Ep 14 - Copilot Readiness with Michal Pisarek

Is your tenant ready for Copilot? Do you know what needs to be done to be ready? This week, Kevin was joined by Michal Pisarek, CEO and Janitor at Orchestry, who has been involved in governance in Microsoft 365 for many years. We talk about Copilot Readiness and three key areas for governance:
- Understand the environment
- Review oversharing
- Sort out your sensitivity labels

Zoe and Kevin also cover a few bits of the latest news, including some Drama Llama action with Whats App, some impressive custom copilot creation from Microsoft Research and a plea from Kevin to provide ideas and feedback to Microsoft.

Don't forget that we are holding a Month of Copilot in June and still looking for people to share sessions. Submit your sessions at Month of Copilots: Call for Speakers.

Useful links:

Project VASA - https://www.microsoft.com/en-us/research/project/vasa-1/ 

Lllama 3 launched - Meta steps up AI battle with OpenAI and Google with release of Llama 3 | Meta | The Guardian
Comparison of recent models - https://x.com/rschu/status/1781007659493212215

Meta AI in WhatsApp - How to chat with Meta AI in a group | WhatsApp Help Center

Privacy and Meta AI - https://www.facebook.com/privacy/guide/genai?entry_point=whatsapp_genai

Feedback portal has Copilot for Sales and for Service - Community (microsoft.com)

For more news, subscribe to your newsletter on LinkedIn at LinkedIn

Transcript

[Kevin 00:17]  

Welcome to the Copilot connection.  

[Zoe 00:20]  

We're here to share with you all the news, insights and capabilities of the Microsoft Copilot ecosystem from across the entire Microsoft stack. I'm Zabby Wilson and I'm an executive at Avanade in our modern work business, an MVP for M365-A Regional Director and Aviva Explorer.  

[Kevin 00:38]  

And I'm Kevin McDonnell, I'm an MVP, Viva Explorer and the copilot strategy and modern workplace AI lead at Avanade. And we'll be releasing these episodes as podcasts and we'll be bringing experts from across the community, from Microsoft and talking about what the Co pilots are, the impacts they can make to you and your organization, what you need to do to start implementing and even how you can extend them. And we're on to episode 14 now, Zoe. And I really wish I'd learnt that I didn't have to keep looking up at my notes as we talked through it. But hopefully I'll get there one day.  

[Zoe 01:12]  

14 + a whole heap of mini episodes.  

[Kevin 01:15]  

That's true. And and and the one that we lost with Trisha that we had to redo.  

[Zoe 01:20]  

Yeah, we don't talk about that. Actually we do. I'm sure we've mentioned it on about four other episodes.  

[Kevin 01:25]  

Yeah, yeah, I've learnt the guilts from that one.  

[Zoe 01:29]  

Yeah, so.  

[Kevin 01:30]  

What are we doing this week?  

[Zoe 01:31]  

Yeah, so this week we've got a few bits of news. And actually most of the news is outside of the Microsoft ecosystem, isn't it? But there's a few things that we've seen that we wanted to share with you. And then a great interview with Michael Pisarek that Kevin from Orchestra that Kevin did earlier this week. But first, sorry, go.  

[Kevin 01:50]  

On and just please stick around for that because with really great interview where we talk about kind of governance, but as Michael says it's copilot readiness for that. So we talk about some of the things needed with that, which is really interesting conversation.  

[Zoe 02:04]  

Yeah, a really great interview and it was really good to to have another guest take part. And the first thing we wanted to talk to you about though was the month of Copilot. On our last episode, we announced that we're holding a virtual event for the whole of June where we'll be releasing episodes every day or sessions every day that cover all of the different Co pilots. We opened our call for speakers and we've had tons of really great sessions submitted from across all of the different products, but mostly for Microsoft 365. And despite coming from that Microsoft 365 background, actually what we really want is good sessions from across all of the other Co pilots as well.  

[Kevin 02:47]  

Yeah. And we want to see that breadth. You know, we'd love to see some demos and people talking about the other Co pilots. It was lovely having Rouen in the last episode talking about Copilot for security, but kind of talking about what you need to do to prepare for Copilot, how you get that adoption out there, how you measure success across the different Co pilots, the differences types of Co pilots. So with copilot and power platform, you know, it's it's kind of just there GitHub copilot, it's just kind of there. Let's hear some stories about that. Love to hear anyone who's looking at things like copilot for finance, copilot for service, copilot for sales. It's like hint there Trisha Sinclair of who we can get along and hear from those different things. So please, if you're in that space, we would love to hear from you. If if you're not, don't let the hold you back. We still want to hear from you, get that submission in, but we want to encourage people from the other areas as well.  

[Zoe 03:40]  

Yeah, 100%. So the call for speakers closes at the end of this month and we're about to follow Donna Sarka's example and start volunteering people. So please get your sessions in. Like Kevin said, we we'd really love to hear from you and I'm sure everybody else will do as well. But we are. We're going to hunt you down and find you if we know if we know you're working with Copilot.  

[Kevin 04:04]  

Absolutely. But if we don't hear from people and we want to get a range of different voices and different people talking. Actually, this fits quite nicely with our first news article. I think we we both spotted I, I saw Tom Arbuthnot share it first. And this is about, I'm going to say it's Vassar. Vassar one as a lifelike audio driven talking faces generated in real time. So those who are watching along can see this. You can upload a single image, upload an audio clip, give it some control signals. I'm, I'm kind of assuming you can give it some what you want it to say as well and it will generate some really quite impressive clips.  

[Zoe 04:48]  

So, you know, sometimes nothing happens and sometimes everything happens all at once and you just kind of deal with it.  

[Kevin 04:55]  

So I won't play a lot of these different clips, but that is an AI generated face from a single image. It's learned from that audio to create something on there, which is kind of scary.  

[Zoe 05:08]  

Yeah, it's, I mean, it is very scary. And I was, I know I use this phrase to you earlier, Kevin, it is still a little bit uncanny valley. And when you watch some of these longer clicks there are you can just tell there's something not quite right in in kind of the way it's moving, but it's still high. It's still incredibly realistic.  

[Kevin 05:29]  

Hang on for if anyone is behind the times, as I am this morning, Uncanny valley, What? What's it? What does that mean?  

[Zoe 05:37]  

So I'm just finding the Wikipedia definition that I actually sent to you this morning. So the uncanny valley is a common unsettling feeling that people experience when androids are humanoid robots and audio visual simulations closely resemble humans in many respects, but aren't quite convincing the real estate. So it's something like this where it's hyper realistic, but you can you can just sense that there's something that's a little bit off still.  

[Kevin 06:04]  

Yeah, absolutely. And, and I think, you know, for people who've played computer games, some of those computer games have been a bit like that, but they haven't really been based on real people. And not with this ease of creating from a single image, a single bit of audio as well. So it'd be amazing for computer games that you could mock up a face and have it generate something from there. Would be really powerful.  

[Zoe 06:32]  

I was just thinking. I was literally just thinking the same thing. It would be awesome for computer games.  

[Kevin 06:38]  

So sorry, have you downloaded and tried this out yet?  

[Zoe 06:42]  

Unfortunately, not so this this has been published by Microsoft Research. And if you just Scroll down to the bottom of the page, Kevin, for those who are watching at the bottom of this page, we'll put the link in the show notes. Microsoft do talk about the the risks and responsible AI considerations. And as I'm sure you can imagine, there's a lot of potential for misuse with something like this. We started having a really interesting internal conversation about could you, could you personalize mandatory training by, you know, recording it and then giving it photos of famous people that people might relate to and, and have this recording delivered by Ryan Reynolds or Sakura or someone like that.  

[Kevin 07:26]  

I loved that idea that you could get, you know, you're really in depth compliance training done by Deadpool. Whether it would have that own personality too, I don't know.  

[Zoe 07:36]  

Yeah, So, so, so, and of course that that introduces all kinds of like copyright and, you know, personal identity infringement and stuff like that. Yeah, it's just so there's a lot of potential for misuse. It was a shame though, because I was hoping, like I, I've actually told this person that I was, I wanted to do this, but I was hoping that I could take a picture of Chris Hunticford and his voice and create a video where he talks about how SharePoint lists are the only data source that you should consider for power apps and just put that out into the wild.  

[Kevin 08:10]  

But I should say that you didn't because of ethical reasons.  

[Zoe 08:14]  

As well, Obviously so.  

[Kevin 08:15]  

That's that's what stopped you from.  

[Zoe 08:17]  

There. It would have been funny though.  

[Kevin 08:20]  

Yeah, it would for those watching the video. I'm going to play this clip in a SEC because this did entertain me. But it reminds me of what's the Netflix show where they talk about kind of crazy ideas and it's got that guy who used this Channel 4 show who used to news. Oh my gosh, Charlie Brooker white something wipe, isn't it? Back Mirror, Black Mirror. Thank you. Yes.  

[Zoe 08:50]  

Black Mirror and then you completely threw me off then.  

[Kevin 08:52]  

White Wipe was something White was his old show, but Black Mirror and there was one, I think it was the last series at the beginning where they they kind of had someone whose life was suddenly on TV and they heated it all from there. And they said, oh, it's OK, you signed up when you signed up for the the latest Netflix details, then you agreed to have this happen and, and things like that. And it's that kind of ethical things. And I love Black Mirror for exploring those. And this is almost the reality is not that far behind what's happening with air, which is terrifying. But let's get a bit of comedy. And I will play the sounds of this for those who are listening. If you could imagine the Mona Lisa singing this. And that's what the video is.  

[Zoe 09:35]  

Yo, I'm a paparazzi. I don't play no Yahtzee. I go my cameras up your crotch. See. I tell the truth from what I see and sell it to.  

[Kevin 09:44]  

Permanently. I do notice it talks there about the camera up a certain area. I'm hoping that we're still keeping in the clean podcast section for for Apple. It is terrifying. You can I can look about and I can imagine, you know, I'm not going to be able to see the Mona Lisa in the same way from that. We can see some of these other videos where they've taken it as kind of cartoon type or animated characters from there. So I think what scares me most about this is I love what Microsoft research have done in that they haven't made this available, but the reality is there's going to be others who have put this out there. You know, we've seen Joe Biden videos, Donald Trump, we've got various elections coming up. There is going to be so much misinformation that happens about this. So I hope that people like Twitter and Meta are keeping up to date with this.  

[Zoe 10:33]  

Yeah. I mean, it's, it's a, it's a real and present concern, isn't it? Yeah.  

[Kevin 10:39]  

It is absolutely from there.  

[Zoe 10:41]  

Yeah. OK. So so moving on then in other in other AI related news for all of you drama llamas who have got lots of WhatsApp chats, Meta have announced Llama three and there's a new Meta AI bot for WhatsApp now. Unfortunately, it's only available in limited countries. So let me find a list of countries. So we've got, I think, the United States.  

[Kevin 11:11]  

Canada and, and a lot of African countries, it was just very interesting. And, and I know that Whatsapp's a very popular tool there. So whether that's kind of driven the usage, oh, should I say this, I'm going to go there. Whether there's some belief that there's less kind of control over what can be put out there, I don't know. It's, it was a very interesting set of countries that not not many people expected. And I'm going to be honest, there was a little bit of the British ego that went, we were always first, whether we get the toys first to play with so that.  

[Zoe 11:46]  

Was yeah, yeah. So, so US, Australia, New Zealand, a lot of countries in Africa, we've got Pakistan, Jamaica, Canada and I, you know, I don't know what the data protection rules are like in all of those countries. I would have thought Canada might have had.  

[Kevin 12:05]  

Yeah, that's true. Actually New Zealand as well I thought would control. So yeah, maybe I'm completely off there and it's it's more about the usage and the availability of people using it so.  

[Zoe 12:19]  

Yeah, so, so, so it was actually added into a group that we were in this morning, wasn't it, Kevin, the the Meta AI bot? And my instant question was, you know, what happens with our data. If you're adding this into a chat, does that mean that everything in that chat from that point onwards is fair game.  

[Kevin 12:38]  

So you were being one of those charm alarms in in that group straight away.  

[Zoe 12:42]  

Weren't you? Yeah, yeah, yeah, I was. And, and I know it's probably an unrealistic concern considering that WhatsApp is owned by Meta anyway. And if they need to get in the chats, you know, for security reasons or anything like that, they will do. But at the same time, you know, if I'm having like off the book conversations on WhatsApp, you know, I've got a lot of group chats where I would, I wouldn't want anybody to see those conversations. And then people just randomly start adding AI bots into it. And I know a lot, but what's happened to?  

[Kevin 13:15]  

Taught a lot about being encrypted. And you know this, this has been the whole conversation around the UK government, the Scottish government's having to release there and they're having to go to the individuals and get their own backups because it's not available. So it's always felt like a safe place for things like that. And suddenly if people are dropping bots in there, as you say, is it still a safe place? Because that's that bot just appeared there and someone kind of at mentioned it from the US And what was it they were comparing the Metra AI and copilot wasn't it with the description?  

[Zoe 13:52]  

Yeah. So it's a few.  

[Kevin 13:53]  

Things there.  

[Zoe 13:54]  

Yeah. And I know for those of you who like us who are in lots of big group, lots of big group chats, you know, there's often a joke about needing a copilot for WhatsApp to help you catch up if you're in a meeting and you miss like 300 messages. And we did a little bit of digging and found that actually it will only apparently the meta AI will only record messages where you @it. And you can only do that if you even though you can only do that if you're in one of those countries where it's been deployed. So that would mean then that the rest of the conversation it hopefully can't see. And what that means is that actually it's basically a copilot in WhatsApp. It's a it's a copilot that's there for you to ask questions and it will go off to other sources rather than a copilot for what attack, which will help me catch up to date with 300 messages that I've missed.  

[Kevin 14:47]  

So I'm kind of suddenly connecting things back. So this this has no capability to go back and review that set of messages.  

[Zoe 14:58]  

No, it well, no, So so So what it says I because I, I had a different experience on on my desktop app to what I did on the mobile and I scrolled back up through the chat on the desktop app and could see that at the point where the bot was added, it says it says that Meta AI will only have access. Yeah. Only messages that mention at Meta AI are sent to Meta. Meta can't read any of the messages in this chat. Some may be inaccurate or inappropriate. Click to learn more. OK, let me click that.  

[Kevin 15:35]  

And, and I think from what I'm showing on the screen that basically you've kind of got a way of that the same way as you ask copilot a question, you can do that within a chat group. So, you know, I guess what worries me is occasionally in chat groups, people say things that aren't true. So what this could do is answer that question, whether that's correct or not. Gosh, this could only raise more debates and things on there I do not see. There's also that ability to to AI generate an image. I really hope they've got the ethical boundaries on that because yeah, there's going to be some fairly rude images based on many WhatsApp chat, None of mine. Mine are never like that.  

[Zoe 16:16]  

Yeah.  

[Kevin 16:17]  

Could be some interesting things if if that doesn't have the the right bit within there.  

[Zoe 16:22]  

Yeah, I mean, personally, I'm even having read it and, you know, seeing the the statement that it's only looking at the chats where it's mentioned I'm. I still don't really feel like Meta or Facebook is a company that I have a lot of trust in and I still don't really feel that having their AI bots in my group chats where there are lots of highly inappropriate conversations and jokes that are made and you know even.  

[Kevin 16:50]  

I think again, anything that happens, maybe UK government, that could be talked about as well.  

[Zoe 16:56]  

Yeah, so, so yeah, I'm not, I'm still not convinced. And I believe someone else was checking and it only uses, it doesn't use the latest version of Llama that was announced, does it? It's still using.  

[Kevin 17:09]  

The one, that's why.  

[Zoe 17:10]  

It's not the newest one, so.  

[Kevin 17:11]  

I don't think so, yeah. And the other one that you saw, I don't. Yeah, it was this link that you can remove details, but only from a chat. So it wasn't entirely it. It seemed to be that when you have asked it questions, obviously it's going to train the next version with that as well, but you can reset that, but you can't remove your data. There was also something in here and we'll put this link in the show notes how you can download information about you from chat with AI so you can see what it is saving within. Then you can download it, but there seems to be no way to request that to be removed and things on there. So it's good that they're sharing and being clear about what they are doing. I I'm still with you. I'm not that comfortable with what is there. Just quickly on Llama 3 because I think what's very interesting about that and I've seen a couple of posts this afternoon that I mentioned as well. But the new Lama three with I, I think it's these different types of model and the number of parameters involved in that. It's kind of ahead of some of the Mistral ones. The Llama three with the 70 was it 70? Llama 370 B is ahead of Gemini Pro 1.5 and the the Claude 3. I'm not going to pretend I'm an expert in each of these different models, but all I can see is it's way ahead of many of those. And this is one that you can download and use in your desktop. I've already seen actually someone download and build a Visual Studio codes add in. So you can almost have a GitHub copilot style experience, not quite the same, but that style experience using just what's on your machine. So you can work offline and have that access and know your data's going nowhere else on there, which I think is very interesting and certainly opens up a lot of possibilities. We're seeing these models. It feels like there's another jump in what's happening with these. We know that GPT, obviously GPT 5 is being talked about and we hopefully will see something happen from that soon. So it's very interesting seeing this happen and and what's going to come next from these as well.  

[Zoe 19:24]  

Yeah, I couldn't agree more. And even though we're predominantly in that Microsoft space, I think it's just it's really interesting watching what the industry as a whole is doing.  

[Kevin 19:34]  

Yeah, yeah, absolutely. Because, and also looking at these models, looking at these capabilities like the multi modal, because that starts to come to the next, the next session as well. Final bits that I wanted to talk about was a little bit about feedback on there. Not talking, you know, we, we make sure our audio's good here. I'm talking about providing feedback to Microsoft and submitting those ideas from there. And this all stems from, funny enough, a WhatsApp group. Conversation on there that was someone saying there's no longer a way to submit ideas to Microsoft. And I pointed out, yes, there is. There was the old user voice that was decommissioned. It's now the feedback portal.microsoft.com. And there's all sorts of things. We can see Teams in there, Microsoft to do Excel, and I'm going to quickly go through these, the Microsoft 365 app within there. But I did notice and I thought I remember which page it was on, but it seems to have changed the order there is Copilot for sales is on there and also copilot for service. Not got too much feedback on this or note if you are using those to provide the feedback. But I think it's really good and I really want to raise awareness for people to go back there, provide that feedback and add those things on there. And if there's something you don't see copilot for Microsoft 365, there may be suggest to Microsoft to that that should exist there as a place to provide feedback. I know Co partner for Microsoft 365 has that thumbs up, that thumbs down. You could provide that feedback but being able to say ideas what do I want it to do I think.  

[Zoe 21:14]  

What's missing?  

[Kevin 21:16]  

Yeah, I messaged you about this, Zoe. I've got copilot, Microsoft 365, my own tenants and my kids were asking today, how many days is it until we go to Disneyland? And I was like, do you know? I don't know, but I'm going to ask. copilot and copilot didn't know the answer. It's in my calendar. It's 40 days to go today apparently. But it didn't know the answer to that. Once I told it what date it was on, it could do that calculation, but it couldn't seem to find that far ahead. So it's that kind of feedback I'd love to be able to put into there as well.  

[Zoe 21:48]  

Yeah, I couldn't agree more. And like like you say, I think, you know, they've got the they've probably got a huge amount of feedback coming through those in app feedback options. But actually it's it's the other, it's the other stuff that we can't do in the in app feedback. Like you say, the idea is the things that are missing the cross app integrations and you know, the things that are just going to make it better as well. Brilliant. OK, So we've got other news since our last recording as well and that will all be in the show notes and for the things that are really.  

[Kevin 22:22]  

Sorry, not just the show notes, it's also in the LinkedIn newsletter. So hopefully you subscribe to our nice new LinkedIn newsletter. I'll make sure that the link to that newsletter is in the show notes, which will have all the show notes and the newsletter. All this is feeling like a big circle.  

[Zoe 22:38]  

Going on.  

[Kevin 22:39]  

There, Yeah, yeah, exactly. But yeah, do check that out.  

[Zoe 22:43]  

Yeah, brilliant. And with that, Kevin, I think it's time to go to our interview. So before we before we kick off with your interview with Michael, do you want to just give us like an overview of what you talked about?  

[Kevin 23:00]  

Yeah. So we, we kind of covered three key areas is it's what you need to think about for governance or as as Michael calls it, Copilot readiness to, to prepare your environment. It's very focused on Copilot for Microsoft 365 and you're, you're here at the end. I could have noted that particularly, but said a lot of the the themes that we talk about actually apply to to other things as well. So I'll let you get to that point when we hear it also talked about what orchestrate can do. I I'm not a big fan of pushing marketing type materials and I hope you don't feel like it's like that. You'll hear the story. I'll talk about why I got introduced to Michael because I loved what he said and then realised that he was building a product to do it. So I like what he's going in there. I hope you hear me challenge and you know, talk about what he doesn't do, what it does as well. So please, if you think it's too much on marketing push, let us know. We're not here just to push products. We're not getting paid by them. I just love hearing what Michael talks about. But then also at the end he talks about what what sort of AI they're doing within orchestra itself and how they're using that. And that was really, really interesting, some of the ideas and things they're looking at and, and in fact, they're not trying to jump onto a bandwagon. They're not just putting generative AI in for the sake of it. They've got some really interesting ones on that. So yeah, I think should we roll that and see what people think?  

[Zoe 24:22]  

That's you.  

[Kevin 24:23]  

Michael. So hello, I'm here with Michael. Michael, do you want to introduce yourself?  

[Michal 24:31]  

Hi everyone, my name is Michael Pisarek. I'm one of the founders and the CEO here at Orchestra Software based out of Vancouver, Canada. I always have to say I don't sound Canadian. I'm originally from Australia, but I've been here for actually over 15 years now, so I could, I could call myself a honorary Canadian. We are a governance adoption and automation platform for Microsoft 365, really helping customers in a number of different ways to take better advantage of Microsoft 365. But in terms of Copilot, obviously everyone's hot topic of the the day, we really help customers kind of understand their environment from a security and a governance posture perspective and help them clean up their environment as well. Whether that's retroactively applying things like sensitivity labels, looking at permissions, or even some of the provisioning capabilities that we have. So, you know, long term, you have all of those security measures in place. But yeah, super excited for this conversation today, Kevin. So let's get rocking and.  

[Kevin 25:37]  

Wrong. So I think it's probably I know for those who've been around my podcast history, we we spoke on great happy Princess previously on there and I think I shared the story of how I got connected with you and I think it's really good. One is listening to my big inspiration of the Microsoft Cloud show podcast and you're on there and as as some my fellow podcast host. No, I tend to listen to podcasts in the supermarket and the ones that really get my attention, I sit there and end up walking on going yes, yes, and and you're one you did with Andrew and CJ was exactly that because you talked about governance and I think you were just starting Augustry when you spoke there. But the kind of fundamental part was what you need to do with governance with Microsoft 365, what you need to plan for, how you engage people with it. And then you just subtly dropped in the end. Oh, by the way, I'm starting a company which is going to help with this. And I was like, I mean, speak to this man, this is amazing. I love this. And and so been following your journey with orchestrates since, since then, which is gosh, probably four or five years ago now I think.  

[Michal 26:43]  

That's not that long, I think. I think time has speed up with everything. No, it was almost like 3-3 years ago, funny enough now. So CJ is one of the one of our early stage investors in orchestrate and we're doing a webinar with Andrew Conroy next week. So. Oh.  

[Kevin 27:00]  

That's a lovely circle going on, yeah.  

[Michal 27:02]  

What's, what's old is, is new again. And I would also say, Kevin, that in the, the kind of theme of what's old is new is I would definitely say that Microsoft 365 governance has been rebranded Copilot readiness as well, bit more of a sexier name now and and definitely much more, I think appealing to organizations. But yeah, no, it's been, it's been a fun journey. I meant the last three years have seen obviously significant changes within the Microsoft 365 ecosystem, but really we're seeing still a lot of organizations struggle with those fun fundamentals. And it's not an easy thing to do technically. It's not an easy thing to do organizationally, but definitely over the last six months we've seen organizations really start to understand the impact that poor permissions and obviously over sharing and even the lack of training. That's one of the things we see, even the lack of training around using a new tool like Copilot and some of the impacts that they can have that can have.  

[Kevin 28:02]  

Yeah, absolutely. And and it's funny talking about that past. I was chatting with someone about copilot studio, how you can point at a website pointed at my blog and it reminded me I had a post from beginning of 2018 about what I see as the big topics for 2018 and one of those was governance is going to make a resurgence. So, you know, these conversations have been happening for a long time and I think people like yourself and many others have been trying to get organisations tidying up their data, looking about oversharing, doing the right thing with it. And exactly as you say, copilot, suddenly it's, it's almost, it's given them the carrot to, to sort out their environment versus the stick of, well, you should do this in case something happens. So it's kind of nice to be able to have those conversations here again with people.  

[Michal 28:51]  

Yeah. I mean, we see it on the orchestra side in a couple of different fronts. So I think traditionally what we saw was IT admins coming to us and saying our environment is a huge mess, our users can't find anything, you know, we need at all a third party solution to help to help manage this. Now, of course you can do it out-of-the-box, but as you know, trying to go through the various different admin centres and writing PowerShell scripts and everything else can be can, can be really, really difficult. But traditionally up to.  

[Kevin 29:22]  

And sorry just quickly on that and not just doing it once, keeping that up to date and with all the changes happening as well and.  

[Michal 29:28]  

Doing it at scale, I mean, that's one of the things we kind of laugh about. You know, it's quite easy to write a PowerShell scripts to go and it'll write 100 teams or SharePoint sites. It's a very different thing when you've got organizations that have 2053 hundred 500,000 groups and teams right about the pastor's group just starts to fail after, you know, at that scale, unless you put in a whole bunch of additional kind of logic in there. But as I was doing it traditionally, it was IT managers coming to us, right, saying, you know, we've got these issues and we're annoyed and maybe we're paying for additional storage and our end users are annoyed. So, you know, can we want to look at a third party tool, but then they have to go and sell that up to the executive and of course the executives like governance. This doesn't make any sense to us. We don't really care. So it's a little bit more of a harder sell. In the last six months in particular, we've seen that completely flip and what's happening in ours. A lot of CEOs and CI OS, they're hearing that AI is coming. They're hearing how much more effective it can make their organization. They're hearing it from Microsoft, they're hearing it from the outside world. You know, we have a lot of conversations with CEOs in particular who are saying like, we know we need to get ready for this because we see our competitors doing this and we see them becoming more effective. We see them also losing staff right as well, but keeping that same level of productivity. So we need to get copilot ready. And that message then goes down to the IT team. the IT team are like, great, what does that mean? And they come for a company like us. So that I think the cell is, is, is a little bit easier, at least from a budgetary perspective also around that. I think the second part is you've alluded to is, you know, copilot in particular makes things much more visible, which were a little bit harder, I think to conceptualize for a lot of organizations. And I think Copilot does it in two distinct ways. So the first one is around search and information retrieval, right? So this is not a new problem about people going sharing content and people have content, you know, access to a bunch of things that they never knew knew about. But really this concept of security by obscurity worked pretty well in kind of that search world because again, you know, search was good. But if you really wanted to drill down into search, you'd have to understand, you know, how search works and maybe those common, common terms. And it wasn't really as smart as something like, you know, a copilot for Microsoft 365 years where it's not only searching for the keywords that you enter in, but it's smart enough to know those relationships with the graph. So what knows that salary could be linked to something like a pay stub for instance and return that that information. So it's much easier now to find information that you didn't necessarily know that you have access to. So of course, the typical thing is I'll go ahead and I.  

[Kevin 32:13]  

Was going to say, but it also surfaces that chunk of information. So before you may have searched for a phrase and it would show you that document, but not necessarily the bit of document that was relevant. Now it can pull that bit out. So you're kind of highlighting that that tiny element which could be the important bit, which could be salary, could be someone's name, etcetera, etcetera.  

[Michal 32:35]  

Correct. I mean, that's semantic index. The whole magic of this is it understands those connections between salary and pay stub and who's working on that. So again, it can make those connections much quicker and provide you with a better set of results whether you have access or not. So again, at the moment, it's much easier to be able to find things that you didn't necessarily know that you have access to. So I think that's one of the things. The second thing that we're seeing a lot is around Copilot hallucination, right? Particularly from a generative point of view where someone is going in and they're saying, and we've seen this with a number of different customers where someone from, let's say Singapore goes in and wants to create AHR document or a, or a document relating to the benefits related to Singapore, for instance. And that HR person only has content relevant to them. But maybe they have a shared policy from someone within another region within Germany or the US, for example. And then Copilot knows what you're looking for and knows you have access to, but it generates something based off those documents, which is a completely different set of laws and rules and regulations as opposed to something within Singapore. So again, you get this hallucination happening. And again, we see that a lot with our customers. And it's nothing to do with Copilot. It's simply to do with the fact that, you know, users have over shared content and they don't and they have content, they have access to content, sorry that they don't know that they necessarily do. But again, Copilot is using all of that information. And again, it makes mistakes. It's not perfect obviously, but it can also be really difficult. Like if you get this beautifully generated document for you, a lot of people now are like, great, this information must be correct. And if you notice.  

[Kevin 34:19]  

A quick look, yeah, that seems about right.  

[Michal 34:21]  

Yeah. And if you notice down the bottom everywhere within Microsoft 365, it always says, right, make sure that you review this this kind of generated content for you. So I think that that's why we're seeing this huge interest around obviously copilot readiness. But you know, words that we never really heard before in the Microsoft Lexian have become really, really popular. Oversharing is a huge one, right? For, for instance, where we talk about over shared content, but really that's content or people having access to things that they shouldn't. And of course, if you've ever worked with SharePoint or Microsoft Teams, you know that's pretty easy to do, particularly because the the default options are so open, right? And lots of organizations don't lock that down from the very beginning.  

[Kevin 35:05]  

Yeah, absolutely. And I mean, it's something I see a lot is obviously as a client based organization, we work with a lot of projects where we'll be talking about policies and talking about how you should do things. And when I try and look about how we do things in our organization, I get an awful lot of content of how we recommend other people, which which is obviously exactly how we stick to what we do ourselves within. But occasionally they start slight nuances that come up from others. And it's it's having that understanding between the two. And I think, you know, you touch them over sharing there as being content that you shouldn't see. But I think there's also a relevance, isn't it, to content that shouldn't matter to you so much that maybe you can access, but it almost maybe shouldn't be all quiet. It needs to be lower priority. Are you seeing that the kind of governance story about being that? Because I know prior to Copilot a lot of it was getting that search relevance, getting the right information to the right people at the right time, and having goodwill managed content helps that happen as well.  

[Michal 36:11]  

Yeah, I think across the board I would say that people are terrible at managing their content in Microsoft 365I mean, I'm sure you see that as well. Now it could be a byproduct that they only come to us because they have challenges and maybe there's thousands of customers that manage their content really, really well. I, I would hazard as a guest, that's probably not not the case. But I think again, it's, it's basic governance and its basic content management. And particularly from the point of, you know, we have a lot of capabilities within orchestrate about basic life cycle management. You know, pretty common thing in terms of people create these teams and these SharePoint sites, they use them for a period of time. You know, there's lots of processes within organizations which are time bound. So whether it's, you know, responding to an RFP, responding to a project or working on a project, maybe even onboarding and an employee, which could be time bound. A lot of the time that content gets created and then it sits there essentially forever. So this idea that having a process to be able to go back and kind of clean that content up and like you said, Kevin, archive that content and make it a little bit lower priority. It's something, again, that a lot of organizations haven't necessarily done. And A, it's because it requires A considerable amount of effort, B. And let's be honest, you know, Microsoft hasn't really made this process that easy, right? And when you think about it from Microsoft's perspective, because I always talk to customers and like, you know, why do they make things so easy to be able to go and create these things? Well, you know Microsoft is in the business of you creating your content and getting your content into the Microsoft cloud with this little friction as possible and as fast as possible, right?  

[Kevin 37:51]  

So and if they didn't then people will go elsewhere where it is easy because they want things to be frictionless as well. People aren't there to be secure, sadly so.  

[Michal 38:01]  

Exactly. Yeah, that would go to Dropbox. And you know, some of these tools which are extremely easy, easy to do. But again, there needs to be some sort of mechanism, particularly even on the basic side, you know, if we have content that's been there for 10 to 15 years, doesn't really need to have to be there. You know, we see a lot of organizations as well at the moment running into storage challenges because they've got this massive content, lots of versions happening and nothing gets cleaned up as well. So I think that's on the the life cycle side or the back end, The front end as well was really organisations now really starting to plan around things like sensitivity, labels, proper security, you know, reviews and reviews. We're starting to find not just of content or of SharePoint sites, but things like review of users, user membership and also around guests, right? That's the other big thing that we're seeing is around guest management as well and having guest reviewed and Microsoft is releasing a whole range of additional tools and capabilities, unfortunately always part of additional licenses. But again, the Entre ID Governance license and there's a new SKU for that, which is Entre ID governance for guests, I think it's called. Some really great tools around managing the life cycle of those of those users as well.  

[Kevin 39:18]  

OK, sure. And but I'm guessing the guest side that doesn't hit copilot quite so much because if you're a guest, I'm assuming you're not going to be spending $30 per user per month on a guest to have access there. But it's part of the general governance.  

[Michal 39:33]  

It is. And you know, you might think it's not, but if you're a guest in most teams, you can create content, you can delete content, right? You can go and do a lot of these things and a lot of cases. So again, it's, it's, it's one of those things as well that I think has has to be considered. But yeah, they're not going to be. You're not going to give you Guests Copilot licenses unless you're a very, very wealthy organization or you really like your guests.  

[Kevin 40:01]  

Absolutely. So what what sort of things you does orchestry? Do you know, do you touch that data life cycle? You touch on that, that kind of guest management? That's that's the areas that's that you've helped people with to to kind of make it easier to do the right thing.  

[Michal 40:18]  

Yeah. So I think, I think the first thing is really helping people understand the content that they have holistically, not just for the amount of content that they have. So we've got some really nice features in terms of understanding inactivity and how old the content is. You know, how many documents are within SharePoint sites and groups and everything else, and also around around permissions. So there's lots of stuff coming to Orchestra fairly soon around being able to see what's over shared, where permissions are broken and everything else through there. So with most organisations, their starting point for Copilot readiness as well. You know, firstly, we've got a whole bunch of content that we have within the organization. Now, do we really want copilot to consider organization that's obviously redundant, outdated or trivial as part of its processing now? Probably not. And as you know, most organizations have got this kind of whole bunch of old crap, right that never kind of cleaned up. It's kind of sitting there and taking up taking up storage. So but.  

[Kevin 41:19]  

It's also the the definition of what old and trivial is because something could be 30 years old and still be absolutely relevant. You know, there's some great albums out there from 30 years ago, for example. But the problem is that people don't spend the time to kind of look at that and they go, I'm not sure we better keep it just in case it's, it's getting that behaviour as well, isn't?  

[Michal 41:42]  

It correct, yeah. And again, the way that we do it with Orchestra is lots of capabilities. For instance, to start a process that you can go and ask, let's say your workspace owners, teams and SharePoint sites, you've got this content, it hasn't been touched on. This team hasn't had any action for, you know, six months, a year, five years for for instance, there's 37 documents in here. None of them have been accessed for years. Do you want to still keep this content or not? And again, in a lot of organizations we're seeing a whole range of, of purging of content, just deleting that content also. But the other thing is a lot of organizations just deleting stuff just doesn't make any sense. I mean, we, we know we want that. So a lot of organizations as well use orchestry from an archiving perspective and we do some really neat things. For instance, you can archive a document, but remove all of the previous versions. So you can save on a ton of storage that way. So I think the first step is really understanding that environment, what you have permissions and kind of everything else at that kind of broad level. So I would say the container level there there Kevin. The next thing is really around over sharing and that's.  

[Kevin 42:47]  

Sorry, Michael, just just come back to archiving because I love some of the thing such an orchestra fanboy, which I am on there. I think that the kind of definition of what people think about archiving, as you say, there's a certain amount of storage, but there's also just removing the permissions and stopping most people seeing that content as well. It's kind of putting to an area where you know, it's an archive or even just maybe you've got read access but not write access. And having that taken that active decision about what level you want to do either as an owner of that data, as an organization, what is it? Is it because you're being driven by storage costs? Is it because you want to tidy up your data? Is it because you want to keep that information but make sure no one changes it? It's encouraging organisations to have the think about their different types of content and what they want to be able to do with that as well and making that decision. It's not just throwing it from one pile to another and look at my office going. I'm being a bit of hypocrite talking about that as well, but it is making that active decision about what you want to do and being tidy with it really.  

[Michal 43:55]  

And honestly, Kevin, I think you've seen this. The worst way that you can get users to archive content is if you give them one, if you want to give them two decisions, keep it or delete it, yes, then no one deletes any, right? It just people are not not going to. And again, with Microsoft 365 out-of-the-box, you know, there's some other capabilities now with Orchestra, just as a quick plug, we got a whole bunch of features around this and I'm going to talk about this because you you set me up for it. But again, all the things that you might want to want to do. So, you know, removing previous versions to save on storage, like you said, changing permissions, right? That ability to say, well, let's remove access for the members, maybe remove access for, for, for some of the owners. Definitely get the guests out of there, right? Guests don't need to have access to Teams and SharePoint sites. They haven't. But the biggest thing, and it's kind of funny because we had this functionality in there already. We had the option with an orchestra where if you archive the SharePoint site, you could remove it from search now as you know and now if you remove it from search, you also remove it from copilot processing. So us being the product company that wants to dive on every trend now change that label from just remove from copilot processing as as well, but that we find that.  

[Kevin 45:04]  

Ability, Michael, you're rename. You're just becoming Microsoft at this.  

[Michal 45:08]  

Time why not? But but even even those things we find and I don't have the stats, but but we know probably 99% of cases unless people are just like, I really don't need this stuff. Let's let's just delete it. That ability to get someone to say, yeah, I don't need this content. So let's get rid of the previous versions. Maybe remove it from search change permissions like you said, Kevin, make it read only and even change the name of that team or SharePoint site so people know that it's been archived. Super, super useful. And when you look at it from a Copilot perspective, you know, the more massive content that you have for Copilot to work on, the more of a chance it is that it could get it wrong, right? So again, you want to reduce that corpus of information that you have to the stuff that you know is going to be the most relevant, relevant to you. So again, that first step for a lot of companies that come to us is really that kind of what do we have and let's at least clean this up. And again, lots of advantages from, you know, a security perspective definitely from a storage perspective there there as well. But again, the smaller the haystack, the easier the needle is going to be defined. And that's really what you're trying to do with Copilot to give it the best chance of finding that that needle there. Yeah.  

[Kevin 46:18]  

And you've got that notion. I, I get forced by my family to watch those shows where people kind of declutter their houses and there's that kind of go and put it in storage, leave it there for a few months, work out if you still need it. Does it still bring joy to your life? And if it does, great, keep it. I think it's that same logic, isn't it put into that area, just it removes that, that kind of feeling that, that, that kind of behaviour, that habit around it, that, well, I need it just in case you've got it there, you haven't lost it. It's just taken a little bit further away. And then if you don't need it, great, then it can go and be sold off. And I think it's, it's that same mentality, isn't it? To break that habit that people have, that addiction that people have to keeping things.  

[Michal 47:00]  

It is, yeah. I mean, for for us, we have a feature called two stage archival, which we you might know about, but really the first stages, keep it where it is, get rid of the versions, get rid of all of it. And then you can set up orchestra to say, well, if this isn't touched one year, two year, three years, then you can actually go into that deletion process. But I I think as a general tip is, again, if you want to run a process of getting people to clean up their content, if their only option is just to delete all of it, they're just going to keep all of it, right. And that's where we see these steps. That has to be that middle, that middle ground there. Yeah, absolutely. So that so.  

[Kevin 47:37]  

Yeah, go on. So.  

[Michal 47:38]  

So kind of let's say we've gone through, we've deleted a whole bunch of bunch of stuff, we've gone engaged our content owners, they've made the the process around that. So that's great. The second part is really what I think is, is quite popular at the moment, this concept of oversharing, right? And we talk about it from an oversharing perspective, really it's it's permissions management, right? I mean, when you share something, you break permissions on it. And then people have access to things that they might not know that they have access to, for instance. And there's a lot of reasons that this happens and a lot of it is just because the technical complexity of SharePoint, you know, sharing things with individuals or with groups, the different link types, for instance, there. So again, the next step that a lot of organisations that you know, we see doing these Copilot engagements is again, understanding what's been shared, understanding how people are sharing as well. And there's lots of different ways to share content within Microsoft 365. Unfortunately, the default is the share with anyone link. So that's where I would say 90% of organizations are. And the problem with that shared with anyone link is again, it's not the best way to share. You're not sharing it with an individual, you're sharing it with anyone. So conceptually, if anyone gets access to that link, they'll be able to have access to that, to that content. Now, I'm going to mention this about Copilot in particular because this is really important to to know if you're using that shed with anyone link, it doesn't mean that every piece of content or every piece of content that was shared with anyone is automatically going to be available to everyone there. If that was the case, right, it would be a disaster for people because most people, even in Orchestrate, we were doing that because that was the default link type. If that was the case, I think Copilot would be in real trouble because people would just have access to a whole bunch of stuff that hasn't. So the way that it works is you have to have of actually access that piece of content with that shared with anyone link before Copilot is going to be included as part of any results for that individual. So that's so that's a good, that's a really, really good good thing. And I really wonder sometimes, Kevin, let me know what you think. I wonder if Microsoft had it the other way around. They're like, Oh yeah, just let you know if it's shared with anyone, let Copilot process it for anyone because I think they would have turned that around pretty quickly. Just probably looking at some of the data that we see from Mark customers. So again, trying to understand what that default link type is, change it from shared with anyone to something a little bit lower for for instance. And again, really trying to understand what's been shared within within your organization. Now this is a pretty complicated thing to do. SharePoint Premium does this quite well in terms of you can get the reports out and you can see what's shared with with when. I don't know if this is out for SharePoint Premium yet. But again, then you can start what's called, I think it's like a sharing links review or a site review, they're calling it, where you can send now a review to an owner and say owner of this particular SharePoint site. There's a whole bunch of link types there, right? Do you want to keep them or, or, or or remove them or not? So I'm not sure if that's available yet, but it's it's definitely coming.  

[Kevin 50:47]  

We do the same thing, very quiet because I can't remember what's been announced, what hasn't. So I'll let you get in trouble and then.  

[Michal 50:54]  

I think you're.  

[Kevin 50:56]  

Right. Yeah.  

[Michal 50:56]  

No, yeah, I remember Chris Mcnulty's session when they first announced all of this stuff. So there's definitely, there's, there's definitely, definitely happening. So again, kind of that understanding now at the moment, that can be a really cumbersome process as well because you can go to the reports, there's things been shared with anyone. There's not really an easy way to get that information out of the Microsoft Graph and you have to go and run PowerShell scripts or you go to the individual SharePoint site and you see all of the things shared with and it's all kind of problematic there as well. But that's the second thing that most organisations are are doing. We're building a ton of flexibility or a ton of functionality around orchestrates specifically for that. It is a really difficult thing to build because there's no graph API call of get all of the stuff that has broken permissions within the SharePoint site. You literally having to go through item by item, which can get very, very problematic, particularly for larger tenants. But we've got this under control. That's that's why that's why we're here to do to do the hard stuff there. But that's the other thing that we find is, is really understanding those kind of shared with anyone those broken links permissions. And again, when we speak to customers, usually they're like, Oh my God, why do we have this as the default link type? Let's change it and put it a level down, whether it's sharing people within an organization or share with individuals for, for, for, for example. So that's the second thing we find. And then the third thing is this kind of retroactive?  

[Kevin 52:25]  

Sorry, just got to come back a SEC because you bought up SharePoint Premium on there and I think that's a lot of people's initial one is let's go with Microsoft. Microsoft does something, let's do what they do. And certainly for me, one of the challenges that I've seen with SharePoint Premium, love it as a tool, so much capability, but it has that limit of 10,000 sites is, is that something, you know, you've talked a lot about scale on there, assuming that's something that orchestrate and I hate this. I hate being a marketing person because I'm pushing there, but I know that's something that's that we've talked about before that you've looked at scale as well.  

[Michal 53:01]  

Yeah, Yeah. I mean we definitely got over 1010. I would say the vast majority of our customers have more than 10,000 sites. And really Kevin, Microsoft will change that. I mean that it, it doesn't make any sense to to limit it. I would say one thing and this is going to sound really odd and I might get in trouble for it, but it is what it is. Probably the best favour that Microsoft have given third party vendors like us is with the introduction of something like SharePoint Premium. Because I think previously the general assumption was as long as you get to E5, you don't need anything else, you don't need any products, you don't need all this nonsense. We're going to be able to do it to do yourself. But you've noticed over the last six months there's a lot of these add-on premium products, right, that have come out, everything from SharePoint Premium, there's a Planner Premium now, God forbid, who would have thought there would have been a plan of a premium? And there's all these additional license SKU's as well there. So I think Microsoft has really started to admit that, you know, there's a great set of fantastic core capabilities that M365 provides with E3 or E5, but you need some of these additional licenses to manage some of these some of these pieces. So for us and for every product vendor there as well, you know, whether you're on the intranet side or on the government side or on the security side, it's really how we can leverage what Microsoft provides us and kind of build additional capabilities on top of that as well. But I think SharePoint premiums are great option. You know, with us, what we're trying to do with Orchestra is, is build more of a cohesive set of of capabilities. So you might not have to buy SharePoint Premium and Ontra ID and Teams Premium just for the collaboration reporting. You know, maybe you can look at a tool like orchestra or some of the other other of some of the many others to be able to get those those those.  

[Kevin 54:49]  

And that's another reason I love speaking to you because all the things you're saying here are good ideas on how you do governance Orchestra does. You don't have to do this if you want to go and write your own PowerShell script. And trust me, I have to look at oversharing and see how quickly that runs. You can definitely bring these things and take these ideas and look at them your yourselves on there and then come back and chat to other people once you kind of go, oh goodness, this is horrible.  

[Michal 55:17]  

Yes, and and a lot of customers that come to us to be honest with you, it's it's organisations where someone has tried to do something that in theory seems really, really easy. And it's like this is just this is just a nightmare, right. Whether it's provisioning, whether it's life cycle management, you know, whether it's again.  

[Kevin 55:34]  

Using the graph.  

[Michal 55:35]  

Just using the graph. So let's go about the third thing, because I think this is another one which is really, really interesting. So let's say we've gone through the process of reducing all of our redundant, outdated and trivial contacts. We've cleaned that up. Then we've started to understand the security landscape through shared links and everything else. The third thing that most organisations doing now is, is applying security at the container level or even at the document level, but mostly at the container level through sensitivity levels and sensitivity labels are, you know, become, I would say, Kevin, you probably agree with me, the thing or the way to manage access to content at that container level from a security perspective for Copilot, right? Lots of improvements and sensitivity labels, you hear Microsoft talking about them all of the time. Most of our customers, you know, they've hardly ever used them or they thought about using them. But now it's a process of understanding what sensitivity labels can, can, can do. And they get to a bunch around, you know, managing permissions around guests and default sharing and rights management and a whole bunch of good stuff, good stuff there. And then again going back and typically retroactively applying those sensitivity labels at least at a container based level, right?  

[Kevin 56:51]  

And, and, but even at that kind of lower level, being able to kind of tag a document, say there's something secret in here, top secret, highly confidential, and having that come through to Copilot because you get the icon that makes clear that this information may be confidential. It just helps the individual make a decision about what they're doing with that data. Plus that ability to actually restrict it from Copilot if you set specific tags within your labels to sort of say if it's tagged with this, I don't want copilot touching that. Which is great. Gives people a lot of control as well.  

[Michal 57:26]  

Correct, Yeah. And again, there's lots of great capabilities in Microsoft 365 around the application of sensitivity labels now, now as well, particularly in the E5 license. So in the E5 license, you're going to get that ability to set a default sensitivity label for a document library. So if someone goes in and someone adds a document there, you can have that default sensitivity label set set through there as well. So again, lots of, lots of work around that in order to, you know, let's go back and apply the sensitivity labels, at least at the site level, like you said, Kevin, at the document level as well. Super, super useful. But most of our customers do it at the site level first and then do it at the document level later. And then also apply that to a, to a provisioning process. Because I think the, the concept of copilot readiness is not a one and done thing, right? It's not a project. It really is a, a, a, a program or a strategy that you have to adopt. So again, it's OK with cleaning all of this stuff up. And of course it never happens because things keep changing and people keep creating stuff. So particularly from a provisioning perspective as well, it's making sure that you have that plan in place. And for a lot of organizations, they start off with provisioning because they know that the cleanup is going to take so long. They're like, listen, at least we can come up with a provisioning strategy that if someone requests A-Team or a SharePoint site, there is a sensitivity label applied that's going to, you know, if it's marking something like Top secret for for, for instance, and you're using something like SharePoint Premium, maybe you can use that restricted, what's it called? Restricted site access feature which again only locks down the sharing to people.  

[Kevin 59:07]  

Site search You mean or no?  

[Michal 59:08]  

No, no, there's a, there's, there's another, there's another thing in SharePoint Premium. It's actually really, really powerful. It's the ability that you can tag a SharePoint site and you can't share that content outside of anyone within that group. So again, as Kevin mentioned, if you've got stuff that's top secret, you know, create a label, add that feature on a when it's only on SharePoint, SharePoint Premium. But again, that's going to give you the best idea. And then like you mentioned as well, Kevin, you can even do things like, you know, choose to remove that whole site from copilot processing whatsoever. The problem with that is that as well. So which is I think really, really problematic. Basically yes.  

[Kevin 59:53]  

Yeah, I'm not a big fan. Well a it's called RSS, which just makes me think, oh I can get blogs from it. So that confused me for a start. But I don't like it that it's, it's almost, it's the same attitude that we had of let's just turn off Delve, let's stop people doing it because it looks like a problem. Let's not sort out the underlying issue. Let's just turn it off. And so if you are going to use that restricted search, at least have a plan of how you've got to tidy things up so you can disable that again later.  

[Michal 01:00:23]  

Yeah, it's very bizarre for me. I, I, I agree. And you can only have a maximum of 100 sites, Like who chooses 100? And then what happens if you got 50,000 sites? Like 100 is not that many, but I think it, it, it might be, I might be a minimal viable product maybe for, for now, you know, Microsoft, Microsoft's new way is released this stuff and we'll make it better and better over over time. I'm sure that's going to that's going to improve over time.  

[Kevin 01:00:50]  

I hope so. I hope so, yeah. So the the other one I want to talk about is obviously you know, we're talking about preparing for copilot to enable that magic that AI and specifically Gen. AI brings. But if you're, if you're dealing with a lot of these things, I know governance to a lot of people is specific rules, but there must be some, some vagaries. So are you looking at either AI or generative AI and with you know to for orchestrator use itself as a kind of tool?  

[Michal 01:01:21]  

We are, and I think we, we talked about this before. So I don't want to be the type of product company that jumps on every single bandwagon that runs on, right. So everyone's an AI company now. All of a sudden, just like three years ago, everyone was a Web three company, for instance, right? So.  

[Kevin 01:01:40]  

So you're not going to have mesh avatars or anything like that. We're.  

[Michal 01:01:43]  

Not we're not building a mesh plug in for now unfortunately I think that ship has sailed, but we are. So we definitely are looking at it from the perspective of a cup of a couple of things. So generative AI for Orchestrate, probably not that useful. I mean, there's applications there in order that someone can just type in and say, you know, I'm looking for all of the guests that haven't logged in in the last 60 days and maybe have access to content that's tagged with a particular label. The problem with that is, is that you're assuming that users know the questions that they want to ask. And in a lot of cases, most users don't, right? That's why they buy a product like Orchestra, because we can tell them what they need to know without them knowing it. So I think there's definitely potential there, but we're not looking at it too closely there. copilot plugins for us is a pretty interesting one also. So the ability that again, someone can do some natural language stuff through through Orchestrate where we're really viewing Orchestrate from the AI perspective within the product itself is much more I think from the machine learning kind of perspective, right? So helping us understand trends of collaboration and security within an organization and presenting that to to users. So one of the things that I would love to do, every organization comes to us and say, like we've got 5010 thousand, 30,000 users, right? We know that a bunch of these teams and SharePoint sites, they perform really well. People are engaged, they're having conversations there, they're not having conversations and 1:00 to 1:00 chat. People understand versioning. God forbid someone's added some metadata to a document library like all of the all of those things. So one of the things we are looking at orchestra is can we use some machine learning and some AI to be able to tell you as a customer that these are the teams or the SharePoint sites that perform the best and these are the commonalities between them. These are maybe the owners or the members or maybe how they're structured because ultimately when people.  

[Kevin 01:03:43]  

Come back that, make that work, that sort of thing.  

[Michal 01:03:47]  

Correct. People want to understand whether an organization, what makes a good collaborative team, what are the commonalities? Because if you can find them, you can build off that, right? And a lot of it is who's in the team. A lot of it could be maybe how things are named. So again, using AI and some machine learning to try to understand those pieces. That's that's probably the first thing. The second thing is around templating and, and business processes also. So Orchestri started off as a, as a, as a provisioning tool. We've moved far past that, but in a lot of organizations, if we can run some AI and some analysis to say, based on the 510, fifty, 500,000 teams and SharePoint sites you have, we've identified these four main types of templates. You know, you're doing projects, you're doing departments, you're responding to RFPs and maybe you're doing, doing something else. Well, imagine if by magic we can come back to them and say, here are the four main types of templates that that, that that you have. Here's the structure that we've seen work that work the best and use that as a starting point for other, other pieces. And then the third one is really around Anomaly, the detection. This to me is the most interesting one because again, it's can we look at the patterns that users are doing around collaboration and start to determine anomalies in there? So a really good example is, and I know this is terrible, but I think it's really good from a kind of training perspective. Imagine if we could identify that we have a user and instead of, let's say, sharing a document with people, they always use that shared with anyone, right? And they're always using that. And we see a sudden influx of that usage all the time. Remember how great could be even from a training perspective, if we can go in and we can say, hey, we've noticed that you're doing this. Here's some training material to help you understand why you shouldn't use the shed with anyone link that you should use some something else as well. So that's a normally.  

[Kevin 01:05:40]  

Nudge. That isn't it. It's kind of trying to encourage behaviours by nudging people towards the right thing rather than just so saying stop doing this, stop doing this. It's oh, you might want to try this here, It's a good idea.  

[Michal 01:05:53]  

Exactly. Yeah. Or obviously it's on the other side, which is more of the security side where, you know, a really common thing is you might have a guest who comes in and starts deleting a bunch of content, right. And again, in a lot of organization, you might not not might not know that. Or you might have a guest that hasn't logged in for two to three years.  

[Kevin 01:06:11]  

Hang on Michael, you, you were saying a minute ago that people shouldn't should be deleting content, not holding onto it. This sounds perfect.  

[Michal 01:06:17]  

It does. It does, yeah. So, you know, I think again, for us, we, we want to make sure that we understand the technology fully. And I, you know, for us, we're definitely investing into, into AI. But everyone is saying that and I think people are just talking a whole bunch of nonsense. I mean, we view orchestra as a tool to get you ready for AI, but for us, I think it's the most interesting part for me, as well as how can we really glean insides into how organizations are using Microsoft 365 and help them understand that better from an adoption and also from a, from a security perspective. Also, like, I would love to be able to find out, you know, you know, Kevin, there's always one or two idiots within an organization. Like they don't know what they're what they're doing, right? So they're either breaking permissions or they're creating staff or they're adding lists somewhere where they, where they, where they, where they shouldn't, You know, that ability for us to kind of find those people or maybe find the actions and kind of nudge them in the right path. I think that's a really good example of using machine learning and AI to kind of pour commonalities out or see anomalies within an organization. And Kevin has frozen. Oh, and Kevin has left. So there we go, Kevin will be back. Everyone. In the meantime, since this is still recording, I hope everyone is doing well. It is here, Vancouver. It is 7:51 AM. It is a pretty miserable day here in Vancouver, but have you ever been to Vancouver? You'll understand why. So let's just wait until Kevin comes back. All right, everyone, with Kevin's untimely demise here, thank you for joining myself and Kevin for this week's copilot collection, Copilot Connection. Sorry. Hopefully you learn something. If you are interested in getting in touch with me, you're getting in touch with orchestry. Follow me on LinkedIn or follow me on Twitter. But let's all hope that Kevin comes back to life here. Oh, Kevin, you're back. Hello. I was. I was keeping the audience entertained. Why? Why you had left?  

[Kevin 01:10:15]  

Thank you very much for that copilot for sorting out your Wi-Fi and your kids coming home from school would be very, very appreciated right now. I think so, sorry for that.  

[Michal 01:10:26]  

You, you, you know, and hopefully this gets in into the show. You know what the killer feature of Copilot is? And I'm shocked no one's developed it. It's taking people automatically off mute if they're off mute in a Teams meeting. Like surely it should be smart enough. Now I know people are going to say, well, what happens if you're screaming at your kids or you're swearing or you're, you're chewing, chewing lunch? That's why we have AI, baby. AI is going to understand your kind of usage patterns and all of that sort of stuff.  

[Kevin 01:10:53]  

Usage patterns. That would be nice, yeah. I want to be able to mail you ****** idiots and that sort of thing without it's kind of coming off that that would be my nerves around it and and it will probably mute too many other things I said so without it realising.  

[Michal 01:11:08]  

The idea though, the stats we really need from Microsoft, there's not that we're creating 20 petabytes of data a day. It's how many times in all the Teams meetings across the world someone is saying you're on mute, that that's the data.  

[Kevin 01:11:22]  

We love that. I would love that. And can you see my screen? Yeah, no, that, that would be good. I think even Cosmos DB doesn't stretch to that scale. That's the the problem.  

[Michal 01:11:33]  

But no, I think.  

[Kevin 01:11:33]  

That would be great.  

[Michal 01:11:34]  

Billions a day basically of people of people saying that.  

[Kevin 01:11:39]  

So a hugely appreciate you kind of covering some things there and I'll go back and make sure that that's all made sense and do a little bit of trimming forward put it out. But thank you for that one. One thing I would say is a lot of what we talked about, obviously we're focusing on copilot for Microsoft 365, but a lot of this logic will fit in with the other Co pilots tidying up your environments for power platform and looking at the governance there, looking at Dynamics and even SAP making sure you have the suitable permissions, you're not over sharing within that as well. So a lot of the logic should apply whatever the copilot is within there. So it's it's worth thinking about these different things and bringing those into each, each and every copilot within securities. A great example where generally you'll have people there who have access to everything, but not always. You might want to kind of segregate those different details. Fancy. That might be a question for Ruth Campbell. I didn't ask about permissions on that. But no, no thank you. And I hope, I hope that's got people thinking a bit about their their governance and sorry, their copilot readiness.  

[Michal 01:12:43]  

And then one more thing, I know we're running out of time here. I think this is this is the fourth thing. And I really want to emphasize that this as well. I think the biggest barrier to Copilot at the moment for organizations is the lack of training that people are giving them. And I can really speak to my example because I thought with quite like I would put into copilot create me a document around best practices for governance and then that's it. Click a button.  

[Kevin 01:13:07]  

And the you.  

[Michal 01:13:09]  

Know it would come out all right. I mean, it's a great starting, starting point, but understanding how prompt engineering works and putting in, you know, now instead of my prompts being a couple of lines, it's like half a paragraph because.  

[Kevin 01:13:22]  

You can tell a bit of story, don't you? Yeah.  

[Michal 01:13:26]  

Around Microsoft governance best practices, assuming that I choose.  

[Kevin 01:13:31]  

It for.  

[Michal 01:13:32]  

Right. You know, keep the tone light and conversational instead of formal. Make sure that you cover these points and include a couple of images that gives you a completely different result. And I think one of the things that we've, you know, probably one of the things I think for users is really there has to be some training around that. We found that internally, I'm sure you found that as well. You know, when we're pretty skilled people within Microsoft 365, you know, we have our own Copilot tip channel and it started off where it used to be 1 liners and now it's these kind of more paragraph of things which is so much more powerful. So not just show me the tasks that I have outstanding is show me the tasks that I have outstanding grouped by by day. Focus on the tasks which are high priority and give me all of the communication details for everyone that I need to contact as part of those tasks. Then you get the magic of what copilot is and I think a lot of people don't understand that. So again, from a training perspective, make sure you educate your users on that. And honestly, I don't think there's that much good training around copilot prompting out there because it's so new. Hopefully that's going to change here in the next three to six months as well.  

[Kevin 01:14:45]  

I'm trying to say too much, but obviously there are some organizations, some large Microsoft partners who can absolutely help with that, but that's a topic for another time. What I would pick up on what you said there about creating that team that you have internally build that community that I think is the most powerful thing to kind of share those good ideas because those good ideas are not the same good ideas for everyone. They can be within your area, whether that's an organization such as orchestry, whether that's your finance departments, your people who like cats within your groups on there. Those different ones will have those little niches that work really well together. And if you can harvest those communities effectively, that's where you get the really interesting stuff out. But no, absolutely agree on that, that training and getting people to understand those capabilities. Completely agree. Well, thank you very much. Really, really, really looking forward to going and listening to those last five minutes as as well as some points. But really enjoy, always have a good chat with you, Michael and really enjoyed that one today. So I will hand back to myself and Zoe as we wrap up. But yeah, any any last words from you?  

[Michal 01:15:55]  

No, thanks for having me on Kevin. I think everyone appreciates what what you know you folks are doing with this and and helping organizations with Copilot. So yeah, thank you for for having me on.  

[Kevin 01:16:06]  

Actually, final one, where can people find you? Are you out and about? Any conferences?  

[Michal 01:16:12]  

We need, yeah. So we are going to the M365 conference. Is that what it's called in Orlando in a couple of weeks? Orlando. So the orchestra or some of the orchestra team will be there with a, with a booth. But yeah, if you want to get in touch with me, feel free to get in touch with the either through LinkedIn, Twitter, or just go to the orchestra website, soorchestra.com. Again, if you're interested in, in, in, in having a look, Kevin knows we get a bunch of smart people and very, very nice people as well. So just get in touch with us and we'll we'll show you what we have available.  

[Kevin 01:16:43]  

Yeah, I'm saying and and do get in contact. That's How I Met Michael and I think it shows how much I've enjoyed it as well. So thank you very much everyone. And Kevin and Zoe, back to you.  

[Zoe 01:16:56]  

Thanks, Kevin. That was such a great interview with Michael. I'm really good to hear about what orchestra have been doing, what they're seeing with their customers and and how they can help I guess as well.  

[Kevin 01:17:06]  

Yeah, yeah, absolutely. And and again, things for people to think about for what they need to do, whether you use orchestry, whether you go your own route and decide to rewrite some of those scripts. Yeah, don't go with other third parties. There's so much work to do there. It's it's not worth it. But I hope he's got people thinking there as well.  

[Zoe 01:17:26]  

Brilliant. Well, I think we're almost at the end, aren't we? It's been another really great episode. There's just so much news coming out at the moment. Hopefully we're helping you keep up with some of that. Just a reminder, I know we mentioned this at the start, but I call for speakers for month of Copilot is open now. Please, please, please submit. We want to hear from you. We want to get your sessions. We want to broadcast them through the month of June and just help spread the word about all things Copilot.  

[Kevin 01:17:55]  

Absolutely. And I'd also mention that we'll both be at the collab summits obviously collab Summit EU in Wiesbaden in May. So do come and say hello. I'm sure we'll be wearing our copilot connection tops at some point there. So come and say hello if you see us there. Fantastic. But yeah, otherwise if you want to hear more, please, please, please smash that subscribe button, go into YouTube, subscribe to us there, put us on the podcast, follow us on LinkedIn, Twitter, Stroke X on there. And even if you fancy, just leave a little review on there because it all helps us get a bit more awareness from that. And you know, mention it, mention it to your friends, mention it to your colleagues, mention it to your clients, even your family. Spread that word and get people talking about Copilot.  

[Zoe 01:18:47]  

Thanks very much for listening. Look forward to talking more soon.  

[Kevin 01:18:51]  

Thanks a lot. Bye bye.