Welcome to episode 354 of the Microsoft Cloud IT Pro Podcast. Recorded live on September 22nd, 2023. This is a show about Microsoft 365 and Azure from the perspective of it pros and end users where we discuss a topic or recent news and how it relates to you. We have some news today. Finally, we have a release date for Microsoft 365 co-pilot.
We also discuss a few other topics from the Microsoft Surface and AI event where co-pilot was announced as well as the acquisition of Splunk by Cisco, which was also riddled with reasons around ai. The only thing that wasn't AI in this show was Scott and I discussing it all. Hey, we're back and better than ever. We are. We're not better than ever are we? Why is my. You're dark. I'm so, so low. Yeah, that's a ISO auto. I don't know why it's low teams are kind of dark too.
It's like your camera's just not wanting to adjust. I can't let any more light into this lens if I wanted to. You know what it is, Scott? What is it? It's becoming later in the summer and the sun's further away. So it's darker. . I, I don't even, I don't know. So funny enough, I have three different pictures. I have the one in teams, I have the one in invite cam and then I have the one in Discord and they all get Progressive.
And I guess I have the one on my camera 'cause I have the monitor up in front of me. Uhhuh. . They all get progressively darker like, Hey, we're fine. We're ready to go on the camera. And by the time it gets to Discord, nice and dark. So I'm gonna blame it on your side. Okay. Even though I shouldn't. Okay. News stuff, random news. Stuff. Where you wanna start? How about. Cisco? Cisco. Cisco's buying Splunk. Cisco has some money to burn . Hefty 20, was it $28 billion?
Something along those lines. And interesting enough, so this was announced. Yesterday, right? It was just like recently. Yeah, yesterday or the day before. Yesterday. The 21st. San Jose in San Francisco. Ha ha off the presses. I kind of forget about this. Sometimes with W with Splunk and where they sit, like I still tend to think of them as just a, an observability play and, and being able to go back and kind of do logging and, and really like logging. Right? Like sim, right?
Like you always think of Splunk and a sim as my initial reaction. I don't know. 'cause I don't touch Splunk every day. I always kind of forget they're there. And I really tend to, I guess, downplay it in my head as to like capabilities and what they are. So like I'm, you know, a little naive about it. I'm like, oh they're logging and then Cisco buys them and kind of the top build item is we wanna do like this whole integrated security detection and and
threat detection thing. So to your point, very much like a sim seam, whatever you wanna call 'em and putting that all together, which is like makes perfect sense, right? Like that's what Splunk is. Yeah. But yeah, just , I don't know. I completely lost sight of it. And then to see Cisco buying them was interesting. It's like you, like you said, money to burn someplace. Yeah. And it's an interesting play, right? Because like my first thought was why in the world would Cisco buy
Splunk? 'cause on the other flip side of it, I always think of Cisco as I guess telephony but then networking. Like they've always been networking hardware to me where Splunk is, this sim seems software and it's, I guess my initial thought was everything goes through your switches. So if you have a bunch of logging with Cisco switches and they can somehow integrate Splunk to do a bunch of stuff there. Or maybe it's just Cisco not wanting to be quite so,
such a narrow pillar. Wanting them to try to diversify a little bit. Like arguably I'm not in the Cisco and Splunk space every day. Like you I sit in the Microsoft one. So it was just an interesting acquisition for me of why I think more than anything else, I think. It's another good example of consolidation in the marketplace around tighter integrations for hardware and software. So Cisco has kind of its software stack, right?
Like you have the other iOS that runs on Cisco switches and routers and things like that, but they really didn't have the software story that Splunk has. So now you marry those two together, let 'em marinate a little bit and see what comes out the other side. That's a probably goodness thing for customers. Who knows? We'll see licensing and all that stuff tends to change as these types
of things execute and go all the way through. But yeah, I think it just speaks to that further consolidation that you see across the technology space where kind of key players that are entrenched in, in one area only software only hardware, they do see these kind of natural synergies by coming together and building potentially more holistic products. And, and you see it across tech right there.
There's certainly the apples of the world. You know, we could talk about like Microsoft doing it, like to the degree they do with hardware that you've got things like this. You even see it in like cars, Tesla, rivian, like electric car manufacturers kind of trying to own this entire stack and, and have it go and, and end to end and get it all together. So I think this one will be kind of cool, you know, Splunk's all over the place.
While I don't touch it day to day, I run into customers every single day. who use it and ask for like an integration that doesn't exist. . Exactly. You know, trying to get them going and get all that stuff going. I I I think it'll be all very good. So it, it not only gives Cisco like a solid software play and really into that like threat detection observability market, it also gives them an interesting inroad to cloud providers, right?
Think about like a W S and G C P like and having sync points and Splunk to multiple offerings within, you know, the hyperscalers and public cloud. It starts to become kind of an interesting play and potentially exposes Cisco to new customers as well. Yeah, we'll see 28 billion is a fair chunk of change. I thought it was also interesting that they expected to be cashflow positive by the end of their first fiscal year after they close.
So even though they're gonna spend it like they think they're gonna make it all back rapidly. I. Agree. And oh I can't quite, ooh, let's do this. So I'm curious Scott, this is the other thing. You know what else speaks to, oh they didn't put it in there quite as much as I thought they did.
It's also interesting that speaking to the world we're in today, how many times they mention that this is mention AI in the article, lead in security and observability in the age of AI to establish leaders in ai, which I don't know that I ever thought of Cisco as an established leader in ai. AI enabled security and observability data and AI to deliver customer outcomes. They certainly don't make GPUs . That is not something that's happening. But what they do have is data.
And by buying Splunk they get a lot more data. And as we've seen with things like large language models, LLMs, you know, open ai, all that kind of stuff, the more data you have and the larger the corpus, the potential is there to do more interesting things with it. And it really, if you, if you go and read between the lines on this one, like that's what this one is. Like there is potential there get things going. Yes. It is interesting.
So other than that it just doesn't impact the Microsoft cloud a whole lot unless like you said there are plenty of people running stuff in Microsoft that are using Splunk for their sim. Everybody tells me it's SIM in the us, Scott and SI in other countries. I still don't know . I was talking to somebody else the other day that's like it's sim, it's always been sim I've never heard si.
But then a bunch of other people say si whatever it is people, yes people running Microsoft stuff that do run Splunk instead of Sentinel. Although I have been running into Sentinel more and more. Is that what it is? It's a US versus Europe thing. Yeah. Maybe it's a pronunciation of the vowels. Yeah. The first time I encountered Sentinel I was doing training over in Europe. . Got it. I learned it as seam. So that explains a whole bunch and yeah. There you go. That is the best explanation I have.
Well the only explanation I have ever gotten on whether it's a SI versus a sim, is well country-based. They spent some money. They did, they did some things. You know what else you can do here. Here's a good pivot for you if you're ready to spend money in other areas. Ooh and maybe not on Cisco and Splunk. I like spending money. Yeah. Do. You? No, we were chatting kind of before we started recording but we have a rapid private preview to GA for copilot in Microsoft 365.
Yes, this was, it kind of came out of the blue and we had some theories about how it kinda came out of the blue but the other day, right what day was this one? September 21. Same day as Cisco was announced, there was an event in New York where Microsoft announced their vision for copilot. So we finally now have a release of co-pilot or release date.
So they said first step they're unveiling a new visual identity, a k a, an icon for co-pilot and we're gonna start seeing that icon roll out across co-pilots that are already there because there are some co-pilots already there, right? Like Power Apps already has some co-pilot stuff. Dynamics. Dynamics. They're, they're all. Out there and playing around.
But then the Microsoft 365 co-pilot, which is the one that everybody seems to be waiting for until they found out they had to spend $30 per user per month on it, is going to be generally available for enterprise customers on November one. So this does not appear, and I have not found it anywhere in this article, that there is going to be any sort of public preview of co-pilot.
It is going immediately from that private preview, like you said, straight into general availability on the 1st of November. So you have like five or six weeks to wait Scott, and you can get copilot if your company will pay for it.
It's interesting for this one, if you look at how it's being integrated and rolling out how they're using copilot in Windows as a potential loss leader to build you into things like binging chat and the enterprise variations of it and then everything else that comes along the way with M 365 copilot integration with teams, integration with all the M 365 apps, you know, the better, better together kind of security story around like security,
compliance, all that kind of stuff. But , yeah, copilot and Windows is a little, a little trippy and in your face if you have a chance, like I'd encourage anybody who didn't see the event, it was kind of this weird mixed hardware and software event kind of thing that happened in New York around surface devices and then all these copilot announcements and all those kinds of things that
came along. So yeah, it does give you a sense for where it's integrating, where it's rolling into Windows, how it's going to start to look and kind of manifest across office apps and things like that, which we've seen in the past.
And then they had another interesting one that they announced, which I think this is nothing but goodness, I don't know how, how you found learning how to treat LLMs or how to kind of give them better prompts so that they give you your desired output and all those kinds of things. It turns out that getting an L L M to give you a desirable output is you know, a little bit harder than going in and just asking a question and firing and
forgetting and seeing what happens. So they also announced co-pilot labs, which is gonna be a web-based experience to get hands-on with particularly M 365 copilot, which I get the sense like it's actually like gonna need some kind of poking and prodding given the fixed meta prompt and maybe if you wanted to go in a slightly different direction, things like that along the way. So they're also gonna be actively training users around copilots.
I think that'll be interesting to see how many folks feel like they've all of a sudden had to become programmers and get away from some of their, some of their other stuff that's out there and available. So maybe with a little bit of lab time and getting hands on across the product stack, that stuff starts to manifest and light up and we can all see what it does. Do you feel overwhelmed by trying to manage your Office 365 environment?
Are you facing unexpected issues that disrupt your company's productivity? Intelligent is here to help much like you take your car to the mechanic that has specialized knowledge on how to best keep your car running Intelligent helps you with your Microsoft Cloud environment because that's their expertise. Intelligent keeps up with the latest updates on the Microsoft Cloud to help keep your business running smoothly and ahead of the curve.
Whether you are a small organization with just a few users up to an organization of several thousand employees they want to partner with you to implement and administer your Microsoft Cloud technology, visit them at intelligent.com/podcast. That's I N T E L L I G I N k.com/podcast for more information or to schedule a 30 minute call to get started with them today. Remember intelligent focuses on the Microsoft cloud so you can focus on your business.
I will be interested to play with that 'cause it is like the whole meta prompts and all of that is still something I've been learning about playing with and by no means an expert in. But that is definitely I think the biggest learning curve. People think co-pilot's gonna come out. AI is here, you can just start asking it questions, which to some extent you'll be able to like Outlook is gonna have summarize an email thread for me or you're gonna be able to ask co-pilot to rewrite a paragraph.
But then you're gonna have to think through how do you want it to rewrite it? What type of verse voice do you want? Do you want it neutral, casual, professional. So I think it's gonna v vary too and and then like there will be basic things you'll be able to do but I think to really take advantage of it, there is definitely going to be a learning component of how do you tell AI to actually do what you want it to. I've got a little tip for you just 'cause you mentioned the tone thing. Yep.
I found tone to be very interesting with all the kind of variations of models that are out there. And I think it's one of the complaints that folks have when they see these models in their outputs is they look at 'em and they go, oh well that sounds kind of generic or it sounds a little bit like a machine or it reads like a machine wrote it kind of thing.
So a trick that I've been playing around with, I saw it on a YouTube video, is setting tone by providing examples of your own writing and then having them generated like with that style. So are you familiar with how you can use variables inside of OpenAI and Chat G P T? Oh I have not done anything with variables yet. See I'm still learning. You can have a variable. So say I wanted to create a variable like for like my writing style.
So I tell the engine, Hey I want you to store some data about the way that I write inside the variable, you know Scott underscore writing style. Okay, whatever it happens to be.
And then you say I'm gonna provide you some examples and then you just say, okay, example one of Scott's writing style and, and the way I've been doing it is just copying and pasting blog posts that I have out there or kind of longer form form posts that I've had that I think capture my writing style and conversational style and putting those in and you say okay, store these in, you know that variable like Scott writing under style.
And then once you're done with that and it's got it stored and it understands, you go back and you say, okay now I want you to do whatever the work is that I want you to do but I want you to respond using the style that was stored in the previous variable Scott writing style and then it spits back something that like I'm finding needs a little bit less editing. Okay.
Both in chat G P D 3.5, the the three five turbo and in the 4.0 although I see I I find myself still sticking to three five Turbo these days as well. I'm not seeing a lot of additive benefit to being on 4.0 yet. But we'll see. I think it's a cool thing. So you should go explore variables if you haven't done that. Like if you've been working on prompts and conversational styles and and tonality and things like that.
Particularly for you 'cause and maybe for folks who listen to this like if you ever write like papers, anything like that internally, like something that you can provide to the language model to help it understand like your style it's a little bit better and nothing but goodness. I'm curious when you've been using the different models like the three five, the three five turbo, the four, what have you been using it for? Has it been for summarizing stuff or for generating like net new
ideas or rewriting stuff? When do you use those models? Lots of summarization. Okay. And I think that's probably gonna be the place that I use uh, co-pilot. Like should they. Like the co-pilot a lot broadly. Roll it out and make it available to everyone at Microsoft. I think that's the place that I would use it the most. Particularly in my world where I spend a lot of time in email Word and teams often talking about and working towards the same kinds of things.
Like for me I think it'll be super powerful to have like a meeting summarization that also incorporates like the document we were discussing as an input to be able to you know, get to whatever output that I'm, I'm kind of looking for along the way. But Summarizations are a great place. I don't like using them for answers to questions. Like I'm really, really super scared of the hallucination, hallucination kind of thing and getting that going. So that's been all good.
And then the other place I use it a lot is on the reference side of things, particularly when it comes to like syntax and not coding but like day-to-day kinds of things. So PowerShell use it a lot for K Q L and I have a feeling I'm gonna use it a whole bunch in Excel as well. Like once that rolls out a little bit more but we shall see. Got it.
So I was curious 'cause I heard the 4.0 model and I've used four G P t four for summarizing like podcast transcripts that four tends to do a little bit better with a summarization of that type of content. But I haven't done enough of 'em yet where I've sat down and actually compared to see what I like better. But I'm with you, I'm probably gonna use it for a lot of summarization. I have used it to ask questions for it but like you, I'm very worried about hallucinations.
So when I am asking for answers I tend to use the binging chat for Enterprise because not only does it give me an answer but then it provides my references. So I treat that as a little bit of summarize your search results for me is how I look at binging chat for Enterprise or Bing Chat is summarize the answer to my question from what you know and what you can find online.
But since it provides my references, I still always go back and check the references it provided to make sure it didn't misinterpret things or I've seen where it pulls from like two or three sources and get some kind of mixed up and comes with an answer based on those three. But it like it combined to them very poorly. Kinda like a kid misinterprets what their parents says or alters a word or two.
I have seen Binging Chat do something similar so when I do ask questions I do it someplace where I can actually go verify the references and leverage it a little bit more as a summarization of a search. I mean I'd be interested to know like this is another feedback thing that I'd love to hear from folks out there. Just like where are you using it to use it in similar ways or are you finding power in the whole G P T four and the ability to have plugins in use and all that kind of thing.
Like I'm really not using any of that stuff today either. So it'd be I think interesting to hear some learnings from other folks. Oh while we're talking about the AI thing here. Okay. I think it's important to kind of touch on, you know, the way that it sounds like you're using chat G P T and I'm using it the same way.
I'm certainly using it to augment and potentially make things a little bit easier and I say potentially, 'cause again you gotta go and learn this whole kind of meta prompt model and and figure out how to do prompts and it's almost like learning a new programming language like or picking up a new skill. Like it it, it is taking a little bit to like wrap my head around and and be consistent with. But if you go and watch the event from the Surface and AI
event, one of the interesting things of it is the tone. Like a, if you remember when LLMs came out, everybody was talking about how there's the potential for these things to take jobs away. Yep. And not really be like augmenters like over time like oh if I can teach an l l M to do this thing then maybe I can, you know, replace that person who's sitting at a desk, a frontline worker call center, things like that. I think they're like everybody's trying to walk that back a little bit.
If you go listen to the tone and go read some of the blog posts and things, everything is about helping you do your work now and being very like crystal clear that this isn't replacing your job, it's making your job better, faster, stronger. And you shouldn't worry so much about, oh a computer is gonna step in one day and magically take your place kind of thing. Which I thought that was a kind of nifty pivot, right.
Just to hear everybody actually say it out loud and and kind of commit words to blog posts and and paper and things like that, that LLMs are here but yeah they're not gonna take your jobs. I. Absolutely agree with that tone and watching some of the videos and listening to some of the interview questions and how they talked through it about even having AI and meetings and all of that.
It's absolutely help you do your job better like in a meeting, focus on the meeting and let AI take care of the notes and the summarization and some of that stuff that before like, and I know I do it in a meeting, you're trying to say you're trying to listen, you're trying to ask questions, you're trying to take notes at the same time where letting that AI help you do that note taking process, do the transcriptions, all of that.
That type of tone seemed to be like you said a lot of the focus on this. So the other interesting thing, pivoting a little bit was all of this came out of the surface event and I had kind of seen that a surface event was happening and then all of a sudden this copilot stuff came out the other day and there was one headline that said an AI event and I was like I didn't see anything about an AI event and it was the surface event.
So there we got new surface devices but I don't believe the man behind the surface was actually there for this event. So it was, we were talking a little bit about it's Panos, right? That's how you say his first name. Mm-hmm . Panos Penne is leaving Microsoft after 19 years, the guy behind a whole bunch of the surface and Windows 11 and all of that. Like literally left days before or announced it like days before this surface event that happened yesterday the 21st.
He is going to Amazon and all of a sudden this seemed to be a lot bigger focus on co-pilot, less on Surface. But we do have some more surface devices too. No duo, sorry Scott. No more Surface Duo but. I don't think you're ever gonna see another one of those. Uh, who knows. I hope they just quit trying to do phones. I'm tired of this. Let's do a phone like Nokia's then Windows phone and let's not, no, let's do a phone with Surface Duo and no let's not.
But there was a new Surface Hub which that one actually surprised me a little bit too because I'm sure you have a bunch of 'em at Microsoft but I do not see surface hubs in the wild that much. I don't go into an office , I'm a you don't, I'm a remote employee so , I don't When. Are you gonna start going into the office Scott? I don't see them either. You don't see. Them either. I do not but. We did get like a new Surface go and a new Surface Pro. And what else was there?
Was there a new Surface laptop Surface Pro 10 Surface go Surface laptop Surface Studio and the new Surface Hub. So if you're a Surface person, I am not as, I'm sure everybody that listens regularly is well aware you have some new surface devices to go spend your money on as well. I lost your link with this announcement. I don't know, did I overwrite it with a search I had? Well you had a link. You sent me a link. Now it's gone. I. Will put it back. We'll put one in the show notes, show.
Notes for you. And the Discord chat. And over to Discord. One more for you before we kind of move off surface and AI stuff. Oh okay. And just on the topic of kind of like LLMs and, and what's going on there, there's this super good podcast from today explained folks at the New York Times and it's titled, we Don't Know How AI Works .
This has been one of the things that's kind of niggling in the back of my head is we've got these LLMs but like do we actually understand like holistically and end-to-end end how they generate their outputs? Are they doing truly like formulaic responses? Are they just going off the deep ends and hallucinating and and all the, all those kinds of things. So I recommend everyone go and and listen to that .
I thought it was kind of scary eye-opening, all those kinds of things for where this stuff is headed. I will have to go listen to that one as well 'cause that would be interesting to listen to that take on. I guess is it their take the researchers, do they interview people or is it the New York Times, like regular podcasters? I do not listen to today explained on any sort of regular basis. It's them doing some interviews so it, it's actually a two part thing.
So the first one was we don't know how AI works and the second one was call was called we're trusting it anyway. So I, I recommend giving both of those a listen today Explained is not New York Times, I guess Fox Media, neither here nor there, but it is, yeah, it kind of walks through, you know, both sides of the story and gets that stuff going. So eye-opening especially like if you don't build these things you're just wondering how they come together, how they work and what all that is.
Sounds good. Oh, anything else Scott that we want to talk about besides Cisco AI surface and all of that? I know there's some more news. There's always. Always news. Stuff out there. That might be a good spot. Yeah, I don't know. Yeah, you're standing between me and a waterpark right now. Scott . I, yeah, I'm between you and debugging some power shell and then your vacation. So. Yeah, we're not gonna talk about that. You can remind me, Scott put that on your list to talk about next week because
we can talk about me and my power shell. I made a mistake. All. Right. So just f y i next. Week and PowerShell. My PowerShell script is still running. Scott, it was not Quick , we can talk about some PowerShell fun. Lots of restores from Recycle Bin is not just a quick metadata operation. Yes, that's neither here nor there. We'll talk about it next week. Thanks for the time. Alright, sounds good. And enjoy the waterpark. Thank you, I will do that. Enjoy your weekend and we'll talk to you later.
Thanks. Ben. If you enjoyed the podcast, go leave us a five star rating in iTunes. It helps to get the word out so more it pros can learn about Office 365 and Azure. If you have any questions you want us to address on the show or feedback about the show, feel free to reach out via our website, Twitter, or Facebook. Thanks again for listening and have a great day.