The BCN Podcast

AI Trends for 2025

• BCN

Share episode

2024 was a pivotal year for artificial intelligence, and as 2025 unfolds, AI’s impact on businesses and daily life is accelerating. In this episode of the BCN podcast, Fraser Dear and Andy James explore the latest trends, demystify AI concepts, and discuss its transformative potential.

From generative AI reshaping workflows to the rise of multimodal models enhancing versatility, AI is enabling creativity, efficiency, and predictive capabilities across industries. Microsoft’s push for agentic AI and personalised systems highlights the growing integration of smart technologies into everyday operations.

Tune in to discover practical strategies for AI adoption, explore funding opportunities, and learn how businesses can use AI to streamline processes and unlock innovation. Whether you're a tech enthusiast or business leader, this episode offers valuable insights to adopt AI and get the most out of it.

Peter Filitz:

Hello and welcome to the BCN podcast for 2025. This is our first episode and I have today the pleasure of Fraser Dear and Andy James joining me. Today, we'll be discussing where 2024 went in terms of AI evolution and what we saw, and we're going to discuss what is in store for 2025. We want to understand more about what AI is, what and where it's going, and understand a little more in terms of how businesses like yourselves can really take advantage of this emerging technology. The purpose of this podcast is really to drive more awareness in terms of what we're seeing in the AI space, engage with businesses to understand their business use cases and have a look at how the market is changing on that front. So, without further ado, I'd like to introduce, I guess, starting with you, fraser. Fraser is our Head of AI and Data Innovation at BCN. Over to you, fraser. Why don't you tell us a little bit about yourself?

Fraser Dear:

Hi everyone. Yeah, my name is Fraser. I look after our AI and data innovation here at BCN. I've worn many hats over my professional career, from laser physics and big data sets all the way through to, more recently, heading up a power platform team and also business central. So lots and lots of business processes and lots and lots of insight into how we get the best out of our data. Obviously, now with heading up to AI and data innovation, it really is focusing around getting value from investment into businesses so we can see that return of investment for AI.

Peter Filitz:

Excellent, Fraser. Thank you so much for joining us and an old favorite, Andy James, welcome back. Good to see you, Andy. Do you want to give us a quick introduction?

Andy James:

Happily, I look after our productivity solutions team so we make solutions based on the Microsoft platform and the Power platform to make tools to smooth over business processes, so that's apps and automations and chatbots. We're working closely with Fraser and wider across the other BCN teams to try and include AI, to try and remove administrative burden, speed tasks up and just make life easier for people.

Peter Filitz:

So I guess a good place to start is really dispelling the hype and the myths around AI and really bringing it back to earth, so to speak, around AI and really bringing it back to earth, so to speak. So I guess one for you probably.

Fraser Dear:

Fraser, what do we mean when we say AI? It's a great question, peter, because you only have to go onto the internet. Ai is everywhere. Ai is inside everything, and if it doesn't have AI in it, it's obviously not kind of current or up to date. But at BCNcn we believe and we define ai as the capability of a machine to imitate intelligent human behavior, whatever that behavior may be. So it's important to think that through, and what we don't mean is like machine intelligence being sort of somehow cognizant or aware. It's not about having kind of like the ability for a machine just to be completely autonomous and go off and do its own thing. No, no, no, no. It's trying to remove and I think, as Andy's just said, remove those kind of more mundane elements so that we can release people's time to do more value-added elements. And I think as we go through this conversation today, we will talk through about what we see in 2025. And autonomy is something that is going to come through. We believe in this year, but it's all very much coming out now.

Fraser Dear:

When we think about AI, it involves that kind of use of algorithms and models to kind of perform tasks that do require human intelligence, like visual perception or speech recognition, things like language translation and decision making.

Fraser Dear:

Traditionally, those are the sorts of things that you would sort of ask a specialist can you translate this? More recently, we've got kind of like, oh, we could type that into google and it'll give us a translation, but the reality is, we can now build ai that's able to do that on the fly. Ai is integrated across so many different platforms and services, and so, more often than not, it is true when they say this is AI enabled. But have a look at one of our other podcasts around. Not all AI is equal, because this is where we want to start to draw out these lines. You know, something in Power Platform is AI, something in Cognitive Services is AI and something in Azure is also AI, and actually one of the challenges actually one of the challenges and one of the opportunities we've got is actually to stitch all these things together to give a business a solution that actually is embedded in multiple different platforms communicating between different data sets.

Peter Filitz:

That's a really great explanation and I think that just you know sort of brings it back to reality. Shall say that what we're talking about is leveraging these evolutionary services to drive more intelligence and efficiency into the more mundane manual processes that businesses have in place today. As you say, giving people back more time to focus on the more important and strategic elements of their job, which so many businesses really need, right. So I guess that really brings us on to the next topic, to set the scene for what's to come in the coming year. What were the trends that we saw in 2024 in the space?

Andy James:

So the buzzwords through 2024 was all about GPT, which is generating content, and again that sounds quite a single specific thing. I'm generating content, but there was a wealth of opportunity for what that could mean. So we had co-pilot in every single conversation that I think we had for the last six months. We were talking primarily to clients about chat, GPT and how that kind of fit into this change of mindset as to a kind of AI adoption for them. So the generative AI, generating content, was everything from. Here's what I need to do. Give me some ideas, help me think of the best way to do this. What are my options? What's some information? And allowing ai to search for a load of information, review a load of information or documents. You gave it for it to generate some content. Other might be here's all of my documents and reports and conversations I've had and I want you to generate an email or a social media posting campaign and all this different stuff.

Andy James:

So it was all around using AI to generate content to try and make life a bit easier. Give us a first draft and we'll come on to that. That will be like a broken record, but giving us that first draft of information that I, as the professional can then review, decide if I'm happy with, if I need to do something more with it, enhance, change or whatever it might be, because that's still my job, my role. But the AI is there to support me, speed things along and make life a bit easier With all of the work and all of the kind of enhancements and developments going on with AI.

Andy James:

The other kind of thing that we started to see was around large language models, because GPT and all of this runs on large language models. But then there were smaller language models as well, which have been really good to see, kind of Microsoft again investing time and effort to try and reduce the amount of kind of resources and consumption that's needed to run these models so that you can have smaller models that are cheaper to host and run but still give us the benefit. So it's looking at how even the AI can be more effective and efficient in what it does. So that was kind of what we saw over 2024.

Peter Filitz:

Yeah, that's really interesting and I guess I know that's been a big concern from an environmental perspective is obviously the compute needed to support a lot of this AI utilization globally. It is a concern, but it's great to see that Microsoft is making steps in the right direction to help optimize usage around that and obviously again scale the models based on what their function and need is, which is great. Fraser, I know you've got some thoughts on that as well. Why don't you share what you saw in 2024 and what really we need to?

Fraser Dear:

take note of. Building on top of what Andy's kind of said there, I think what seemed to happen then towards the latter part of the year was around kind of like multimodal AI models. So when we talk about models and, andy, you mentioned GPT there are thousands of models out there, all kind of tailored and focused and tweaked, with different sets of different volumes, different parameters, and actually they're usually targeted around slightly different use cases. Now, with multimodal AI models, what we then started to see was actually we're bringing perhaps like text and image together for a model to decipher what it means and then perhaps create an audio output, or take audio and text and then create an image, or these sorts of models are able to operate on different levels, using different modes of input to yield different modes of output, and actually that was quite a big step forward in the kind of model generation piece. And I think the other words that kind of springs to mind is democratization, because in the early part of 2024 when we talked around generative AI, it was big business doing big scary things with these proprietary models that were all a bit kind of what is this strange thing called an LLM? How do we access it, how do we use it and within the year we've actually, for example, ai being embedded in things like science and medical.

Fraser Dear:

So in that sector there's so much innovation happening with AI embedded in it. So if you think about, like, predictive analytics for disease prevention, you know we can now, with lots and lots and lots of data, identify predictions for what the outcome may be for a patient. So AI has been instrumental in shifting that focus from diagnose and treat to predict and prevent, because actually we've got that machine learning ability to overlay on top of the data sets and then even on a more basic level, if you think about kind of image processing. So if you've ever been for an x-ray or if you've had a CT scan or an MRI, it always has to go to that clinical person to start to review every single element and they can all now be processed by AI and are being processed by AI. So we've gone from big, scary new thing to actually embedded in our businesses and our organizations, adding value within that one calendar year. And to me that's the big trend it's the speed of new research, ability, availability, embedding actually into business solutions.

Peter Filitz:

Yeah, and not only just business solutions. I mean even in personal lives and solutions. I mean, I'm a big advocate of co-pilot, personal as well. It's made significant changes to me personally, which is great to see and, as you say, I think the speed at which it's evolving and changing and being fundamentally embedded in pretty much everything we're seeing is exciting, but also scary, I think, for some people and businesses out there, and this is why I think it's again bringing home the key message that AI doesn't have to be scary or daunting. It is accessible now to everyone and every business, from any shape or size. So it's here to stay and you need to embrace it and look to how you and your business can use it to be more effective in what you do and ultimately maintain some degree of competitiveness in your sector. Now I guess, looking to the future and when I say future, we're already at 2025, talking about 2025, I can't believe we are already. What should we and our audience be aware of? What's coming out, what has been announced?

Fraser Dear:

Yeah, and I think this is the space where Microsoft are really pushing hard. So when we think about AI, the immediate thing that springs to mind about the here and now is the agentic era that Microsoft kind of started to talk about in October and November time last year. We're now starting to see the fruits of that theme and that topic, because Copilot is central to Microsoft's AI play, and AI powered agents are the bit that kind of fits behind that. So, in the same way as we talk about AI having a meaning, agents have a very specific meaning in this context. So we could talk about an agent being like a bot or like a chatbot.

Fraser Dear:

That could be described as an agent, but in reality, when we're talking about agentic experiences, what we mean is these chatbots, these independent entities, actually now have the ability to talk to each other, and so we're starting to enter into a phase whereby we no longer program for one use case.

Fraser Dear:

We create lots of use cases and allow Copilot to communicate with the ones that are relevant to whatever's happening with the and allow Copilot to communicate with the ones that are relevant to whatever's happening with the user in Copilot at that time. So that, for me, is one of the big things that Microsoft have announced and we are now starting to see those being deployed. So Microsoft announced that they're actually going to deploy 10 of these agents into the Dynamics stack, nine of which will be in Dynamics 365, and one of those will be in Business Central. But if we then start to think about that, in the rest of the Microsoft stack, we're already hearing about what else. So agents are actually going to become part of the Microsoft 365 ecosystem, and already Microsoft have launched one of those agents, which I think Andy will talk about.

Andy James:

Yeah, no, absolutely. What we've seen pop up in the last couple of days really has been SharePoint agents. So those of us who have been lucky enough to beg, borrow and steal Copilot 365 licenses will have seen whenever we open anything Teams or Word or whatever there's always a little Copilot symbol and it opens the chat and you can ask you questions about whatever it is that you're inside of. So the microsoft 365 co-pilot experience in word will answer questions about that word document and generate content in there and is amazing for those who are using the free version of co-pilot in edge similar kind of thing you can open that up and ask questions about the site you're on, or it has access to the internet to kind of generate content from wherever. That same kind of experience is now available in SharePoint and it's specific to that SharePoint site and that SharePoint content for the most part.

Andy James:

When Copilot was first launched, it had access to everything that you had access to, which was a bit of a double-edged sword, but it was brilliant.

Andy James:

And then one of the changes that happened kind of near the back end of 2024 is that you could then start to restrict or manage the SharePoint sites that Copilot had access to, because some of them might be sensitive HR information, you know, financial sensitive information and such the like.

Andy James:

So we've now kind of gone the same kind of thing but the other way around.

Andy James:

So instead of having a Copilot that has access to everything and you need to manage what it can and can't see, there's now co-pilot, specifically in sharepoint sites, that can give you access to just that information. Those co-pilots can be customized and configured to kind of maybe stretch a bit wider where needed to, but their kind of foundation is to look at that site. So if it was a, you know, a site around your sales details and with all your collateral and all the different services that you might provide, there's a co-pilot already in there or an agent that's already in there. You can ask it a question what do we do around Power Platform? It will interrogate all of your information and give you an answer, and then you can kind of surface that click through to get to that reference documentation and so on. So it's a really nice way to be able to access that information that's now just kind of integrated directly into sharepoint so really interesting to see and I think that's exactly what andrew just said there.

Fraser Dear:

That's that's the key piece is that from microsoft, copilot is the user interface for the entire ai agentic experience. It should be one interface that connects through all of these agents, and that's the kind of vision for Microsoft on this piece and that's what we're going to see coming through very, very strongly. Copilot Studio enables us to create customizations in something that's published for Microsoft, but it also enables us to create something completely unique that's relevant only to that very specific business or that specific process, which really then does kind of superpower, that kind of oh, it can know what's in my SharePoint site, great, but then what Can it do? Actions, can it do something when I'm saying, actually I'm looking for this, would you like to create an entry in a database someplace or update the status of a record, or perhaps send something for authorization or approval or review, for example, if it's within that sharepoint space?

Fraser Dear:

so agentic experience can actually do all of that and I think, if we start to think that through, would we have been talking about agents this time last year?

Fraser Dear:

Well, no, we wouldn't have been.

Fraser Dear:

And I think, when we start to think about some of the trends that we're seeing inside microsoft, I think that's going to sort of drop a few breadcrumbs for what might happen in the rest of 2025, because Microsoft announced something called Core AI Platform and it's essentially the dev team and the joining of that UI UX piece, as well as joining together that with the AI team at Microsoft, into one core platform and one core platform and that core platform, as I say, called core AI again lots of terms that we've heard already meaning something now completely different that is going to supercharge the whole Microsoft platform, because I believe that that's really going to focus that end-to-end co-pilot in AI stack to bring faster insights to businesses across different platforms. And I think these advancements will kind of really revolutionize SaaS applications and this whole category, because if it's serviced as a software, that then turns it a little bit on its head and it will make bespoke applications and bespoke processes even more powerful. I can foresee very, very clearly some big changes coming out from Microsoft over the next year.

Peter Filitz:

That sounds all very exciting, and you hit the nail on the head there earlier in terms of would we have thought agents in general would be available now last year, this time? No, because I know for a fact that we have projects that we've scoped for clients that we were essentially going to build our apps for, where now we don't. They can subscribe quite easily to just a straightforward agent to essentially do what we would have built in an app to do, and therefore they've got that ability to do it today at a fraction of the cost of what it was 12 months ago. So, no, it's exciting to see the cost of what it was 12 months ago. So, no, it's exciting to see, and I think, what Microsoft is bringing to the table with their 365 ecosystem and the AI developments in it. It's really revolutionizing the modern workplace for businesses as they are today. So, fraser, that sounds great for big businesses, but what about the end users, so to speak, and what have they got to look forward to?

Andy James:

Well, I think you said it yourself earlier, peter that you're a big advocate of kind of the personal AI bits in your life. I've had a number of conversations and led a number of workshops about AI and productivity opportunities and I make it very clear, and I'll say it again anyone who tries to take my Microsoft 365 Coppilot license will have a fight on their hands. It's just changed everything. For us. Our house is probably no different to others where we've got Google hubs and friends of mine who have Alexa devices and bits, so we've all kind of adopted personal AI bits into our lives over the last couple of years, knowingly or not. Recently on my I don't want to start a fight about android and iphone but on my samsung, new updates to the operating system have come with uh gemini ai, which is their version of generative bits, which is great.

Andy James:

Unsurprisingly, I've got copilot on my phone for personal use, so that's kind of my go-to, but we used it for helping us with the christmas shopping. You know, looking at holidays, my boy's birthday's coming up. Well, no doubt it will play a part in there, but with them there are kind of updates with Copilot Vision, where you can then take pictures of things and show items and it will help you search for and shop and you know describe and all kinds of really cool things shop and you know describe and all kinds of really cool things. So if I think back to when I got my first kind of google home hub and philips hue smart light bulb and I was trying to have it the light bulb come on at certain times and voice commands and stuff and mrs james probably rolls her eyes a lot over it and just wonders why we can't use a light switch but all of the benefits and the efficiencies that came with that being being able to set all of these schedules and bits and pieces it's the same kind of thing. For those of us adopting it. We'll be able to use that AI in our day-to-day life.

Andy James:

Copilot on my phone for my personal bits just makes life generally easier. Maths has changed since I went to school, so being able to help my son with his homework, be able to ask questions and I'm sure we'll have conversations around the proper and ethical use of ai but in the same way we would google an answer or we'd google asking questions about a certain topic to then read up and make a decision the way we can use ai to have almost a conversation with that information to learn about it, instead of googling to find a source to read it, to kind of extract the information. But just having a conversation with it to learn more, I think is going to be quite transformative. So that's definitely something we do in our personal lives because of the experience and the exposure we've had in professional. So I think those two things are going to kind of come together and be really helpful. No, absolutely.

Peter Filitz:

I think you hit the nail on the head there and it really just you know the simple terms is AI is essentially to make everything a little simpler, right, take away some of the unnecessary complexities and time needing to be spent to do a particular task or find relevant information, whether it be in the workspace or in your personal life. So that is great to see. Just to sort of round up conversations on today's chat if a company was totally new to AI, how would they get started?

Fraser Dear:

So I would suggest a very careful consideration of a basic use case what's causing the biggest pain in your business process, what's causing the biggest amount of administration? Where are your biggest amount of errors? And use that as a vessel, because getting a return of investment for AI at the beginning is really, really important. If you start building a large language model for your business with no real intent, then you're going to spend an awful lot of money and an awful lot of time with very little return. And actually getting that one early win and showcasing really what AI can do for your business can then set up a whole series of activities with adoption, with kind of governance and security and control All of those things are really, really important. But actually getting that confidence level that AI is going to work for your business is really really important.

Peter Filitz:

Fraser. Just following on from that, then, am I right in saying that there is some funding available from Microsoft if businesses are looking to embark on this AI journey?

Fraser Dear:

Yeah, that's right. Because of BCN's status with Microsoft in the UK, we can actually engage on a number of different funding opportunities with regards to different use cases. So, for example, copilot or perhaps maybe some of the more complex data science workloads, there is absolutely funding available to start to look at kind of needs assessments and understanding the use cases that businesses have today.

Peter Filitz:

Well, gentlemen, I think that sort of brings us to the end of today's interesting conversation, very topical. I want to thank you guys for taking the time to join me and cover all of these interesting topics. With that being said, we hope that you've enjoyed today's conversation around AI. If you or your business is interested in understanding more in terms of how BCN, in partnership with Microsoft, can help you on this journey, or if you've already made a start to this, then please don't hesitate to get in touch. As we've said, given our great relationship with Microsoft, we can help complete these assessments on your behalf and see if you're entitled to any funding, but also really provide you with guidance on where to start and what to focus on. Please don't forget to visit our website at wwwbcncouk. There, you'll find a wealth of information, product updates, service updates, as well as events that we've got coming over the coming year. And don't forget to like and subscribe. Thank you for joining. Cheers, guys. Thank you, thanks, bye-bye.