Right. You could, you could ask Generative, you could ask ChatGPT, write me, write me a blog post on X topic. Right. But you could also search for a blog post that's been written about that topic, if that makes sense. Right. So I feel like for knowledge gathering it is still a lot quicker and really accurate. So I'm not discounting it, but I think the real magic comes from when you are looking at your own data. So like your co pilot as an example, looking at all my days, it, knowing all your emails and being able to search for all your emails and being able to pull this random presentation that was shared with you like nine months ago that you'd never have found unless you search for it in Copilot. That, that to me is truly the magic of it and that is the productivity gain, isn't it? You know, from, from that operational side.
Yeah, I think it's that. But also it's the, the meeting in Insights and things like that. Now that's got a lot better.
Yeah, yeah. And kudos to Microsoft is they have applied all of that into the enterprise, haven't they? Yeah, there is a standalone experience which is. Oh, it's great. Right. But like you say, when it's integrated into those services, it is mind blowing. Some, some it doesn't get it right all the time. I'm going to. Let's just not be complete fanboys for a minute, I suppose. But yeah, so I do think that, I do think that it's not, we're not at a stage there where it doesn't need human interaction and intervention, you know, and oversight. Right. I think it still needs somebody to validate what it's creating, you know, and I think Microsoft do a good, a good job of that, of basically saying, you know, a human needs to check this by the way. But you know, if you, if you are a subject matter expert or you've got some relative experience in what it is, you know, generating and talking about, you can pull out the bits that you need or the gaps that it's filling in for you and it's really, it's a real helpful assistant at that point, isn't it? You know.
Yeah, exactly. I think it's like 80%, 90% of the job done with. Yeah, yeah. 20% of refining and correcting kind of thing. I think that's fair.
Yeah, exactly. Yeah. And for me, not having to go to learn and look through Microsoft documents and documentation and getting it to pull from there is. It's been really valuable to me if I want to find something niche. The only thing is I do find if you're looking for sort of technical help with it, the models can be so old that it's trained way back and it's not very useful if that makes sense because when it's pulling information around Defender of a Cloud it's calling it Azure Security center and things like that and you're like this was definitely trained too long ago for what I need basically if that makes sense. So I do think you have to be a bit careful on the age of the data that's there. That's why I tried GROK the other day because their claim is that it's a lot more up to date basically and the jury's out on that at the moment. I don't have a concrete viewpoint of that just yet. So yeah. Any other tech Alan for 2024 that sort of blown your mind?
The not I can't think of anything generally. I mean from a Microsoft perspective, you know exposure managements, that sort of stuff was quite call to come out and obviously security co pilot officially being fully launched I think this. This. This year as well and being more operationalized I'm sure it's probably fair to say for it. Yeah. Didn't it go GA this year? It. Yeah it was still in private over because wasn't it like early this year that it went public or ga? Yeah, I think it was.
I think it was yeah. His private secure or something like that, won't it? Yeah it was that. Yeah. And GSA as well. Oh yeah, yeah GSA was a good one this year.
Yeah I've seen such. Such good feedback on GSA just as a product. Right. We've. Well I say we've used it. You've used it a lot more. Right. But I've seen the output of that and it is very impressive. I don't really know much other tech in that space. It's quite a new technology to me if that makes sense. So I can't benchmark it with its peers but from what I've seen it's pretty damn good. Yeah it's got improvements to have but Pro accesses. Yeah. Just works.
Yeah, yeah exactly. Yeah. I also think one area for me is the new consolidated secopics experience. I think Microsoft's push into these, what we calling it now, two or three portals now, something like that, you know because you got. Oh well I don't know if you Include Entra. But instead of having loads of different portals, they're amalgamating loads of different blades into one place, aren't they?
So yeah, yes, the, the consolidation to into Defender XDR port has been hard, you know, been lots of that. That integration has been a lot this year.
Yeah, definitely, yeah. And I said the new Purview portal, great experience compared to the old compliance portal. The fact that the menu is dynamic on the left hand side. Oh, so much less scrolling. I don't wouldn't say it's any quicker. So a bit of feedback there, but definitely. And I've seen the only downside I think on the data side is that they've sort of merged the structured and unstructured data together and I think that has caused a lot of confusion for people. When you're on E5 and you go into the Purview portal and it says you're on a free plan and you're like what? I'm not on a free plan. I'm paying quite a lot of money for this. But it's like no, no, no. For this you're on a free plan. So I do, I do think that, I don't think it's just been slapped together, I just think it's a, I think it's another new horizon that maybe customers aren't thinking about today. You know, maybe they've only really just started to get their heads around their unstructured data. Right. And a lot of that's being driven by copilot, etc. So structured data is like another mind bending concept again, isn't it? You know, you know, what about the data that's sitting in Azure SQL? Well, never really thought about the lineage of that data and labeling it, etc. So I mean, I know, I know a lot of that. There are lots of customers out there that are thinking about that. A lot of enterprises that do, you know, have had that for a long time and they've got people in their organizations that are focused on that and managing those risks. But I think that's more of the rarity than anything else. Right. So.
Yeah, yeah. Yeah. I don't really know what else from the Microsoft site. Yeah, exposure, I think exposure management and the new data. Is it called data exposure management now? I think it's. Is that the security posture management data.
Security posture management, yeah, sorry, yeah. I think those types of consolidated posture tools are going to be quite big for customers because posture information has been spread in multiple areas, hasn't it, for a long time. So like they've done with the administrative side. Well, not just him and the secop side of the Defender XDR merge, I think we are going to see a lot more posture merging and I think that's where exposure management and the data side is now come from. Right?
Yeah, yeah, exactly. So I think that's the two kind of exposure management side of things. Yeah. So yeah, we'll see where that goes. But yeah, you're right, they've brought it all into one place which is good because like you've, you had the, the Defender cloud portal to see the CSPM side of things, which is still there, but a lot of it is now feeding into the exposure management sort of side of things as well. Especially attack attack paths and things like that.
Yeah, yeah, Attack paths are attack path simulation is of attack path mapping is, is big now, isn't it? Know, and I think that is, that really resonates with people as well because it's very visual, isn't it? You know, it breaks down sort of, I think it communicates sort of defense in depth really well, you know. And why? Because people can see it, you know, they can say oh this, this, this resource is linked to this identity or this X, Y and Z and you can physically map it out and where you might want to put extra protections and focus in place.
Yeah, well it's, it, you just kind of hit it on the head there that you, it might expose communication that you're unaware of as well between services.
Without an absolute doubt. Yeah, yeah, yeah, yeah. Because I, I did some Azure work for a, for a customer in the year and they were really concerned about reuse of credentials and plain text credentials in their environment. It was like a big topic of theirs and, and Defender for Cloud attack path. So that actually pulled a couple of instances out of it and they went and remediated those really quickly. That was a big thing for them in their environment and, and I do think when people see it and.
Was.
It more apparent because it was attack path, you know, analysis? Maybe that's what it felt like to me anyway. I've got no hard evidence for that, but it was just sitting there sort of in the portal and it sort of been overlooked I would say because there were a bunch of attack paths parts in there that were false positives and they hadn't been cleaned up, if that makes sense. So I do think it is a bit of data overload and information overload for customers. But also there are some really rich, you know, findings in there and. Yeah, but I do think we need to get to a stage with customers where they're. Any organization just say customers, but organizations where they're thinking about posture management as a continual exercise, you know, because it can, it can change at any moment, can't it? So it's something that needs to be monitored and understood.
Yeah, exactly. And it's. It's also when you're building new, new services and things like that as well, isn't it? Or M and A's, things like that. You know, you bring that into the. Into the. Into the. Org and then your posture increases or reduces either way, you know, because of just. Yeah, that side of it. So, yeah, we do see some customers. Not customers, but organizations at least, like you said, do it once and then maybe do it in a couple of years time and realize there's change there as well.
Yeah, exactly. Yeah. No, nice. Yeah. Right. What else were we gonna talk about? Oh, worst fail of 20, 24. Can be personal, can be corporate. I'll leave it to design. Up to you, Alan. What, what are you going for? I'm gonna go for the obvious one and I guess that would be the crowdstrike incident. Yeah. That didn't help anyone really at all, did it, to be fair? Just unfortunate. It was them, I think.
Yeah, Yeah, I completely agree, you know, with the amount of access antivirus and EDR systems have to, you know, end points, it's kind of almost a matter of time, wasn't it, until this, this type of incident happened, you know. Yeah. I mean, it has brought in some. I think Microsoft brought in some new frameworks around it so that if organizations do have that sort of level, they've got to comply with a. This framework now.
Yeah. To try and reduce that incident or that similar incident happening. Yeah. I don't think it's tough to completely, you know, mitigate it, but I think it's just adding in some more best practice sort of scenarios, I guess, around it. Yeah, I totally agree. Yeah, I totally agree. Any others this year? I can't think of any. I'm trying to think of as anything else out there.
I'm gonna. I don't know what to really call out, to be totally honest with you. I think Security co pilot or no, sorry, it's now called Copilot for security. No, no, Security Copilot. It's gone back. Sorry. Right. Yeah.
I think it had a bit of a rocky ride this year, I would say. I would say that in the quest to. And, and I. I don't doubt that AI augmented security operations is a positive thing. Right. Because I, I believe that most technical roles can be. The productivity can be improved by using AI or generative AI or whatever you want to call it, machine learning. But I kind of feel like it was a product that was a bit ahead of its time, if that makes sense. Like as in it was almost. It.
Was almost too big to start off with. You know, they, they connected so many systems to it and I, the, the people that I spoke to about it were in a bit of shock about how they would actually utilize it, if that makes sense. People that hadn't consumed, people that hadn't consumed the defender suite, slash, E5 didn't get enough value out of it because not all of their systems were connected to get the holistic viewpoints. And then people that had maybe didn't have big enough teams to really warrant the cost of it. You know, that's of what I've seen. So I, I think it's a victim of its own. I think it's just ahead of its time. I think is my, is my gut because the more recently as I've used it, it's been way more responsive. The, the answers that you get out of it are a lot more accurate than what it was originally. So I do think it's improving and it's moving in the right direction. I just think it's going to take some time to, to really show its true value. Is my gut on it.
Yeah, I think you're right. It's. You had to have the, the data to ground it in place to be able to make, get the most value out of it. And I think it was that you said it wasn't.
Well, we're still trying to convince people that have purchased things like Defender for Identity to even enable it and deploy it, you know, and I just, you know, and it comes back to the whole purview data security conversation as well, how mature organizations actually are today. If you took the average organization, where are they? You know, I, I wouldn't like to call that, but you know, how, how much further than EDR have they gone, if that makes sense, you know. Yeah.
You know, how many as a percentage of all organizations that have, you know, I don't know, more than 500 end points. How many of them are using a CASB solution today? I don't know. But you know, it's pretty rare to see somebody that's sort of ticked every box in terms of E5. Right. There are a lot of people out there. I'm not saying that hasn't. That doesn't happen. But it's, I think it's a rarity more than it is a norm, you know. Is my girl.
Yeah, exactly. And using the full suite as well. Yeah, it's a rarity. Yeah. Because there are, you know they do have organizations do have other solutions in front of or using them instead of Defender products.
Yeah, yeah. And just, just to take it back to Security Copilot is a lot of people that I've talked to aren't doing those day to day sec admin duties if that makes sense. So it's hard. It's not something they're doing today. Maybe they know they need to do it, but they might also be very resource constrained and they might be like I, you know, I can't take this on. You know this is just, this is like a next year and beyond the thing because I don't have the, the budget to even all the time to even consider it if, if that makes sense so. Because I, I do think that internal teams are trying to keep up with advances in technology, in advances in different threat actors and their, their attack vectors. You know it's a, it's a constantly changing and evolving landscape. So you know it is, is a real struggle for them to keep sort of up to date.
Yeah. Yeah. I don't know if there's anything else that you've. I think the only other one that I've just had taken a quick look at and it's not necessarily a, a fail per se but maybe they were just concerned when it was launched and that's co pilot recall. Yes, true. Obviously had quite a. Yeah. Backlash when it came out. Yeah. Around the screen grabbing every five seconds of your desktop and that at least at initial launch some users were able to access that information. Yeah.
Yeah. Again pass, you know, doing stuff, you know, the non normal user would do. So faking that if even words. But yeah, the concept I think was good. I believe it's still potentially an active thing but I think a lot of organizations initially are disabling if they've got a Copilot enabled cpu. Well, is it an npu? Yeah. Neural processing unit? Yeah, yeah, yeah. So yeah, yeah.
But I, I think for me like the amount of good products that you know, specifically Microsoft in this example put out versus their fails or near misses is. It's a very good ratio, isn't it? Right. Like we've just, we've seen so much, haven't we? Like this year. Like it's just, it's just every year it, I don't know if it gets better More crazy. But it just feels like there's just been a lot of different things this year that have landed. So kudos to the engineering teams and product teams.
Yeah, definitely. So let's, let's stop bashing Microsoft even though it's fun and easy to do what our episodes. Alan. Favorite episode of this year. So we did 42. So I will do stats in a minute. Yeah, so my, my couple of favorite episodes of. Of doing, you know, what ones we'd done, unfortunately are mine. Oh, of course they would be. Yeah.
So my favorite ones of doing was the token. Reducing the risk of token. Token replay. Yeah, Token theft. That side things. I thought that was a really good episode because it's good that we're talking about the actual sort of attacks, the play, you know, the processes or the scenario kind of thing. And then yeah. Not only just going, you know, this is really bad, you know, this is out there sort of thing, but actually here's a, a potential resolution around it or resolutions around it at least to reduce risk. I thought that was a really good episode that we did. And I think the other episode that came to mind was probably. Where was it? I think it's around the. The Glow score access side of things. I think that was. I quite enjoyed talking about that once we, once I was allowed to talk about it wanted to come out of private preview and stuff like that. Yeah, I think that was a couple, my couple of episodes that I did enjoy sort of doing. What about yourself.
Number one's got to be Azure data box, hasn't it? Let's be honest. Got to be quite possibly the most. I'm going to say ridiculous, but in a good way. Microsoft service. You need terabytes worth of data moved into region. Instead of sending that across, you know, fiber optic channels and paying for that, put it on a crate and send it. Send it into us. I think that was really cool. Data API Builder was a. One of those things that I found that I couldn't believe that I had missed basically, which was really good. Nice. I am actually using that at home every day, which is really cool. And then the other one. And I did pick one of yours, Alan, because I'm a nice guy like that. I like the episode on the CCP community.
Yeah. Okay, because.
And the reason why I say that is, is because things like that are programs and extracurriculars in Microsoft's ecosystem that I don't think a lot of people know about or think that they can. That they think they can't get access to. If that Makes sense. And yeah, those types of communities are really good because as far as I'm aware, it's just an active NDA with Microsoft as a prerequisite base, essentially. Right. But it is open to essentially anybody, if that makes sense. Customer partner, like it doesn't. As long as the NDA is there, you can request access to it, can't you?
So, yeah, as far as I'm aware.
So. And that is probably the best source of knowledge, inspiration, access, I would say, to Microsoft, because you can naively now legitimately critique something that Microsoft's creating and have calls with product teams to fill in your knowledge. Because they are benchmarking their changes with that community, aren't they? Like, how many times have you have you filled out a feedback thing, Alan, and you've actually got a call back from a product team member. It happens quite a lot, doesn't it? You know, if you're willing to go through that, you know, sometimes you can't do that because of availability and things like that. But they're very hot on that, you know, that side of things, aren't they?
Oh, definitely. I mean, I've had product group actually just contact me out of the blue now. Yeah, exactly.
To ask for feedback and things like that. But it's not just that, like you said, it is like testing out the Pro previews and things like that, but it's also helping shape the products before they go out giving that feedback that your customers or yourself is going to say. Basically they can either change the comms or, you know, slightly change the product to. Or add the features that are needed first. Prioritize.
Yeah, exactly, yeah. And you know, if you're being cynical, you'll be like, well, you're sort of doing Microsoft's job for them. But you get such an advantage knowing what's coming. You know, you don't know everything because they still keep things up their sleeve, don't they, that nobody sees. And even you as an mvp, you don't see them until late in the process. Right. Even if you even see them early, if that makes sense. So they do keep stuff back. But you, you, you, you understand context of what's happening, what's changing, why they're changing it as well. Because I think what's also really good in those communities is like you understanding why they're changing things. It might be like, oh, what was the one? What I love getting is the emails from like the data, the purview one. Okay, this might not sort of be your thing, Alan. Right. But what's great is when something pops in because a lot of that sort of data program, they email you. Right. Because you. They'll email you and they'll be like, we're testing, like, what was it? Copying DLP policies, which was an absolute pain in the backside for a long time. Right. Because for a long time you couldn't copy DLP policies. But then you'd get this email or you'd see a thing about a survey, or there'd be a private preview around copying DLP policies and you're like, God, that's a great idea. Thank God that's coming. And it's almost like. It is almost like a safety blanket for you because you're like, oh, I've got to deal with this now. But I know they're actually wanting to change things, if that makes sense. So, yeah, I think it's sort of good. It's good. It feels good if you're in and around the technology to sort of peer into the over the horizon, essentially.
Yeah. And it helps you prepare for the release of it as well. Yeah. So it might, like you said, if there's. If there's changes to your. Your work pattern, your bau because of it. Yeah. At least you can be prepared for it and start, you know, getting ready for it kind of thing. Because some of it sort of goes public preview or it goes. Some of it even goes straight into ga. Oh, yeah, yeah. You don't even see. Yeah, yeah, exactly. So. So, yeah.
Right. Alan, what about some stats for this, this, this year? 40, if we're including this one. 42 episodes this year. Yep. So that is exactly the same amount as last year when we did two seasons. Wow. But technically we had two finales, so we've technically done one extra episode of content in theory. Wow. That is insane. That is insane. Well, well done, both of us. So we only had 10. 10 weeks off. Well, it's just not good enough, is it? I know.
No. Yeah. It has been, I would say, for a long. I think it only got a bit disrupted during. I think we had a period over summer, I think, when we were just sort of focused on our three or four days of actual summer that we had this year. And I think in and around ignite time, things just got busy for us, didn't they? Work wise. And also because you were traveling a little bit more. And I think I also had some time off and things like that, so. But for a long part of the year, a long part of the year, we were very consistent, weren't we? We were Just sort of on the treadmill, sort of doing it. So yeah, sort of fair play from. From that perspective. Most popular episodes this year, Alan, what's. What's been the pick of the bunch?
I think GSA was up there. AI Studio. I think it was the biggest one this year. Yeah. I'm looking at the analytics now. I've got AI Studio as the top one. Kind of makes sense because that's gonna have a lot of crossover with other search terms, I would assume. But I think second I've got is Managed Identities for Azure resources. Yeah.
Which is interesting. I don't really understand. I know that area in Azure is quite. Not hidden and complex, but it kind of is, isn't it? Because I do. No, I don't know. I don't know whether that's fair for me to say. I don't see a lot of it being used but I have also seen a lot of it being used so I don't really know. But maybe it's a big concern for people. I think it's because you've always been. It's always been the norm to do client ID in secret. True.
That's been the norm to do. To access services across Azure and things like that. Yeah, that is true. And that you could at one point with like Azure Automation accounts, I think you could use. You could put in your credentials somewhere and use them sort of thing and that all got nuked out and it was to basically you can use Manage Identities or something like that. I think eventually. So I think I all sort of pushed it. Yeah. Yeah. The next one I've got is Microsoft Updates in March.
I don't know Microsoft Secure, wouldn't it? Yeah, could have been. Yeah. No, was that just before? No, we've been. March would have been end of. End of it. Yeah. There would been all the Microsoft Secure stuff, wouldn't it? Next to I've got is also Azure Backup and Azure Key Vault. So lots of Azure based. Well, does kind of make sense the name of our. A podcast but.
Yeah, yeah but that's. They are. They're going to be. They are sort of. I'm not going to say legacy services but they've been. Services have been there a long time. Yeah. Back up in Kivo, aren't they? So it's. It's just bringing them back into the light, I think.
Yeah, yeah, definitely. Yeah. No, yeah, we've. Yeah, there's been a lot of like really, you know, popular episodes I'd say some, definitely some standouts but you know it's all relatively consistent in terms of her listeners. So thank you ever so much for listening. That's all I'd say. Anything on the stats front that you want to call out? Alan, I was just going to say the, the, the all time top episode is hello world. The very first one. So not by much though. And the next one, Entra.
Okay. All right. And then Copart Solutions, then AI Studio. Okay. There's only three more across the whole, you know, all of our episodes. 118, 119 worth of episodes that we've.
Done ever listener locations. I currently only have access to last six months worth of data. I believe I can only do. Let me just see if I can go back further but I think I'm limited to. Oh no, I can. I've just got to put a custom filter on. Okay. Sorry. Two Sex. So United States is the top and it is three times as high as the United Kingdom, which is second, which is twice. Well just over twice as much as our third biggest location which is Australia and then followed relatively shortly afterwards with Canada and Germany. So thank you ever so much to our US listeners for your support for the show. But yeah, definitely you US and UK are all our top two.
Yeah. Just looking at the. All of the countries, Stephanie. Quite a few. It was definitely, you know, we do get around. Yeah. And also thank you for our single listen from Cuba, whoever you are. If you're also listening to this at the bottom of the list. Yeah. Nice. Right, Alan, the next thing I had on my list was predictions for 2025. We need to make a note of these and we need to. We should review them in the next finale.
Yeah. So I guess the obvious one's going to be more AI in our lives, specifically. What? More AI. You're just going to say just more AI. Just AI plus more AI. No, I think it's going to be more. I reckon it's going to be more AI doing stuff for us to actually take in some actions. Okay. Scenarios.
Yeah. Okay. Yeah, yeah, I'm gonna go for. Local AI is getting more and more prevalent. I think people like Apple are pushing. We've had refreshed MacBooks. Well the refreshed Mac line with minimum memory being bumped to I think 16 gig across the board now to support Apple in. Is it called Apple Intelligence? I think it's called Apple Intelligence. So I personally think that we're now going to see more of a divergence from more of a focus on local processing. I think that we've seen this with the copilot is It Are they called co pilot plus machines?
Yeah. Are they? No. Qualcomm arm SOCs that are going into these laptops, I think we're going to see more and more local NPU processing power so that you can drive your own micro models. Client side essentially. Yeah. Intel have got some now, haven't they? I think. Or at least coming out for laptops anywhere. I don't know about normal desktops, but.
The other day I watched a video of somebody buying because the new Mac Mini is like really cheap, the base model and he'd bought a bunch of, well he didn't just buy the base model ones but he'd bought like a bunch of different variants because they've all got Thunderbolt 5 on them and you can do Thunderbolt 5 networking between them basically. So he had built like a, an AI cluster for it. And what I really learned from that video was was that because Macs have a unified memory, they can dynamically shift the amount of memory that is available to their GPUs across their die. So you know, traditionally if you wanted to run with any sort of performance, a local model, you needed a graphics card with like a lot of, you know, video memory. Right? And that costs a lot of money. You know what's a 4090 nowadays? 2000 pounds, 1500 pound, 1800 pound, whatever you want to call it, that'll get you 24 gigs of video memory. And the amount of video memory is sort of how bigger model or more sophisticated model shall we say? That's an oversimplification but let's just go with that for the moment. You know, the more memory you have, the in theory the better the output of the model. Right. That isn't technically true, but let's just run with it as an easy thing for us to digest. Now what Apple can do is Apple can say on their silicon they can say so if you spec a MacBook Pro, a MacBook Pro can now have 128 gig of RAM in it and it can dynamically allocate a lot of that memory to the GPU. So actually even though the GPUs aren't as strong, the size models that you can run on those machines are basically incredible what you can do. So. So I do think that we're going to see more of that in other areas. I don't know how these co Pilot plus machines work. I only really thought about this or under. I wouldn't really heard about this the other day and sort of it popped into my head. So I think we're going to see a lot more of that coming through so that you don't need a $20, you know, or $3,000, $10,000 a month subscription to these types of services. Your local machine is going to be able to take some of that processing away and allow people to scale. That's my gut for AI.
Yeah, that's probably fair and like you said, I think. Yeah. The heart from a hardware side. Yeah. MPUs are going to be a standard thing because you can imagine the use with them as well just for like gaming that side of things. Yeah. Running your, running your single player games, the AI just pretty much beasting you.
Yeah, well we, I think we've seen, I think we've seen examples of games now that are fully AI generated. Like as in the worlds are pre generated dynamically as you walk through them basically. So everybody gets a different experience of the game and you're just playing a model basically and it's just generating the environment around you. So be interesting to see if generative worlds, you know, become a thing. I assume they will do because it's kind of here to stay, isn't it? Yeah, definitely.
Any call outs of any microsofty sort of things sort of come. More consolidation of portal. Yeah. Getting rid of other portals. Yeah. Yeah. Cope co pilot being like I said, more, more in everyone's day to day. Yeah. I think because what they calling them now agents, aren't they? It's like, it's like the Matrix basically. Project agent.
Yeah. Stop bringing that specific one up. It's like you're making a statement. He's not the. Yeah, I think those, I think to your point earlier, those agents are going to start to do more for us, if that makes sense. So I think that's going to be a very powerful, very powerful thing, that's for sure. I did actually see today on one of the SharePoint sites I could create the SharePoint agent on it. Could you? Yeah, I was like oh do I better check. No need to find out what the.
Cost is or if it's included. Yeah, I don't really want to make any more security type predictions, I don't think. I just think like we see every year there's going to be new novel threats that appear. There's going to be a whack a mole situation with them I believe. And I think it's just going to be a never ending sort of security battle I think personally. Yeah. This year was token theft, wasn't it? Really was the key one rising from an attack perspective.
Yeah. And From a risk perspective, I saw a lot of change in terms of data this year. I would say sort of never ending DLP requirements, I would say. So that's been a key. I think that's just going to continue to accelerate. Personally, I think IoT is going to become more and more prevalent as sort of regulation takes over, you know, requirements of monitoring of. And I'm talking ot, when I'm talking iot, I think. Yeah, more than anything. So I think we're going to see a lot of that because. Because a lot of manufacturing is sensitive in terms of what it is, you know, supply chain, global supply chain risk. So I think we're going to see a lot of food, chemical, industrial defense, etc. Etc. Need to layer on additional energy, et cetera, critical infrastructure that's going to need to be prioritized.
Yeah, I definitely agree with that. Yeah. So what's. What are we going to do? God, we've been babbling on for 55 minutes nearly. What's the plan for us next year, Alan? 43 episodes maybe? Yes, I think we're going to stick it to stick to the same one season for the year. Yeah, it worked out. Yeah, we didn't take a break even though we kind of did at the same time. But in mid year. But yeah, yeah, 40th. We're gonna aim for an extra episode, are we? Well, maybe, maybe not. Yeah.
I think one thing that we're also going to start to look at is more specific features and focus areas of things that are popular in any moment, if that makes sense. You know, I think your episode on Token Replay was well received and it was informative. Not just of. It was an actual real attack vector, a mechanism that you need to protect against. Right. So I think we're going to work on some more real world examples of different things, I would say.
Yeah, I was going to say that's probably some good ones that we. That was definitely a good one to sort of showcase that sort of type of episode, wouldn't it? So. Exactly. Definitely more of those.
Yeah, I, I'd really love for us to get more of what we're seeing day to day and that is, you know, really truly in front of people. I think it is good giving people grounded information in terms of different products, new products, releases, those types of things. We are sort of. We're not coming to the end of the things that we can talk about. We're coming to the end of the things that we want to talk about. I think is probably a better answer to that. So I think we need to start focusing on real things that we're seeing day to day. Real challenges we're seeing in the field and sort of sharing that as far and wide as we. We possibly can. So I don't think we want to sort of chase clickbaity titles and scary situations, but, you know, pragmatically and logically raising topics that we feel like we should be sharing, you know, with each other. Because I think, you know, we do need to remember that the reason why we started this podcast is these were the conversations we were having with each other. Well, you telling me stuff to upskill me, essentially. So I think we can need to continue doing that and just try and make that a bit more real, if that makes sense for people.
Yeah, definitely. Yeah. Yeah. I don't think it's probably anything else. We don't need a rebrand yet, do we? We're not. You know, it's been a couple of years since we rebranded all the.
Yeah, yeah, I'm. I think we should sit down over the Christmas period and just review all of that. What we want to do. I know we added YouTube this year, didn't we? So. Which has been going really well. So we just need to think about. Yeah, think about that, I think. Yeah. Oh, we did also have. This is completely random and nothing to do. Well, it is kind of to do with what we were talking about, but we had a message from a listener around a quite a cool feature around Linux baseline configurations. So, Andy, thank you very much for pointing us in the direction of that. We had that after Iowa ignite recap. So we'll do a dedicated episode on that technology because it wasn't something that I was aware of, but when Andy highlighted it to us, well, to me especially, it's something I will cover in early in the new year. So I just wanted to call that out. So we do get your feedback, suggestions and notes. So thank you ever so much for sending that in.
Yeah, and I guess it kind of comes back to as well, you know, if. If we do get feedback or suggestions of episodes, we will consider them to then to do because, you know, we do build our own. We do look at what content to sort of talk about. But there's, you know, listeners that have something they, you know, they're struggling with or an area they want some more info on and yeah, we're trying cover it as best we can.
Yeah, yeah, no, definitely. Right. Alan, should we wrap this one up? I think we've been nattering on. I think it's our longest finale episode ever. Probably the most longest with the, the less amount of notes for it. Exactly. Amazing.
Cool. So I guess if you did enjoy this episode, please do consider leaving us a review on Apple, Spotify and YouTube. This really helps us reach out to more people like yourself. If you have any specific feedback like I just talked about around episodes and things like that, we've got links in our show notes and on YouTube to get in contact.
Yeah. And if you've made it this far through this episode and through this season, you know, thank you ever so much for listening. We wish every one of our, you know, listeners a, a good end to the year and, and a break if that's, if that's, if that's what you're doing at the end of the year and yeah, we'll be back in the new year. Yep. Thanks all. Thanks. Bye.