0:04 There is that importance of making sure that you've got those instructions right. Those who are honing into their craft and learning how to use AI to do even better will carry on doing even better and they will not lose their jobs. So it gets complicated and unpredictable as we give it more to do. Ask AI a vague question and you'll get a woolly answer.

0:31 Ask it a complicated one and you'll get a complicated answer back. My guest as Maggie says, that's not new. It's the same rule that's always applied to good leadership. Just sharpen now with AI in the room.

0:44 We get into why working with machine intelligence means we need to become more human, not less. Why the businesses seeing 7,000 hours saved on AI programmes are the ones treating it like a toy. Not a tool, not a saviour. And why the people who hone their craft are the ones who won't lose their jobs to it, while AI slop quietly costs the ones who don't.

1:07 If you've ever wondered what actually separates AI that works from AI that wastes your money, stick around. Today, my guest is Maggie Sarfo. Maggie and I met recently through the Only Business Network and she seemed a great fit for this podcast. Maggie is a founder and CEO of Meris Consult, a global growth and AI adoption consultancy.

1:26 With 20 years of experience supporting scale-ups, SMEs and corporates, she's a Gartner alumni and advises founders and C-suites on growth, AI and the future of work, and her clients see great results, 500% growth for a startup in two years and 7,000 hours saved on an AI programme in just 12 months. She's a keynote TEDx speaker and named on the UK top 100 female founder list in 2025. Maggie enjoys tennis, yoga and family business. Maggie, welcome. Thank you, Shaun.

2:00 It's been a pleasure. So I'm really looking forward to our conversation. So it's really interesting. And just to keeping it real, we spent a little bit of time figuring out the tech, right?

2:12 We had a problem and we had to switch to your phone and we got there in the end. And I bring that up because with our experience, we still have challenges. Technology doesn't necessarily get more reliable. And in fact, sometimes.

2:26 The more complex it gets, the more of a challenge it can be. And we see that with AI, don't we? We see AI is less predictable. So more complex, less predictable.

2:38 Now, how have you seen that come up in terms of with your clients and some of the concerns they have around AI and its ability to be, well, less predictable? I think, you know, and that's what will always be there. And that's why it's important for us to really. Take a step back and look at what we're looking for AI to do for us.

3:02 What I find is the more complex you get. So, for instance, if you're looking at, you know, the type of AI that you're using to generate information or content. Right. If you go into ChatGPT, Claude, Gemini, Copilot and you're asking a question, you will receive an answer. Right. If it's not good enough, you can ask the question again or clarify it or train it again.

3:26 And it will give you the answer. That's the simplest form. As you start to say to it, now, I want you to build this, you know, MVP. I want you to build a solution that I can then link to my CRM and I want the automation built into it.

3:44 It becomes complicated. Right. And those kind of scenarios are the ones that we find you need to make sure you take a step back and get what. You are expecting it to do right and make sure that everything that needs to work in the background is working, but also make sure you have a fallback position. Like recently, we had a lot of issues with regards to some of our AI providers where, you know, I read one on LinkedIn where, you know, one of the automation programmes has actually deleted 100 clients.

4:19 Perfect clients. And so and so, yeah, so it gets complicated and unpredictable. As we give it more to do and it all comes back to really the human in the loop, the human at the start and then making sure the instructions are clear, making sure we've got something that sanity checks and making sure the human comes back to finish that loop as well. Yeah, absolutely.

4:46 And that's something that I actually posted a post on LinkedIn today saying exactly that. So it's really interesting that you bring that up as well. And for those that don't know. Well, then MVP is it's a minimal viral product.

4:58 So something that when you're looking to build a product, what is the minimum products that you can build to prove that? Yeah. And it's interesting when you in terms of when you speak about there's two elements I pick up from from one multiple things. There's lots in there.

5:15 One thing I particularly two things I particularly picked up. One is this element of mistakes can be made. So this and the importance of you, as you say, of the human in the loop, the human. So the whole point of the app is to make sure that we give it the clear instructions.

5:30 So if we are developing something, making sure that we're actually going to practise, right. You know, that that first part of it, that like you said, making let's get the instructions right. Correct. As you say, if it's a bigger project, you're actually building a product. Then, yeah, obviously there is that importance of making sure that you've got those instructions right.

5:51 And I had a situation yesterday. I have got some automation. So. I set up to help for this, the scheduling of the podcast, et cetera. And I'd fix some, I used, um, Claude codes to help me to build some codes to fix something.

6:06 And then I ran, and then I ran it for the last episode and it made a mistake and it made the episode live and it's supposed to schedule it. And that wasn't there before. So, so of course, that's the thing to remember that humans still make those mistakes when we, if humans are building, we still put introduced bugs. We still introduce issues within. Code within things.

6:27 So we can't expect AI to be any better than that. So we shouldn't treat it. I mean, the irony is we're looking at, we talk about it, artificial intelligence, but I had somebody recently call it simulated. It's like simulated being a human.

6:41 And if we make mistakes, then it also can make mistakes as well. A hundred percent. Yeah. Yeah, I agree.

6:50 And the, the interesting thing with, you know, the AI systems is. When it does make a mistake and you spot it, because sometimes it actually goes, oh, I've done it for you. Here it is. And then you go, no, when you check it.

7:06 Right. So if you didn't check it, you wouldn't know because you thought you had given it the right instructions and, and, you know, you said, correct this and do this and do that. And it goes, oh, there you go. I've just done it. Then you look at it and go, no, you haven't. Right.

7:19 So, so again, it goes to show. And I see that quite, quite a bit. Yeah. And I, and I think, um, I mentioned before on the podcast as well, but I know if you find this, but sometimes I get frustrated.

7:31 I forget myself because it's given me, given me speaking to me as if it's a human and I get frustrated and going, no, that's actually not true. And it's reminding myself going, this is a machine that's given me a predicted response that it believes statistically is the right response based upon the information I've given it. And that's actually the other thing that we, we, we, we bring. And it's bias, isn't it as well.

7:55 So if we've asked it for something, it's going to try and give us what we asked for. So coming, like you said, that, that ask has to be, is so important to get it right. And I would say it's almost one of the key skills for, for organisations to be able to grow is the ask is the better the ask. And so how would you, how do you normally advise, um, organisations, people you work with to how to structure that, that ask in a better way? Yeah. Ask.

8:24 I mean, it's what we call prompt engineering, right? But it's even further than prompt engineering. I think before you get anyone to even start prompting, you need to understand what is happening within the business, within the organisation in terms of what their perception of AI is. What, you know, do they want to adopt AI?

8:49 Do they have fears around it? Do they have doubts? And, and, you know, that, do they think. They really do this, right?

8:55 You know, their jobs are going to go away. Those kinds of things, you need to really hone it and help them create their own internal policies that, you know, doesn't blame people for trialling it and making some mistakes and things like that, but also protects the organisation. And so that way, when you have like some of these internal policies in place, because we don't have the external regulations yet, we got the AI EU act coming up. In August, but we don't have them kind of really embedded.

9:29 And I think some of the apprehensions or resistances to adoption is around that. So it's important for you to do that. And it could be like you have a playbook internally that people understand how to use AI. How is AI done around here?

9:46 Over time, you want that to be embedded into their culture. Then coming back to your question again, the ask. The ask. About the prompting. When you ask a complicated question, you will get a complicated answer.

10:03 When you ask a vague question, you'll get a vague answer. And, you know, this is not a new thing. But of course, with AI is a bigger, deeper thing. We know this from other forms of work.

10:17 You know, I mean, I've had the opportunity to coach leaders a lot. And again, it's the same thing. The question will determine the right answer. So prompting starts with being clear about what you need to get out as an outcome.

10:34 We need to really keep looking at the outcome. And then you look at it from your human perspective. And one thing I really encourage is really looking at our own emotional intelligence. So I bring this into conversations all the time.

10:50 We are working with machine intelligence, artificial intelligence. You know that. In order to get the best outcome, we need to become more human. And so you being able to look at it from a human perspective means that you ask the right question that considers not just your perspective, but the perspectives of others.

11:11 That will address the bias kind of conversation you were talking about. Then from there, so you put the question in, and then you are clear. I want the results to be presented in a spreadsheet. With what? With these columns. That's an example, right?

11:27 Give me a list of companies in this region. I'm assuming someone who's kind of trying to build a sales list. For instance, give me a list of companies in this region. Um, and I want you to present it in the form of an Excel spreadsheet.

11:42 I want you to, you know, highlight the name of the CEO. And if they have a LinkedIn profile, add the LinkedIn address, and then something like this will give you a better answer. Unlike someone who asked one question, come back again and say, you know, now add this, now add this, now add this. So that, that is an example of what good prompting could look like as opposed to not good prompting.

12:12 Yeah, absolutely. And thank you for that. I think that's a really great explanation. And it makes me think of, I mean, I'm not the best at drawing, but I have actually had a go.

12:21 I've got a book, like, you know, drawing 30 days and then you've got to day seven. So I need to do that. And it doesn't move on to the rest of the days, but it makes me think about if you're making a drawing and then you are, you could rub this out, rub that out on the ad a bit more, any, but more. And what you can end up with sometimes is something that's messier than if you'd actually spent the time in the first place, thinking about what you wanted the outcome to be.

12:43 And that's what I take from that as well. It's again, it's, it's, you talked about, you know, motion intelligence and it's interesting because yeah, we are working with, uh, a. It's not that we want to interact with as a human, but realise that actually, if we are more human, you know, our approach to doing that, we recognise our frailties. Well, it's been, it's been trained on data around humans.

13:15 So it's going to, it's going to have picked up on those same frailties. I think it's so interesting. You mentioned motion intelligence. You're the first person that, um.

13:24 At the podcast that's actually brought that in as a factor. And I think that's actually really insightful, really insightful because it ties in with the fact that emotional intelligence has been something that. By over the last five to 10 years, but it's been, it's grown in, in terms of understanding that it's an important factor in terms of our personal and our business lives. And actually, and it really is what makes us human understanding ourselves better. Yeah.

13:53 And understanding the emotions of other, other people, the better allows us to interact better in a way. And I love that concept of actually adding in emotional intelligence into your considerations in terms of how you, how you prompt AI. And also just jumping back a little bit, you mentioned about experimentation and how important that is. And I really love what you said about having the playbook saying, actually, yeah, this, this is some of the challenges, isn't it?

14:23 Organisations have, they have the term going out, shadow AI, where in your organisation, you might say, oh, we don't use AI, but there's people using free accounts and people potentially using company data in free accounts, potentially using information that's actually GDPR information. Yeah. Sensitive information. And so actually, yeah, that, that playbook, I just wanted to highlight that because it's so important, even if that is a minimum of saying, you say, this is how you can experiment.

14:52 Yeah. These are some company accounts that we've provided to experiment. This is, you know, this, and these environments are not attached to the data. Here is what you can and can't use.

15:03 And obviously, obviously the, so there's that element of experimentation, but tell me, build, build the, build the confidence. When an organisation, so let's say, yeah, you've, you're working with an organisation. They've, they're in that, they've done that experimentation. What is that?

15:20 What is the next step for them? They've got their, you know, maybe. Maybe they've got a basic playbook and they've done some experimenting and they want to take the next step. What would you, what would you normally, how would you only support them in doing that?

15:32 I'll say they are now ready to, to assess how ready they are. So, you know, when you were doing the intro, you, you kindly kind of introduced the, the AI readiness, AI maturity assessment is only a 10 minute assessment, but it does allow you to look at. The seven areas of your business in relation to AI, because we might sometimes think we are ready, but we might not be ready. Um, and being able to assess your readiness will help you select one use case that you want to start implementing within the organisation with, with that use case.

16:14 Then you are able to then look at governance around it. Look at the play book again, look at things like AI literacy, you know, what is a common language that we want? What is a common language that we want to be using when it comes to AI within the business, within the organisation? And once that's all covered, you're ready to actually implement that solution, whether you're going to be building it or you're going to use existing solutions and, and develop what works best for you.

16:39 So the readiness assessment is important. It covers things like strategy and leadership. It covers things like your tech and infrastructure. It covers things like your data, your risk.

16:53 Um, culture, then I've got a couple of things, um, operations and use case selections. But once you've completed it, it gives you an idea of where you are. We are still in very early stages of AI adoption. And so we find most businesses will be.

17:10 So if we look at this on a rating scale of one to five, five being those who are really kind of going into the agentic mode and doing something big with AI, um, I'll say most organisations and businesses. Um, I'll say most organisations and businesses. Will be at one and two, and that's fine. You want to build the foundations, right.

17:29 In order to step into the future. Okay. Yeah. Thank you. That's really helpful to understand that. And it's what I find interesting from AI in terms of I've worked over 30 years in, in, in technology and what I find interesting with AI.

17:45 So I come from an IT background, so it's, it's very much been the, the IT, you know, the IT, you know, we're the custodians of the technology. That's generally how it's been and AI is changing, changing, changing that narrative because my, you know, with your assessment, it's a holistic assessment, isn't it? It's a, it's a, it's a business wide and it's, and this idea of it, when we come back to the experimentation as well, it's, it's ensuring that it shouldn't just be the IT department and sorry, IT guys, but I'm, I'm an IT guy as well. And we just, I think it's, we, we want to often within IT we're, we're there for, to.

18:22 In short, to protect as well, not just to, to help support, but also protect the data and everything else. And with AI, we still want to continue to do that, but it's so important, isn't it? To have people in their teams that aren't in, in, in IT to be able to have the opportunity to experiment. So understand what is a use case for them?

18:45 Like, and I love that, what you said about the single use case, create an opportunity to go, let's not go huge. Let's go. Here's a business case. A business case that we could do in a, in a fixed period of time that will help you as an organisation grow more confident. And it might have nothing to do with IT as an example, or it might do, but it's just that helping organisations in it to understand this isn't about IT.

19:07 IT is important because there's a, there's a big chunk of that. And you mentioned a number of the, a few of the areas you cover in the assessment are very much, very much IT. But it's, yeah, but it's then making sure that actually it's not limited. We're saying here's an opportunity that in, instead of somebody from the IT department having to get the requirements to build something, then actually that could come from the, from the actual, that department saying, oh, actually we've got this idea.

19:41 We've experimented a bit. This is what we've come up with, come up with. This is kind of as, you know, maybe as far as or comfortable as we, we like to go, but creating the opportunity to explore that using AI. As a, as an inspirational tool.

19:56 And one of the things we we'd spoken about, uh, yeah, previously in a previous conversation was around the, the human element. You mentioned the human wrapper and what's interesting that we talk about the human wrapper, but sometimes that execution piece, the piece in the middle can also be done by humans. Is it something that's making that decision to say, I could create this, let's say it's an image. I could create this AI image.

20:22 Or I actually really enjoy doing that. Actually, I could do that myself. And so there's that consideration about the importance of, I mean, you went, you know, we start with the emotional intelligence. We're also looking at it and going, yeah, we want to be human.

20:39 And the more that we train AI in terms of things like emotional intelligence in, even in areas of, of pushing back and us and saying, well, actually this is, this is something you could do. Why, why, why am I doing it? I mean, as humans, we do that. Wouldn't we, if you go, somebody comes along and they say, they say, oh, can you do this thing for me?

21:00 And you go, no, you can do it. Yeah. You ask them further questions and that helps them understand that they could do it. And then maybe you add something to their idea, but you, you empower them to do it. So it's about that as well.

21:16 Yeah. So this, yeah, so it's definitely important. And how do you. In, in, in your approach, how do you kind of handle that, that idea of making sure that the human is, is still developing?

21:31 Yeah. So, I mean, we, we are looking at outcomes, a big chunk of the work we do at Maggie's Consult is AI business value creation. So we are growth consulting and we're doing AI business value creation with regards to growth. We are looking at outcomes. Right.

21:51 But then, you know, how can you get the best value at the end of the day, you've already invested resources that includes money that includes your people time, but what's also a stick here is, you know, what you have set out to achieve, like as part of the bigger organisational goals. And for us particularly, we look at growth. So sales, marketing, customer experience are the use cases we would look at. So how do you make sure? I think.

22:21 So what we need to recognise is that AI can save time at the end of the day, we want the outcome and we put like measures and KPI measurements and all forms of measurements into the, the, the programmes that we do. And that's how we're able to come up with that 7,000 hours saved in 12 months. So you want to look at the outcomes you want and you look at what is the best way of. Getting there.

22:52 Now, sometimes you have to make a call on whether you're looking at the quickest way, but if it's the quickest way, will it still give you the best value or could it be an outcome that is kind of shoddy and is that what you want? So it's about weighing those scenarios and choosing the type of activities that you want AI to be doing and then the type of activities that you want humans to be doing. What are the things? Things that humans only can do and could you develop the humans further into honing into those skills in order to work with the machines?

23:31 So the machines do the mundane stuff, like the repetitive stuff, but still with a bit of human oversight, I think you can't take the human out of the equation, but could you develop the human better? So if it was to draw the picture, for instance, could you develop them in such a way that they don't just. Draw the picture, but now they can animate the picture because people respond better to videos, right? And so if that's the case, then how do you make sure that AI is able to draw this image to a certain level?

24:08 The human comes in, perfects it to what is required, maybe hands it to AI to automate some of the animation. The human comes in to again, perfect it before it goes out to market. Or whatever it needs to be done, it needs to be released into the world with. Yeah, that's really helpful.

24:30 What's interesting, the word that comes to mind me to here is tool and that recognition that AI is just another tool. Its potential capability is we don't even know the potential capability, but it's just that if we think it was a tool and we say, well, you know, when we're using a hammer, you know, we're using a hammer to hammer in a nail. But I can't remember. I mean, you could automate a hammer, like hammering, hammering in a nail.

24:58 But the point is that that actually allows us to do something. Yeah. You want a chair that you can sit on. That's the outcome. Right.

25:08 So then it's not about the hammer. The hammer is part of the process is the tool. But the outcome you're looking for is a really good solid chair that you can sit on. And the hammer is being used.

25:21 Um, to, to bring that chair alive. Yeah. And that, that, that ability to this idea, this concept we're talking about with AI working alongside humans, that, that, and coming back to what we spoke earlier, you're the one giving the ask, you're one setting the outcome. And then it's sometimes it's AI executing.

25:43 Sometimes it's you executing, but then whoever's executing, there's still the check at the other end to say it's the output and saying, going, well, actually. Yeah, that could be better. Right. Let's, let's re refine our ask for the next time we do that. So then continuously, continuously improve.

26:01 So we can continuously prove, improve our use of AI, continuously improve the human ability as well. It's not, they're not, we're not taking one away from the other. And I've heard more information to say that actually we're not seeing the, reduction in jobs. Related to AI that actually was expected and a lot of the, what the redundancy with jobs we have seen being lost are just related to a, to a actual one, a natural cut because of over recruitment around COVID so, so we've seen, so, so actually seen that, that past that, that, that reset, but we are, we, you do expect to see displacement in specific jobs.

26:47 And one of the interesting ones that is, um, well, sorry, software engineering that can come, has come up and like, oh, software engineering now that it can code. But actually I've heard that there's actually seeing an increase in software developers and, which you kind of think why, why would there be an increase if AI can do it? And I think it comes back to some of the things we've been talking about that if I'm in an organisation, let's say, I don't know, it could be an organisation of 50 people, right?

27:18 I'd probably, I may not have a software developer. I might have my IT. I might have a few people in IT and maybe the rest is outsourced. I may not have a software developer. and then I have and then I work with like Meris and they're going oh well actually we you can and then they think well hold on we could have our own like software development department and

27:35 they go but we wouldn't need 10 developers we could do that with two developers and AI yeah and oh isn't that and so I think it's when we look at it for large organisations it might be the organ they're looking about well okay we've got maybe we've got more than we need but for the smaller organisations it's saying well actually you can bring in some capability

27:58 that you didn't have before but actually for every capability that you bring in using AI that's what we're speaking about you do need to make sure there is somebody that actually knows what's going on knows what knows what good looks like understands what the outcome is looking to achieve that is able to ensure that AI

28:18 AI is providing that you I was had a friend over yesterday and I was showing them some of the things I'll be using Claude Code for and he said to me oh couldn't you just show the people how to do this and I said well I could but the problem is the challenge is not showing them what I'm doing I could show you

28:37 what I'm doing but it's much harder for me to actually tell them my thinking in terms of when I look at code and I go that's good that's not so good or actually AI comes back to me and says I'm going to do this and I'm going to do that and I'm going to do that and I'm going to do that and I'm going to take this approach and I go no no that that there's flaws in that approach and I

28:54 go back and I say there's flaws in that approach this is the problem oh yes you're right he this is a and so I'm often going back to it now that isn't something I can teach that that is that is something that comes from experience and that's another key factor isn't it that we can't we can't just train people to do these things we need to create the opportunity as we talked about

29:16 through experimentation with AI but also we need to the opportunity for people to continue to grow their experience because they are the ones that are defining what good looks like that is true and actually what you're describing we've seen that a lot in businesses so it's all well and good to say I can do this with AI but if that's not your

29:40 field the quality of the work you're producing the quality of the outcomes that you're producing with AI would not be comparable to someone whose field it is and who has honed into their craft and so if you happen to be lazy on your craft the quality of the output will not be a good quality

30:06 those who hone into their crafts they are able to ask the right questions they are able to give the right directions and the output is so much better and so that comes back to that question would people lose their jobs um those who are honing into their craft and learning how to use AI

30:30 to do even better will carry on doing even better and they will not lose their jobs um and so yeah and so then from that perspective it doesn't make sense for you to then decide oh now I'm an architect because I can use AI now and then I'm going to start doing better and I'm going to start doing better and I'm going to

30:49 I'm a software designer because I can use AI you can but you it makes it depends on what you are coding as well and if what you're doing needs to really gain competitive advantage in the world you need the expertise of someone whose world is because they they bring their expertise into the work of AI to take it to that next level and we're seeing that now absolutely it's one of

31:17 the reasons why a lot of um people who were laid off um are being called back in um because you know they're so far that AI can get but there are limits to that in terms of the the knowledge that that human expertise brings to the fore to combine with AI to get to that next level wow that I just think you just you just dropped another nugget there which is the

31:47 importance that human growth the importance that we we continue to hone our skills which is something which is counter to some of the things we're hearing from people is oh we're going to lose our jobs I think it's an encouragement to students so any students that are listening or anyone that's concerned about your current role you hone your skills hone your skills look at what

32:10 other skills will be valuable in in going forward because that is such a powerful statement I think for organisation as well to say don't don't just bring in AI and allow your people to go stale because and that's one of the biggest challenges I think we're seeing is we need we're still going to need the skills we can't lose the skills we want software developers that work with AI that's a

32:36 real encouragement I was working last few weeks at graduations in Bath and the speeches the chance would often mention the challenges with AI and the challenges with AI and I think that's a really And I know there's a lot of concerns in terms of what do I study in terms of what do I do as a job?

32:54 And I just want to, I just think what you've just said, there is an encouragement to say, actually home, nothing has changed in terms of be the best you can be.

33:03 You know, obviously you still need to pick a field that's still going to be valuable. Whoever that field is, be the best you can be in that field.

33:11 And you, you, and you will get the jobs because what, when you'll have seen this yourself, we hear the term AI slop where,

33:19 and now I stop is really where the ask is poor. And while there's no human wrapper, there's ask is poor and nobody's checking the output. It's like, Oh yeah, I'll just create something for out of there, automate it.

33:28 And you'd be improductive. I'll get followers. Yeah. And then they go, but it's,

33:32 but actually if the outcome is more sales, more leads, more brand awareness, often none of those things are happening.

33:41 So, so you're, you're, and so therefore you're not only is AI,

33:45 it's, it might be, it might be saving you time in doing it, but your overall outcome is you're not getting the return on your investment.

33:52 You're spending money in using AI and you're not actually, you're getting those things. So I think that human quality, I think,

34:00 so there's two things, two nuggets, I was numbering nuggets, but I think the key ones I've come up with is that the importance of the emotional

34:05 intelligence and this human quality, but we need to continue to improve as, as in our own skills, in our areas of expertise, that it doesn't,

34:17 AI doesn't become a replacement. And that brings me on to, to this consideration about whether we should, we use AI or not,

34:28 and not in terms of the ethical side of it. I've covered quite a little bit of that in the, in the, the previous episode, episode I did with someone,

34:37 but I'm thinking more in terms of going into an organised. And we talked about this sort of the creative element, but this, this is really about I go in and I, I need some information like what's the,

34:52 yeah, what's the weather today. Now I could just go and use AI to do that. Or I could go to the BBC weather,

34:59 which is what I do because that's, that's the, what I, that's what I trust.

35:06 And that's where I go. Now, obviously if we make the decision to use AI in, in a lazy way, I mean, that's a bit harsh,

35:14 but it's saying in a way, that means we're not doing something ourselves. There is a risk there of reducing the cognitive ability, but the other element of it is the cost and the environmental, you know,

35:27 the sustainability impacts of going, well, actually we, if we don't need to fire up those processes to get information, if I go onto the BBC, whether the information's there,

35:38 there's no processing or very minimal processing needs happen because it's there. In fact, if other people have, if in my area have looked at it as well, it's probably the information be cashed somewhere. So that,

35:49 that whereas if I think if I go and ask AI, then there was a lot more processing happening in doing that. And how do you, how do you approach that when you, when you're working with organisations in terms of the,

36:02 the amount of AI to use and when it's appropriate or not. Yeah. And the sustainability elements of it as, as well as kind of wastage of raw cash as well, because you know,

36:13 the subscriptions are going to go higher and higher. So incidentally, I've been speaking with an organisation and infrastructure company. So this is not an AI company, but tech company, but I guess it's been,

36:23 it's been built to support this particular issue here and what their solution does is sit in between the organisational AI environment and the, say the large language model that they're using or open AI for example, and they're using open AI for that. And so it's, it's a,

36:45 it's a more complex industry, not a much more complex and robust, multi-millionaire industry, not a lot of AI is in the organisation. Um, but I think it's, I think it's a, it's a part of,

36:57 it's a part of the AI, the AI that we're producing is a part of what it's doing. And it's the cost of that. So there are no, there's a, there's no cost there. It's,

37:08 it's not a big deal. It's not, it's not like an educational business. Right. It's, the large language model and and support and add to the answer what this happened what happens is that we are reducing the amount or level of tokens that we're using eventually um you can as an

37:31 organisation or a business go and negotiate this with your your um your ai provider so if they said your subscription is going to go up by 30 you can show them that actually the way you're using your your tokens has meant that there has been a reduction in tokens because the dashboard on the platform will show and you you'll be able to then save money as a result but then there are so

37:58 many other elements sustainability elements as well that you're not getting the machine to process all the time and because that it's all um impacting um you know the the systemic this sustainability um kind of classes that we've said we we want to be adhering to as organisations and as businesses and as people in general

38:23 so yeah so we we're seeing that and um token usage is going through the roof and that's not just on the organisational side i mean these um um say chat gpt um trends that come where you know this is flying all over on social media put this prompt into chat gpt about what you do and see what it comes up with and then comes up with a picture and all of

38:52 those kind of things that uses a lot of tokens and so how many times do you want to do it even if it's valuable do you want to keep doing it because even as an individual or as an organisation if your your employees are using it it's also going within the same token usage and is is costing the planet yeah yeah absolutely and i think that that is

39:16 such a key question to be asking is you know that should i use ar not in a particular context but also what i take from what you said there is there's a there's a real advantage in terms of in organisations that have you know we're talking about the the costs of an organisation how much it costs to run it and ai is obviously get you know becoming a a cost line

39:39 yeah and actually there's an opportunity for organisations to reduce that cost line as you say you know the expectation is to say well okay one of the ways i use predominantly at the moment i'm using ai to replace it to code so i don't have to use ai right so i'm reducing i'm using ai for the minimal if i need some thinking then i use i'm using ai but if i can if i can generate

40:05 something if i can use pure code to do it it means it's going to be a cost line and i'm going to be unpredictable i know the outcome is going to be the same every time and so and then also as you say i'm not i'm not using tokens i'm obviously looking at using um we've got time to cover that but in the previous podcast episode we i've talked to talk with someone about local open source

40:26 models but um but even even then it's it's there there's such an important factor like do we use it do we not use it and i think for organisations it's really important that you don't just accept your ai because it's not the all singing all dancing saviours everything and as you you pointed out we've seen organisations have gone down that route had problems because they lack the people

40:55 and as you bring the people into their into that then there is still you know people are going to are going to find ways to say actually we don't really need ai to do that or it might be more costly to do that right so and so and so and so and so and so and so and so and so and so and so and so And this is one of the challenges with we've kind of come to accept in a strange way that there's a cost.

41:19 Now, consumer-wise, it's been pretty static. But from an enterprise level, what we're seeing, the costs are creeping. And you're going, oh, you're getting less for what you get. I mean, it changes every day.

41:33 Like Opus 5 came out. And it's like, oh, the Opus 5 is, you should use that now because that's going to use less tokens than Opus 4.8. But that's still more than the other models. And I noticed that they changed my default to be the higher model.

41:51 So I was like, what model am I using? Oh, I'm using this model. But that wasn't what I'd set to the default. I mean, we're on this, obviously, so much more we could talk about.

42:04 And I just, I think really to close, it'd be really good to just, for those business leaders, for those CEOs out there that are going, okay, this is all really great. What are some kind of three key questions they should be asking themselves in terms of how they move forward or what's the next approach they should take in terms of, it's not necessarily specifically AI, but we're talking about, obviously, it could be. But I mean, in terms of they've got some outcomes. They want to achieve.

42:35 Yeah. And AI might be a possibility. Absolutely. And that is our sweet spot.

42:40 And we feel that that's where every organisation needs to be all the time to avoid being driven into directions that they don't necessarily want. So the questions we need to be asking as leaders is, what are my business challenges that I'm trying to resolve? Then the second one is, is AI part of the solution? And if so, what are or what is the best AI solution for me?

43:10 And these are three fundamental questions that every business leader needs to be asking. And you want this to be aligning back to whatever strategy you set at the start of the year or for a three year period. You want to be responding to that all the time so that any decisions that you're making, any spend that you're making, any AI solutions, any solutions that you're implementing in your use cases, they're all feeding back to the overall goal that you said you want to be achieving this year.

43:42 And that will save everyone a lot of troubles. Yeah, yeah, absolutely. And I think there's a couple of things to take from that. One is you put the challenges first.

43:54 So you put the identify what you want to be doing first, as opposed to how do we use AI, right? Yeah, 100%. Yeah. That's that first thing, right? Okay.

44:04 That is the challenge. But also you talked about strategy. And so what we're really saying, we're saying is you can't replace having your values, having your mission, your purpose, your strategy, your, you know, setting goals, disseminating them into the organisation, having people behind what you're doing. All of those things are still fundamentally important.

44:24 And I would even argue that even before you consider using AI, if any of those, if you don't have any of those, then, you know, make sure you have those. Because how do you make decisions? Yeah. Particularly using AI in a way that could affect your branding, could affect your, you know, you've, you've got, and that's what we talked about with the human wrappers in there as well is, is you want AI to be the gatekeeper. I was M&S bank.

44:49 I got an, I got an automated letter from a card that I closed the account in 2022, which must have been expiring now. But I got a generic letter that said, oh yes, you know, your car's expiring. And we've, we've, we've done a credit check and decided that, you know, we're not going to renew it. And I was like, oh, I was, first of all, it was a scam.

45:10 I thought, I don't remember this. Then I looked it up and realised, okay, this was something that I cancelled four years ago. Yeah. So then, then I tried to call.

45:20 And I tried to call them. So I was like, oh, I couldn't call. No, no, you called them. They say, use our chat.

45:25 I find like I use your chat and the chat, you know, round and round, they go in the loops and I end up going back to, oh, you should call us. All right. So I call, I try all the different options of the numbers and then not because it, you know, I couldn't, I even tried the credit card number and I wouldn't let me through. I couldn't get anywhere. So in the end, I actually went onto Google.

45:46 I know there's a website called say no to 8087070, where you can find alternative numbers. I found an alternative number, called that for the MLS bank, spoke to a person and they said, oh yeah, this is what's happened. But it took me probably half an hour. So I say this.

46:03 I say that as an encouragement to those considering using AI for their support environments that actually you really, you still want to have a human monitoring that to understand are there experiences happening, but also don't lock people out to, because I was without me going out and finding some other entry point. I could not speak to speak to human to resolve a problem that wasn't even my problem in the first place. And I saw that four years ago. So I'm not missing MLS bank.

46:33 Just seeing it as an example of saying how we can use AI better. And I think, yeah, it's been a fantastic conversation, Maggie. I'm so, so glad that we, we, we got through the tech and we, we actually thought through it, used our brains. We found a way forward and we didn't use ChatGPT to give us the answer.

46:51 So, so thank you so much for your time. And how can, how can people, how should we get in contact with you if they want to learn more, they want to do an assessment or what should they, what should they look? Yeah. So they. Can go for our website, which is marisconsult.com.

47:07 So M E R E S consult.com. Um, and you will find everything that you need to know about us, including, um, um, the, um, AI readiness assessment tool, um, is free of charge to use as well. You can also click a button to book a call to speak to a human if you need to. Um, but you can also connect with me on LinkedIn. I share.

47:33 Lots of, um, our knowledge and experiences on LinkedIn to support other business leaders. Oh, great. Thank you so much. And those that will, will have those links, the links will be, uh, wherever you're watching or listening, they'll be on those platforms as well. So you'll be able to easily go and contact Maggie and, and yeah, and get support and hear the insights that she's got to share.

47:55 So if you've, if you're watching on YouTube, then please subscribe. If you're listening on, on your favourite podcast channel, please, please follow. And, and like, and really, yeah. And if you want to hear of any particular other topics that we haven't covered so far, then please get in contact again with, there'll be, the links will be in the platforms.

48:13 Yeah. And thank you again so much, Maggie. It's been, it's been a pleasure and I'm, I'm going to be really thinking more about the emotional intelligence, the importance that we continue to grow as humans and, and a great, a great way to end, you know, um, a conversation around for the better with humans podcast. So thank you once again. Thank you, Shaun.

48:33 And thank you everyone.