Navigating AI Evolution: Insights on Successful Implementation, Missteps, and Future Trends

Mark Smith: 0:01
Welcome to the Ecosystem Show. We're thrilled to have you with us here. We challenge traditional mindsets and explore innovative approaches to maximizing the value of your software estate. We don't expect you to agree with everything. Challenge us, share your thoughts and let's grow together. Now let's dive in. It's showtime Showtime, Well, welcome everybody. We're back for season two of the Cloud Lighthouse podcast sponsor, which is the Ecosystem Show. We're excited to be back. Everyone's been partying, having a great summer. Unfortunately for me, when you guys are having summer, I'm having winter. Winter in New Zealand is like summer in London, so it's not too bad.

Andrew Welch: 0:45
So it doesn't exist, is what I'm hearing.

Mark Smith: 0:48
Yeah exactly, exactly. It's been a great year. Anyhow, what have you guys been up to?

Andrew Welch: 0:54
William has been doing some exciting things.

William Dorrington: 0:59
They've been so exciting that I can't remember what I've done, because it's just been a lot of going to festivals. I went to Siget in Budapest, which is, you know, first time I've been to a festival of that size. It's 450,000 people go through it in the six days, so it's double the size of Glastonbury and it was insane, insane. But the thing that really struck me and this is showing my age was I didn't have to once queue for the toilet or queue for a drink, and that's what I was more impressed by than anything else, that is, that is actually.

Chris Huntingford: 1:32
That's pretty damn incredible. I'm not gonna lie.

William Dorrington: 1:34
It's impressive.

Chris Huntingford: 1:35
Right the logistics behind sorting that out yeah, because when you're in the uk it's like they purposely make you queue, because I think we just love a queue. It's a hobby. Oh, the brits. The brits love a queue it's like oh shit, there's a queue.

William Dorrington: 1:47
Yeah, and this is what I'm saying. I'm not, I'm not happy about it. I was annoyed by the lack of queues yeah, I'm sorry.

Chris Huntingford: 1:53
Yeah, I was wondering that. That's sort of where I was going with this question. It's like, were you, were you angry that you didn't have to stand in a line and wait for something? Fuming?

Andrew Welch: 2:01
fuming, I would say that there's nothing more british than queuing, if it weren't for, um, you know, being genuinely unhappy about things, that should probably make you happy, that's you know yeah, yeah, that's true hey, you know, chris and anna and I all choose to live in your country, william, that's true.

Chris Huntingford: 2:29
Yes, because we want to be here.

William Dorrington: 2:30
I appreciate you calling it my country as well.

Andrew Welch: 2:35
Well, it's not Mark's country.

Mark Smith: 2:37
Definitely not mine.

Andrew Welch: 2:38
Hey, you claimed it, I was colonized from your country.

Mark Smith: 2:42
Oh, come on, yeah, you did yeah.

William Dorrington: 2:45
Right. Go on, chris. Go yeah Right. Come on, chris. What have you been up to? Come on.

Andrew Welch: 2:51
And we're back for season two.

Mark Smith: 2:53
Let's start the new season strong and of course there's been some. You know we're into a new FY with Microsoft FY25. So they've made all their announcements. I hear pretty strongly that Microsoft is no longer a software company. They are an AI company. Yes, and that is pretty much how they are thinking about themselves internally as an AI company. So I assume that means everybody's scorecard is going to have AI in it, absolutely. And I think what are we up to? 80-something different AI applications coming out of Microsoft.

Andrew Welch: 3:29
I think there's probably 101 co-pilots at last count. I believe. Wow, yes, okay.

Mark Smith: 3:34
So my question is this If you're a seller inside Microsoft, how do you go to the customer and talk about AI when you have got such a conflict in understanding? What are you talking about?

Chris Huntingford: 3:53
Yeah, this is actually a problem, right, because I'm experiencing this now. So somebody came to do I was being interviewed by some sort of a magazine or press company or whatever, and they're like, oh, he's very important, no, I'm not important. And they're like, oh, what is the first thing you need to do when you start thinking about adopting Copilot or releasing Copilot? And I'm like, well, you need to be extremely specific about which Copilot and what part of AI, and I think there's a lot of miscommunication about what AI, what co-pilot is, what specific tools we're looking at. And I'll give you the example. If you're looking at, you know, preparing for AI like that's a huge landscape. That's like that's massive. There's just so many things in there. And I feel a little bit bad sometimes for the Microsoft sellers, because you know they need to get skilled up on all of these new, different things, right, which can be tough, and they have to do it extremely fast, I feel like there's a lot of confusion about what it actually is.

Andrew Welch: 4:51
I am happy, though, that we've seemed to have moved away from. You know, if you go back to this time last year, you heard a lot of oh, this Microsoft specialist is now going to be the co-pilot specialist, which, of course, begs the question. So that means they're going to be an expert in how many dozen products. At the time, that's how many co-pilots we had. So I do. I am heartened by the notion that, at least in most of the geographies that I know, I do see some moving away from this idea that co-pilot is a product rather than. You know, chris, you said this in a call that I heard you on earlier today, which you know. You said what's the use case for Excel? Where you don't think about the use case for Excel? Excel is everywhere, and Copilot is everywhere too, and it seems like we are kind of getting the message there. Yeah, it is true.

Chris Huntingford: 5:40
It is true.

Mark Smith: 5:41
You know, when you look at M365, m365, if you look at what is the fundamental cell story behind M365, is productivity, right? Everyone talks about productivity in that group of applications. In business applications, we talk a lot about building applications or solutions that enable or empower the business that might not be commercially available off the shelf. So you build a solution out on it. How do you pitch AI in this world of now and Will and I were having a discussion off air there is such a level of noise, not just for Microsoft, but globally around AI. There's an element of snake oil being sold with it, right? So every company that's been around for 15 years is all of a sudden. You go to their landing pages and you'd think they'd been born in the cloud type thing or born in AI, right, it seems like they're all AI experts. All of a sudden you were like we just know you're not being in the game long enough to be AI experts.

Andrew Welch: 6:54
So I had a conversation with. I had a conversation with an organization a while back that they had gone so far as to rename their company and I'm not going to say who it is, but so that they would have the word AI in the name of the company. And I asked them, I asked point blank well, what are you doing with AI? And the response from one of their senior C-level leaders to me was oh well, we're doing a lot of work with RPA. We're doing a lot of work with rpa, wow. So here's this company that has gone to the legal trouble to rename itself with ai in the name, but doesn't actually know what the thing it's now named after is, which, which killed me that disturbs me.

Chris Huntingford: 7:39
I actually put a post I mean I put a post out on linkedin saying the same thing. I'm like everyone thinks that they're an AI expert, but like actually, what is? I mean, we've got power platform people kicking about saying, hey, I'm a co-pilot expert. I'm like cool, so can you elaborate on things like data security and purview and that type of stuff? And they're like what's a purview?

Andrew Welch: 7:59
Now I believe that one of was it you, chris, though, who said in one of your recent conference presentations who here is a Power Platform person, and then you said congratulations, you are now an AI extensibility expert.

Chris Huntingford: 8:12
Extensibility. But, not an AI expert, just an extensibility expert, correct. So you got me on a technicality.

Mark Smith: 8:21
Chris.

Andrew Welch: 8:21
I always listen to you.

Mark Smith: 8:23
So here's the thing, though, with this right. So companies now are coming to partners, they're coming to Microsoft and they're going we need AI Right. And you're like okay, so what's the business problem? What do you mean? We need AI right. And it's kind of like how do you take? Because obviously they're going to invest in this, but how do you deconstruct that and go and take them on a journey of really uncovering how AI is going to apply in the context of that organization and really move them from this buzzword you know that we have into something that's practical, that is really going to move the dial for that organization with a client, an organization, maybe about a month ago, and ostensibly we were there assembled to discuss their power platform strategy.

Andrew Welch: 9:55
We spent at least the first four hours of an eight-hour session and we barely talked about power platform at all. We actually talked about their data platform, and one of the things that that I have observed is that oftentimes we go into a room thinking that we're going to talk about about X, and we end up talking about why? Not because we were, we were mistaken in thinking that the topic was X, but but that all of these disciplines and all these technologies now overlay on one another and they require one another in order to really scale and be effective. So you know, one of the things that I find is that when you want to build an AI strategy and you actually want to get the most out of AI within your organization, maybe talk about something else organization maybe talk about something else. Maybe talk about data or talk about your appestate or talk about digital literacy, because AI is not a separate mountain off in the distance that we need to. We should be treating separately.

William Dorrington: 10:57
It's usually the last parliament. If you look at you, you do your foundations to your data, your solutions and then your intelligence, because you need all that quality data flowing through the system to get that. So just to jump in here a bit, I think when people talk about AI and talk about true AI, what they really want us to go in and actually look at that data fabric they've got within their business already. So Mark and I was having this conversation offline. A lot of people still have that dispersed, siloed and segregated data around their entirety of their technology estate. So, bringing that together and then, once you've got that sorted, it's then looking at exactly what Chris always talks about, which is the governance behind it, the classification, the tagging, knowing what data you have, knowing the sensitivity of it, knowing where it's stored, your target data architecture and your current data architecture and the gap between that. And then, once you've got that, I think the other thing that's overlooked, that we don't always talk about, is actually what quality data specialists do you have client side, because it's all well and good partners going in and providing data engineers and data scientists, but actually the projects that are most successful are when the data specialists already exist on the client side. They can work closely and understand and accelerate that growth already, and then it's OK.

William Dorrington: 12:03
Now you've got all that. You've got that policy data, you've got it governed, you've got people who know where it is and how to actually integrate more data into it. What is it you want AI for? And don't tell me it's because you tried it out and you got a recipe for a brownie in the style of Snoop Dogg wrapping, because we're not going to send that out to your clients. So that's when we start understanding your requirements understand your business objectives, your targets that you're trying to get to, your weaknesses, your strengths and how we then can start aligning tech to be better through adopt, extend and build your own. So I went on a full. Thanks for attending my TED Talk, guys.

Ana Welch: 12:34
I would say I would say that, even an advocate for that, and you're on the right path towards a successful problem-solving implementation. So your guy on the client side doesn't necessarily need to be, like 100% a data expert. Absolutely. They need to understand that data is crucial, because the reason why people don't say, oh, actually we need to fix our data is because it's not flash, it's really hard to look at and also because people don't understand what that means. And that wasn't very clear to me until very recently the fact that when you say the word data, people don't understand what it means.

Ana Welch: 13:21
Like just this week I was with a customer who were looking to fix or to repurpose or to re-implement a very, very clever Excel spreadsheet. Like I couldn't stop by going back over and over and over again and asking you know where's this data coming from? Where's that data coming from? And they were almost annoyed with me. They were like we're not here, like the data problem will be fixed by this other team. We need to fix this Excel spreadsheet because it's a risk. So people don't even understand that that Excel spreadsheet is their data.

William Dorrington: 14:02
It's their data exactly.

Ana Welch: 14:05
Exactly, they don't, because for them it's the app, it's the functionality, it's incredible.

Chris Huntingford: 14:12
Yeah, I have the same problem. Right, like we were. I've been speaking to an organization recently about you know, oh, they're looking at copilot-libbing their data. So I'm like, oh, cool, cool, cool. You know like that sounds awesome, what data. And they're looking at copiloting their data. So I'm like, oh, cool, cool, cool, you know like that sounds awesome, what data. And they're like, oh, no, no, we're going to be doing like this data source, that data source, and I'm like, okay, but you are aware that tools like Microsoft 365, copilot, are actually focused on, you know, using mainly unstructured data. Right, like that's where it looks. Structured data, right, like that's where it looks.

Chris Huntingford: 14:43
So people think data and they think like a database, right, but actually unstructured data is really the problem that you've got to worry about here, because you know, organizations without the right retention policies, without the right rules, without the right data loss prevention, without the right file labeling and things like that, unstructured data is far easier to proliferate across the business and cause hallucination with AI. So, like for me, when I think about data, I've had to educate myself backwards and say, no, chris, it's not just structured data inside a database. Actually, semi-structured data, unstructured data, interaction, dates, all that type of thing lives across your ecosystem. I mean, look at the graph, look at all the information that you get from the graph. Right, that's still data. So when you munch those layers of data together, what happens? And I believe personally that this is the first time, I think, in history in tech really we've been able to take unstructured data, structured data and that interactive type graph data and weld them together.

Ana Welch: 15:40
Yep.

Chris Huntingford: 15:40
And that's the power.

Ana Welch: 15:42
That is super interesting what you're saying, and especially the interaction data. I never even thought about that and in theory, this is what we do. Right, that's insane, and Andrew and I even have this problem. When we explain things, it's often that people will say, oh, that's too vague. People want to know exactly how to do things like it's concepts are so complex and remote that they. That's why they're saying we want to do AI, because they can easily get on like Chachapiti and ask it something and it does some. They want to know how to do something. They're not interested in concepts. Concepts are hard.

Chris Huntingford: 16:31
Yeah, that's half the issue, right, because everyone, can I tell you the thing that I hate the most right now when I talk to, I'm going to use Copilot. Right, because actually Copilot is a really good example of AI that works and it's visible. And so people say, chris, what's the use case? I'm like will you stop talking to me about use cases Like what's the use case for Copilot? I'm like dudes, the use case for Copilot is not what you think. Like you're thinking about merging a document. I'm like no, look at your scenario, step back. Even Microsoft rebranded the website to scenarios because actually driving things like ROI against the use case is stupid. It's just stupid.

Chris Huntingford: 17:09
Okay, so, looking at the scenario and stepping back, actually where will artificial intelligence augment the scenario and how will it make it better? Okay, and actually that's how you do it. Right, and I'm glad you brought that up, because what you will start seeing with the agentification of Copilot, with GPT-5, you will step back because agents will remove people from the equation in a lot of aspects. Right, they will become observers. So when you step back and you plug in that agent framework into like a job role, you say where can I use AI in this scenario, and I think that's what's gearing up being. That's what we're being geared up for now agents. We're going to look at many types of data, not just a type of data. So that's what I've been learning. It's wild.

William Dorrington: 17:53
Okay, I'm done on my rants no, it's good we're making turns to do monologues here and I'm enjoying it dude, it's just.

Chris Huntingford: 18:01
You know what it is, man, it's like I get very, I get very passionate about this shit because I'm seeing it live in front of me with companies, and I'm seeing the mistakes and I'm making mistakes, yeah, so I'm learning I think agents is interesting in that we're hearing a lot of talk about agents coming and it's that whole thing.

Mark Smith: 18:20
How far down that path or how far away are we from things? Because November agents I see is is is going to have massive impact. But then I'm also like man, there's so there's so much still need to be, it still needs doing before we can get to that. You know level, but then you know some of the more recent stuff I'm reading in AI is that some of the speed that some of the stuff's potentially coming at is going to just really people are going to go shit. What just happened happen well, it's funny.

Andrew Welch: 18:53
It's funny. You mentioned that, uh, chris and I uh locked ourselves away yesterday working on a, on a, a bit of a project that we won't say too much about, but when it comes to organizations getting ready for ai, we will, uh we will, have more to share in las vegas in about a month's time.

Ana Welch: 19:15
That's my teaser.

Mark Smith: 19:18
Yeah, are you saying that's the next conference season starting?

Andrew Welch: 19:22
Yes, conferences, the new conference season. Well, you know, really it was. The new conference season was waiting on the season two of the Ecosystem Show to get underway.

Ana Welch: 19:36
I thought you were going to say it was waiting on the new season. Of love is blind.

Andrew Welch: 19:41
Oh, oh, I will. I mean love is blind. Uk premiered recently and Anna and I have been saving it, but I think tonight might be the night.

Ana Welch: 19:49
We cannot wait. Like we think, it should be premiered at the same time.

Mark Smith: 19:54
Listen, let's be clear, it's not that great.

Chris Huntingford: 19:57
Have you watched it? Mark has watched it.

William Dorrington: 20:02
You watched Love Island as well.

Mark Smith: 20:05
I saw the US version and everyone was like the UK version is so much better. It's the same thing with different accents. There's no difference. It's crap.

Andrew Welch: 20:15
Chris was really into Love, is Blind, south Africarica it's about true love.

Ana Welch: 20:20
Everything in life it's about true love.

Mark Smith: 20:21
It's not about true love, it's about television and entertainment it's awful.

William Dorrington: 20:27
I'm shocked. You guys watch that. It's just horrible, it's cringy william, there is.

Andrew Welch: 20:33
There's just something about consuming like thoughtless trashy television while drinking a bottle of champagne.

William Dorrington: 20:41
Exactly no no, no. Okay. So if you're introducing champagne, invite me around. I'm more than happy to join on that one. Now. That changes the game a bit.

Ana Welch: 20:50
Yeah. So when he said season, it was either Love is Blind or Bridget Jones.

Chris Huntingford: 20:55
for me You're combining like two of the worst things in the world champagne and Love is Blind, or Bridget Turned for me. You're combining two of the worst things in the world champagne and Love is Blind. Chris, come on.

Mark Smith: 21:03
What's the best champagne you've had, Chris?

Chris Huntingford: 21:05
Mate, I don't even know, but I know it was terrible.

Andrew Welch: 21:08
Out of a shoe. Good day to you, sir. I'm just taking jams.

Chris Huntingford: 21:13
I'm trying to get Anna to throw something.

Andrew Welch: 21:15
Okay, alright, all right. So we've got a new season of the show. We've got a of whichever show it is that you watch, though I presume that most folks are most excited about the new season of the ecosystem show. We've got a new conference season. We have a new year with Microsoft. What are the? I want to know from around the room here, what are the one or a couple things that you are most excited about or that you are most intrigued by in Microsoft land coming up?

Mark Smith: 21:50
I'll tell you what I'm most frustrated about.

Andrew Welch: 21:52
Most frustrated.

Mark Smith: 21:54
Yeah, we're going to put on at the Vegas conference, right, the Power Platform conference.

Andrew Welch: 21:58
Mark is super pissed about this.

Mark Smith: 22:01
We were going to put on a happy hour where we could invite friends et cetera to it. You know we expected to get about 100 to 200 friends to our happy hour event on one of the evenings.

Andrew Welch: 22:14
It's funny how many more friends you have when you're throwing a happy hour.

Mark Smith: 22:25
Yeah. And so the MGM Grand said well, we have to check with the conference if they're allowing you to have this side session. I'm like I just want to book a room, want to book a DJ, want to book some hors d'oeuvres, some drinks. It's a side, it's not on the. You know, the conference organizers came back and hard denied us really and so mgm grand said our hands are tied, we can't give you the facility that's wild.

Andrew Welch: 22:44
I'm always shocked at how difficult people make it to to give them money blows my mind.

Mark Smith: 22:51
But what am I excited about? Tell you I'm excited about ai, and and why I'm excited about AI.

Andrew Welch: 22:57
We need more Mark, we need more, we need more Details.

Mark Smith: 23:01
I have been prompting probably over 100 prompts a day now for the last two to three weeks and I tell you what I thought I was good at prompting three months ago and now I'm like Mark's hitting the gym for his prompts.

William Dorrington: 23:14
He's prompting so hard. Man, how much do you prompt Mark's hitting the gym for his prompts?

Mark Smith: 23:18
He's prompting so hard. Man, how much do you prompt my prompting game? My prompts are essays long in length. That's the starting prompt I have. And see, andrew, I noticed you were like what the? And that's the thing is, man, when you really get into prompting and really understand how this conversation can work with AI, it's a lot more than just a single line question backwards and forwards.

Mark Smith: 23:41
The intelligence you can get out is off the Richter scale, amazing. And when you so what I'm obviously doing is a rag type process with that, that's just amazing. The results that you can. That is possible.

Andrew Welch: 23:55
Hey, you know, no judgment Mark from one verbose, overly verbose fellow to another. Knock yourself out with your essay. Long prompts.

Mark Smith: 24:04
Yeah, it's amazing, I've got a prompt library now.

Chris Huntingford: 24:08
That is very cool. I'm actually quite jealous. That's nerdy.

Ana Welch: 24:11
That's amazing. And do you know what's amazing? What's amazing is that you guys do this, and then you're wondering why all organizations just want to use AI, and they don't know what for? Because it's fun.

Chris Huntingford: 24:25
It is yeah.

Ana Welch: 24:26
Because it's the funnest thing technology has ever ever had and everybody's got access to it. Of course, they don't care about their unstructured data.

Chris Huntingford: 24:38
Can I tell you? I actually agree with you, but can I tell you something? That's awesome. Okay, I like pen testing organizations with shit data. I love it because the crap that I find and I show them like I had. I had a bunch of C-suite people on a call today and I'm on Copilot and I'm like I want to show you all something. So I started asking Copilot about their salaries and stuff their faces. When I started doing it, they were freaking the stuff I know how to find just because I have access to AI. Let me tell you something I'm becoming a rock solid red teamer with this stuff and I love it. I go to companies. I'm like, cool, please let me into your tenant so I can just go and red team your co-pilot and show you what a clusterfuck your entire company is.

Mark Smith: 25:18
Chris, would you classify yourself now as a chaos engineer? Oh, hell, yeah so you know that chaos engineering is a whole specific category. Right, there is. Yeah, it's a thing. There's a tool in Azure for it. Yeah, netflix have come out with a tool. Yeah, spotify were the first ones to in Azure for it.

Ana Welch: 25:36
Yeah, netflix have come out with a tool. Yeah, spotify were the first ones to come up with it.

Chris Huntingford: 25:40
Get out. That's a real thing.

Mark Smith: 25:42
It came up this week in a side conversation around the Power Platform is what chaos engineering tools could you use with it, which is you know? I'll give you a real simple one yes, chaos engineering is how does the platform survive when you introduce an unbelievable high level of latency into the mix? This is wild. You're wanting to basically create cases that stress the system so much that you want to see if it's going to fail or not. So failure is another kind of category, but chaos engineering is an actual category.

Chris Huntingford: 26:16
This is going on LinkedIn.

Mark Smith: 26:17
And people are very serious into that space.

Andrew Welch: 26:20
Yeah, this is like when I was, when I was on small boats in the coast guard and if we were doing you know, some sort of exercise or if, say, a man overboard drill, because we were out and we were bored, and then during the drill I would sort of discreetly push something else into the water just to screw with my crew, just create chaos all around them.

Mark Smith: 26:41
Yes, yeah, and it's, you know, distress to see how things will survive, how you know how robust you know the tech is. But I think the whole read because and I associated Chris with what you're saying with red teaming I think that there needs to be a lot more education around what red teaming is, how it works, what its function is and how any strategy an organization has in ai needs to include the red teaming element to it like and that's not just tools but it's actually specific methodology around red teaming I think, that you've lost.

Ana Welch: 27:15
Chris about five minutes ago.

Andrew Welch: 27:17
Yeah, chris is out. He's just saying, yes, chaos Engineering is his name.

Chris Huntingford: 27:23
I think he's just like his guy. I'm listening, but I'm quite legitimately doing research of Chaos Engineering and I love it. Thank you for that.

Andrew Welch: 27:34
Okay, so Anna, your turn. What are you most excited about or most intrigued by?

Ana Welch: 27:40
I mean, I don't know if I'm like super excited or intrigued about anything, starting with regards to the new year, because, wait, because not that much has changed, but what has changed is good. So the one thing that I do like is that finally they don't segregate at least part platform and dynamics as much. Finally those targets have come together a bit more, and you know, and that's a big deal, that truly says that you can work, you know, towards the best solution for a customer. And that's a big deal because, no matter how well-intentioned the folk at Microsoft were, you know, and how much they wanted to like empower organizations and drive operational efficiencies, they would never put their best work where they weren't incentivized right where their targets weren't met. So by doing that, I think Microsoft has done a good thing, but not enough of a good thing. I will be really excited when I will see like a little Azure quota with these things as well.

Chris Huntingford: 29:00
Yeah, yeah, no, you're right, you're right.

Ana Welch: 29:01
And that's where things are really going to start, you know, to spin up, and that's where people are really going to be empowered to find the best solutions. I think that there's some interesting, there's some signs of life on on this front, right Like yeah, 17 years later, I feel like we've been dragging our feet for at least five years, like it's exhausting, honestly.

Andrew Welch: 29:27
Well, I just I've seen such a cooling of the citizen development fervor around power platform over the last, you know, I would say, six to 12 months, and that's not to say that that's not that citizen development is not still a compelling part of the value proposition.

Andrew Welch: 29:47
But I see some really promising signs. Mark talks about anchor apps and I've now adopted this word, but I've seen some really promising signs of Microsoft significantly reorienting its messaging around business applications and Power Platform, specifically around some of the themes of Power Platform, specifically around some of the themes of Power Platform as an enterprise app dev platform, power Platform is part of the data platform and of the data fabric within an organization. So, you know, I don't know if this year is the year that we're going to go all the way on this. I mean certainly the five of us on the show. We're going all the way way with this, but we've been doing that all along. But I do see some promising signs that that um, you know, the thinking around this is is is starting to change um wholesale mr dorrington, what are your thoughts?

William Dorrington: 30:39
I agree with everything that's been said so far, including anna saying there's just nothing to look forward to. Um but um on top of that now that's now.

Andrew Welch: 30:48
That is, that is our, our, our British co-host for you right there.

Mark Smith: 30:53
Thank you Will.

Andrew Welch: 30:54
Thank you Will for living up to your cultural stereotype.

William Dorrington: 30:57
You are most welcome, but one thing that I don't think we talk about enough is that of process mining. We talk about finding clear objectives, we talk about finding inefficiencies and going in. We talk about finding clear objectives, we talk about finding inefficiencies and going in, but I still think process mining, especially within the realm of Microsoft partners, is so underserved that no one's really cracked it yet to look at actually how, discover those inefficiencies in organizational-wide processes, actually understand how they're really using their systems, their ERPs, their CRMs, their back office, side office, front office, and then looking to plug in.

Andrew Welch: 31:30
Nobody cares about ERP, Will Nobody cares. Nobody cares about ERP.

William Dorrington: 31:34
Yeah, well, you know, there's someone out there that does it used to be me, but even I don't anymore but I think plugging those gaps RPA, dpa, ai, et cetera I think is going to be something that we really do need to crack better, because I'm seeing an absolute surge of process mining companies in uh, in the wild, and I'm getting a lot of clients using it and you know it's something that I'm going to be focusing energy on. And the other thing, and just to do a small shout out, it's just the scottish summit is coming up. Absolutely love that event. It's just around the corner in aberdeen. I'm really looking forward to getting back out there and exploring that event as well. So I think they're the two key things for me in the near future, mark.

Mark Smith: 32:13
Yeah, there's an offer he gave mine at the start. Right, it's written.

Andrew Welch: 32:17
Oh yeah, Mark already did.

Mark Smith: 32:19
Chris, what are you most excited about?

Chris Huntingford: 32:21
I'm most excited about Chaos Engineering Right now.

Chris Huntingford: 32:24
Chaos Engineering At this moment, yeah, right now. No, can I tell you what I'm really excited about? Okay, so I'm excited about agents and I'm excited about responsible AI. Okay, and I'll tell you why Because I think that responsible AI is going to rear its head very, very quickly in the next few months. And let me explain. Think about an agent, right? So what does an agent do? So an agent effectively doesn't replace the human. It gives them the skills to be more observing of what the process does. So, if you have agent A and we're going to call it agent Anna and agent Andrew, so which one of us is agent?

Andrew Welch: 33:00
A yeah, agent Anna. Agent Andrew.

Chris Huntingford: 33:03
No A1 and A2. Anyway, so it doesn't matter, but either way, you're swapping information constantly, right? So, like you know, you're talking to each other. Blah, blah, blah, blah, blah. What happens when Agent Andrew is in a shit mood and causes Agent Anna to be in a shit mood.

Andrew Welch: 33:23
In practical terms, this would be reversed.

Chris Huntingford: 33:24
I'm trying to not get something thrown at me.

Andrew Welch: 33:26
Okay, anyway, anyway, I am a jolly fellow, yeah, anyway.

Chris Huntingford: 33:31
So think about this if your agent uses and what is an agent? Agent uses um, ai and automation to drive the completion of a scenario, okay, that's what an agent is. So if you build the agent incorrectly, okay, and those agents are passing data off between one another, you have a problem. So the other thing that we're not used to doing is we are not used to testing non-deterministic tools. Okay, we are used to testing deterministic tools, so we know the outcome. Like Power Automate is an example, we know the outcome.

Chris Huntingford: 34:03
Gpt generative pre-trained transformers are non-deterministic. So how do you know that your agent is going to do the right thing? That's a problem. So if you have these non-deterministic agents running around your business doing things and they could potentially I would say I call it agent Armageddon infect one another, okay, and you don't have the right observability rules inside your organization from a governance and control perspective, you're actually causing a much bigger problem. Think about things like insider risk management, data exfiltration, proliferation of bad data in your business. You now have a bigger problem, okay, so what will happen is you have to create a thing called a deployment safety board. All right, now, I'm not going to speak more about that here, because actually there's a lot of NDA stuff in there, but that's the only tip that I'm going to give. And the Deployment Safety Board allows you to deploy AI solutions in a safe manner that have been red-teamed and tested. And if you look at the European AI legislation, andrew, how does the European Union make all its money?

Andrew Welch: 35:03
By begging countries for more and suing people. Oh and do they sue people.

Chris Huntingford: 35:08
Well, yeah, so Microsoft had been sued by them before. So if you don't meet those, standards.

Chris Huntingford: 35:13
the companies are not people, chris, sorry, you know what I mean Anyway this is not America, okay, yeah, yeah, true, anyway, I'm going to rant, right. So what will happen? Is that AI governance board correctly? Okay, the fine is 30 million euros, or 7.5 of your base revenue for the year. Okay, so if you don't do this correctly, that's what will happen, right? So now what I'm thinking is no one knows what a deployment safety board is. No one knows how to build agents, no one knows how to test agents okay, and no one understands the output of what will happen if you have agent Armageddon in your business, which is why I like this chaos engineering concept, because I think that if you pen test those agents correctly and you read team correctly and you create the right transparency notes, actually you could have something very powerful on your hands, and that, to me, is the next thing.

Andrew Welch: 36:03
And I think a little bit of a plug here, little little bit of a plug here One I think that this is something that partners that Microsoft partners are terrifically far behind with. If I project out, in this time next year, and maybe even sooner, there's going to be terrific demand for partners that have some degree of expertise here. Don't know why I just got that emoji, thank you but partners that have some degree of expertise here Don't know why I just got that emoji, thank you but partners that have some degree of expertise here and partners that have actually been proactive now in putting together the capability to go help organizations stand these up and do this and do this effectively. So that's something. And here's the shameless plug where Cloud Lighthouse is now working with partners to build that capability and to build those practices, which is something that I'm really excited about over the next 12 months or so.

Mark Smith: 37:00
So we are at time. So I'll just probably take the wrap-out sequence here. Number one Cloud Lighthouse did release their labs capability this last week and that means if you are in a position that you need an idea brought to reality but you don't want to go to a traditional partner, you want to sweat test an idea and validate it back to the business without any forward motion. After that point, check out the lab's offering.

Mark Smith: 37:30
The second thing I just want to bring to your attention Andrew and Anna have written an amazing course around ecosystems and how you really can get into ecosystem development. It's about a seven-module course that will be released in the next six weeks publicly as a self-module course that will be released in the next six weeks publicly as a self-paced training course. So we intend to release that at Vegas, that conference, so that'll be there for you. So great opportunity as we educate the community. We're looking forward to the season ahead. There's so much to cover and discuss. Hopefully you've all gone out and got your five strawberries this week and are excited about the implications of eating those and what that's going to mean for us all going forward.

Andrew Welch: 38:18
This is just. I love. This is how Mark gets results out of his business partners. Is that he announces on a podcast that something is going to be ready in six weeks? There it is.

Mark Smith: 38:28
That's it. Buy yourself some motivation, right? Buy yourself some commitment.

Andrew Welch: 38:35
Yes, that's what I lack, Mark.

Mark Smith: 38:39
Thank you. Send us your questions. We're keen to get feedback from you. Do we need to start a WhatsApp group for the ecosystems where anybody can ask their question and we just check it out? Maybe we'll set that up in the week ahead and we will have a WhatsApp segment which is questions that you want us to answer.

William Dorrington: 38:59
It's what we need, Mark. It's what we need in our life is another WhatsApp group, More WhatsApp groups.

Andrew Welch: 39:05
Mark is just throwing down the gauntlet all over the place here. Deadlines and WhatsApp groups.

William Dorrington: 39:11
Teams, channels. Let's get them all in. Ciao Later, bye guys Later.

Mark Smith: 39:16
Thanks for tuning into the Ecosystem Show. We hope you found today's discussion insightful and thought-provoking, and maybe you had a laugh or two. Remember your feedback and challenges help us all grow, so don't hesitate to share your perspective. Stay connected with us for more innovative ideas and strategies to enhance your software estate. Until next time, keep pushing the boundaries and creating value. See you on the next episode.