Optimising the Operating Model for Continuous AI with Sarah Heffron [085]

Narrator: The Business of AI Podcast, exploring how businesses are using AI to build services and tools, transforming organizations and delighting consumers. Produced by UKAI and hosted by Tim Flagg, AI is our business.

Tim Flagg: Welcome to another episode of the Business of AI podcast. And I'm delighted to welcome Sarah Heffron from Sarah Heffron Consultancy to this edition of the podcast. So, Sarah, welcome.

Sarah Heffron: Thank you very much. I'm very happy to join you.

Tim Flagg: Great to have you here. So I wonder whether to start off with. Could you tell us a little bit about you, uh, your background and then how you ended up doing what you're doing today?

Sarah Heffron: Sure. So, um, I have spent my whole career, more than 20 years, working in various forms of operational transformation, um, at very large multinational, global organizations. Um, and that's financial operations, commercial operations, product operations, um, but fundamentally, in every case, it's really been operating model operations, because in a business there really aren't any true silos. Everything connects to everything else. Um, and so I've seen this at enormous scale. I've seen how things develop over time. Um, I've seen ways of addressing things, um, that are perhaps not, uh, intuitive, um, to people who haven't kind of experienced it and lived it. Um, and when I left corporate about a year and a half ago, I decided that I really wanted to take everything that I had learned and everything that I had experienced working across people, process tools, technology, data incentives, information, and try and package it up. And then I brought it together with, um, AI. I've spent a lot of time on it. I've gotten MIT certification in it. Ironically, my dad was an AI specialist for the last generation of AI. So I kind of grew up around, um, and I pulled it together and I recognized that AI has enormous opportunity and enormous potential. But it's not just a technology and a tool that you sort of throw into an existing business to really get the value from it meant doing that operating model work. Um, and not a lot of organizations know how to really, truly look at their operating model, look at what it's doing, and then understand what that means. What do they need to do, what can they do, what are their opportunities so that they can get the best possible outcomes. Um, whether that's scaling, whether that's AI adoption, whether that's other forms of growth and success.

Tim Flagg: M really interesting just to. Can we just unpack a definition of what we mean by operating model, though? Sure, I think that would really help. Um, you know, I think if you've worked in a consultancy or a large organization that's maybe hired Consultancies. And you're probably familiar with the concept of an operating model. Um, but for the, maybe some people who haven't come across it, um, what do you mean by an operating model and why is it useful?

Sarah Heffron: Sure. So an operating model, um, is something every business has and it's not just for large organizations because it's literally how does work get done? How do you get from your strategy to executing that strategy ideally efficiently and effectively, but it's literally how do things get done? It's your organization, including your people, both their competencies, their capabilities, your organizational structure, your incentives. It's all of the processes. How are things supposed to travel through a process? Um, it's your tools, it's your technology, it's your data, it's your information flows, it's your governance. How are decisions made? Um, it's all of those things combined. So it is essentially everything together about how your business gets things done or does things. Um, because sometimes you do a lot and it feels like you're not getting a lot done.

Tim Flagg: Um, so I think people will recognize all of those component part you talked about there. And you know, I've run small businesses and so all of those things make sense in the components. But I suppose what we don't have often in small and medium businesses is the time to be able to think about it. Um, and we might not call it an operating model, but we know all the component parts of it. So um, how do we actually, how do you sort of start to get, um, businesses to recognize that they, they need to think about that bigger picture when they're maybe just trying to get, you know, the, the day to day existence, making sure that they're remaining growing, etc.

Sarah Heffron: Right. Yeah. For most scaling organizations, um, they're going to reach a moment and they're going to go, gosh, everything feels harder than it should or I've added a bunch of people but it's, it's things aren't getting easier or okay, I've just had an enormous win. This is great, I've won all this business, but actually I don't know that I have what I need to deliver on that because can I grow fast enough? Um, or you have a founder who kind of goes, God, it feels like everything has to run through me. I'm the bottleneck. And that's a signal that a more deliberate and scalable operating model is worth the time of unpacking and investing in. And if you do that again just thinking about how do things get done, how do we do this in a way so that I don't need to add another human being for every new thing so that I can really scale. Um, if you do that again, you're going to set yourself up for this longer term success. Whereas if you kind of just keep bolting on pieces here and there, you're just going to accumulate kind of the, that weight and eventually you'll either reach a breaking point or you're just never really going to achieve the growth that would otherwise be possible. Um, it doesn't have to be something big and heavy and you know what a giant multinational enterprise would have. It just has to be thoughtful, deliberate and specifically designed for your business. What does your business need? There is no magical template. There is no, I can't give you, you know, a toolkit and say, hey, this is exactly what it looks like for every single business. Because it is about your business, your people, your strategy, the industry you're working in, the geography you're working in, the regulatory environment you're working in, um, all of those things. So again, it is big to think about in the sense of it's all of these things together, they all relate to each other. But that doesn't mean it needs to be big and heavy. And so I think it's going to make things more effective and efficient. Now you never sacrifice effectiveness for efficiency because getting the wrong answer very quickly is not success. But with a well crafted operating model, you're going to get the outcomes you want as efficiently as you can. Um, because again, you want to get the best possible outcome, you want to drive your business, whatever success means for your business.

Tim Flagg: Um, so it sounds very foundational in terms of, it sounds like businesses might be, they need to understand the problems, they need to understand how to make those problems then go away. But they can't do that until they've actually understood the model and the processes. So although it might seem a little bit abstracted from the day to day running of the business, it's really important to understand that sort of holistic approach, our systems approach to the way the organization works. And I suppose that actually because it's so fundamental that you can't really get into using some of the AI powered tools and services that I'm sure we're going to come on to. Um, there's no point almost in getting those into your business until you understand what your business does. And this is a sort of a, you know, a lens on a conversation that we often have where businesses race to adopt an AI. Well, an AI as they may call it without really thinking about what it's trying to solve. Um, but this is almost like they're saying, go right back to the foundations, get your foundations right before you do that, which will help frame the business need better. So you can actually then deploy the right technology at the right time.

Sarah Heffron: Um, absolutely. That's absolutely right. If you look at your business, you just take the time. How is it working? Again, not on paper, but actually the people doing the work every single day, how are things getting done, what's working, what's not, what's connected, what's not? Then you can start to see, okay, if I want to adopt AI, which at this point I think just about everybody does, um, you'll have a good understanding of where can AI really add significant value? What are the parts of my business? What are the parts? What are my processes, what's my governance? What are all the bits and pieces where AI is going to add value, but what do I need to do to make sure that it lands on a solid foundation? The other thing that you'll find is there'll be things where you go, gosh, that process, it just doesn't work, or it's broken, or something's not working, or systems aren't connected. AI may be the perfect solution for that. AI offers us the opportunity to solve operational strategy execution problems that historically have been very time consuming, very expensive and very challenging. Um, but again, you just, you need to understand what you're working with and then you just plan thoughtfully and deliberately so that your people are still at the core of your business and you can enable them with an effective use of AI so that then you're really benefiting from that investment and you get a better outcome.

Tim Flagg: Yeah, no, the way you explain it makes a lot of sense now because, um, actually you're equipping organizations with the ability to really make sure that their AI adoption, their AI transformation is going to be really successful because they've asked the right questions, they've understood their own, their own model. So does AI, um, this is all the tools and services that we're seeing. Does it make the whole problem of transformation harder? Does it make it more urgent?

Sarah Heffron: I mean, I think there's definitely a sense of urgency that we're all seeing out there. Everyone's rushing to adopt it. One thing that I want people to consider and that I've started talking about, and I think there's a growing recognition is transformation sort of assumes there's a beginning and an end. I don't think we're in a world of transformation anymore. I think we are in a world of continuous adaptation, which, um, is different, which requires us to think about it differently. So it's not so much a matter of my transformation is more complicated, so much as where do I start and how do I continue moving forward in a way that is continuously evolving, but continuously adding value. Um, we know the technology is moving so quickly, the world is moving. Um, this is going to complicate things like change management, because again, it's a different type of change. It's not a, ah, here was the before and here is my clearly defined after, and here is my timeline with my clear milestones. So I do think it's going to involve changing the way particularly we engage with our people as you talk to them. What does this mean? And I think it's going to involve a little more transparency about the fact that there's a lot we just don't know. So what can you tell them? Are you going to take a human centric approach? Is this about redeploying your people? Is this about freeing them up for other work? Do you have a vision? Do you say, listen, we want to deploy AI? We actually don't know what that looks like yet. And so we're going to figure it out, and hopefully figure it out with your people, because again, I think you will get much better results by doing that, both from ROI and deploying it to solve real problems, but also bringing your people along. Um, you'll face less change resistance. And so I think that that's all, you know, it's just a different way of thinking about it because again, I don't, I don't think we're going to reach the, the end of the AI transformation. I, I think we're just in it now and we don't really know where this is going.

Tim Flagg: Yeah, it's a really interesting way you analyze that. Uh, I think you called it continuous adaptation. Is that right? Yeah, um, and that's a really fascinating way of looking at it because you're right, all of the language talks about, uh, uh, transformation going from A to B, as you put it. And you know, that sort of concept that it's sort of, that there'll be a complete point, but there is really no complete point to. With AI because frankly, we don't know what's going to happen next week, let alone next year in terms of the technology, the fundamental technologies, but also the way in which technology is going to be used. And there's so many other dimensions to it. So, um, but one of the Areas that we've often talked about is creating the mindsets, the sort of culture in an organization to be able to be adaptive and to try and iterate things and test and learn. That's a, uh, whole sort of mindset within an organization. So could you say that actually that the transformation that you're trying to build is more going from not having the mindset to having the mindset and sort of bringing the organization on that journey?

Sarah Heffron: Yeah. No, that's a great way to think about it because it is about changing the nature of your culture so that continuous adaptation becomes something, um, that's comfortable as opposed to, you know, a lot of people are sitting in a lot of fear right now. Yeah. And I think that if you can get to a place where people understand it, we're going to be changing. But that's okay. We're all going to figure it out together. Um, I think the nature of employee engagement, I think the nature of communications, I, um, think as well as making sure that you really are bringing all of your people along in building AI fluency and literacy. Um, not just here's how you write a good prompt. I mean, that's. Obviously they need that, but also, you know, let's really talk about it. Let's get some fundamental understanding what is it so that they can have a better understanding as things continue to evolve, that they can see and recognize what's going on. Um, again, not becoming technical experts, but just conceptually, what is it? What can it do? What are we aiming to do with it so that, again, it becomes more comfortable when we say, ah, here's a new. We're going to put in a new, you know, agentic AI. You know, over here in this part of the business, everyone goes, oh, I know what agentic AI is. Just conceptually, I get it. Um, and it doesn't sound like this scary thing that they're worried is going to take their job.

Tim Flagg: Yeah. I think you touch upon the most, um, important thing in this transformation process, which is that anxiety and fear. Um, so I'm thinking back to, um, Maslow's hierarchy of needs. Right. And I think right at the bottom of that, it's like safety and security and warmth and that can stuff humans. When you, when you think of that level, that's the sort of the most driving motivation that we have to stay safe and secure. And we see this a lot as we go around the country talking to, um, everyday people, uh, everyday employees, um, young people in particular. There is this real sense of anxiety and fear about AI. Um, Partly that's created by the media sensationalisms and some stories. And, you know, we're doing some work to try and bust some of those myths as well. But I think that, you know, this anxiety is there, um, and because of that, then when organizations try and bring in a new tool, they will often see that the usage of that tool, um, is very low, um, and that people are sort of not engaging with it because they kind of fear it a little bit or sometimes. We're starting to see now the apathy. People just, they say, well, so what, you know, and both of those are quite sort of negative, um, ah, sort of ways of approaching it. And then the adoption rate, um, doesn't increase. Although there's often then shadow AI usage where they're using their own tools outside of work, which is kind of on their own agenda, rather than using ones within work. So I don't know, what advice do you have or what sort of things have you seen within organizations where businesses can try and sort of engage more with their employees to help, um, make them feel more comfortable, more confident with using some of those tools?

Sarah Heffron: Yeah, so I think it goes back to sort of where I started. As you're figuring out your operating model, involve those people, include them early. The earlier they're involved in understanding, what are we doing? Where are we going? Hey, I want to understand how you do your job today. What are your pain points? What makes things hard? What slows you down? What makes your job hard? What's the work that is boring and monotonous that you just don't really want to do every single day? What else could you do? Because I think if you involve them early and you really engage them in the idea that AI is there, um, to fix their pain, to make the business operate better, um, to enable them to spend more time on the parts of their job they enjoy or the things that are, you know, require more of that human, uh, judgment and creativity and innovation, um, thoughtfulness. Um, I think you're going to find. What I'm finding is that organizations that really do involve their people early, um, and throughout really figuring out what is the plan for AI going to be, um, and are very transparent with their people, um, are having better adoption because again, people feel like they're part of what's going on. It's not something being done to them, um, which is an important sort of psychological difference in terms of, uh, am I part of the change or is this being done to me? Um, and I think you get earlier sort of change championship and people who say, oh this is great because this thing that we all complain about every day, we're going to fix it with AI, that's going to be great. Um, and saying that to the person who sits next to you every single day carries a lot more weight than some message from the CEO in fancy language saying we're going to become an AI first company and it's going to be great and we're going to have all this benefit.

Tim Flagg: There's different levels for that message though, right? I mean we look at the AI Skills framework as we call it and um, so there's uh, the CEO, um, there's the C suite. Then you've got the management layer with all the different functions and departments and then the everyday employees and across that, those different levels, very different messages like you say the CEO, they'll probably have a sort of top down mission statement, um, SOUNDBITE but it's important for them to set out that agenda, um, and to set that vision and to give permission to the rest of the organization. But then if you're thinking about the C suite, then they've got a different role to play. They've really got to start to interpret that vision into their own different departments and set their own leadership within those teams and same with their managers. Um, and I said there's all sorts of different levels of which that message uh, needs to sort of actually be um, set but also absorbed by the, by the individuals, organization. Um, and I think this comes back to the point you were making before though about making it relevant. So if people can see that it's relevance to their job then they're going to process it because addresses a need they've got. Um, but it's when you go in with a very bland one size fits all, um, uh, training course or kind of like a sweeping statement. This is going to fix everything. That's when people start to switch off I think and so not going to be on board. Um, so I wanted to kind of get to sort of how do you start with this process? Because what you explained is uh, your sort of where you come in and understand the operating model for a business and we talked about some of the solutions for how you sort of drive that uh, ongoing continuous adaptation. Love that phrase. Um, but how do you get started with it? When you go into an organization, what's the first thing you do?

Sarah Heffron: So the first thing is to try and get a holistic picture of, of what's going on. So that starts with three fundamental things. One is I want every piece of Documentation they've got, I've got, I want, you know, anything you have doc documented in terms of process governance. Do you have reports about how previous transformations or changes have gone? Um, what do you have in terms of data availability? Can I see your incentives? Can I see your okrs or KPI's? However you, you do that so that I can kind of understand that. I want to see what is your strategy? Do you have a five year strategy documented? So let me look at that. I want to see everything. There's no such thing as too much information for me. And then there's again two sort of groups of people that are really critical. One is leadership, whether that's management or executives, what is their perception of what is going on, what are the symptoms they're seeing. And then talking to the people who are doing the work, which is really when you start to figure out what's the actual root cause, what's actually because you may see something, you go, oh, well I have a people problem. And oh, you don't have a people problem. You've got a, you know, data availability problem. It's people aren't doing what you want because they can't see what they need. Okay, great, now we can figure that out. Or you've got this beautifully designed process and it turns out everyone does an end run around it because they figured out a way to just get things done more effectively through informal communication. Great, that's really good to know. Because then again, you start building this picture and you really understand what is it today. And then you can put it together and you can say here are, first of all, here are the things that are just potentially not working. And whether you use AI to fix them or whether you want to put AI on top of it, you probably want to be aware of them anyway. Where are your just brilliant opportunities for AI to come in and add value either to address these or again, this would be a great place. AI could add a lot of value here. Here are the very specific things that we're going to want to do that we're going to want to have documented that we're going to need to do with the data, you know, and your people. And we're going to need to rethink the relationship between the people and the AI. Um, again I emphasize taking a human centric approach to this. You're always going to need your people, um, and you come out of what I refer to as a diagnostic with enough information to be very clear and very confident. Ah, here's where I can go here's what I should do next. Here's what I need to do to be successful.

Tim Flagg: Yeah, great. Well, thank you for talking us through those steps to get started. That's a good framework to use. Um, I was fascinated when you were talking about your process. It sounds like it's quite work intensive and obviously you've got a lot of expertise to be able to go through and what you've gathered together all those documents to read through them, to process them, to um, analyze them against some of the methodologies that you and sampling M may experience. But I was wondering, do you use some AI powered tools within that process and are you almost thinking about replicating some of your knowledge and your ability to process all that information by using some AI powered tool?

Sarah Heffron: So I use some AI, um, in terms of particularly data analysis, obviously where I have massive amounts of spreadsheets, I find it incredibly useful to go, okay, you know, here's what I'm looking for or, um, giving an example of a completed analysis going, here are my inputs, here's my personal output, now you do it for me. So I do have a couple of AI tools, um, that I use for that. I also use AI to capture some of the information as it's coming in so that it's captured in a very standard format. Um, but a lot of this honestly AI can't do or can't do yet. Um, so one of the things that I am doing as well is for organizations that want it, essentially I'll take on an apprentice while I do the work so that they have somebody left behind who knows how to do this after I'm gone and they have somebody who can continuously keep an eye on the operating model. Because even if everybody agrees this is really important and we go through and we do it once again, we're going to have this continuous adaptation. And so you need somebody who's not focused on a single function but who knows how to kind of look across at the whole system. Um, and so use AI for where it adds value, for where it adds efficiency, for where it does the analysis. I certainly use it to format all of my documents. Um, I am grateful that I, I never have to worry about font sizes ever again. Um, but right now there's so much of this that really is about the people and it's things that aren't written down and they're not documented. And so you have to look for those things, you have to find those things. Even for a process or governance that's maybe extremely well documented, we have the process, we have here were the questions that came in. Here were the decisions that were made, but the nuance around those decisions, the context of those decisions, the judgment that was applied to those decisions. It's not usually documented anywhere. People don't write down all of the thinking they went through to make a decision. Again, those are the things that at least right now, AI can't really figure out. If it's not written down, if it's not documented, it's sort of invisible. So yeah, I do use AI where I can, but I also recognize that it can't do everything um, and it can't find those hidden pieces.

Tim Flagg: It's a great way of analyzing this and it makes me think of the way in which we see a similar thing happening within generative AI. So um, people will all see and I'm sure the sort of the videos which be created using AI and know this concept of um, uncanny valley, it's getting less bad. Um, but often when you look at faces um, that are being created using Gen AI that there's something not quite right about it. And we've got a couple of members who are actually working on the um, micro emotions. So being able to understand the very, very subtle things on the face that we're reading off each other's faces now. But actually computer haven't yet got the ability to um, process that um, and then to analyze and predict and then model. In a gen AI there's a whole range of companies trying to solve it. From being able to get actors to um, filming actors with those micro expressions, bringing it in, understanding them and then replicating them in avatars. But it's the same sort of thing. It's exactly the same sort of thing that you're talking about, which is the machines, um, when they're processing all the data, they're greater that bit. But when it actually comes to being able to understand the institutional or organizational history, the relationships, the politics, all of those things that go on in the organization which are fundamentally human, the machines haven't got the ability to even listen, um, let alone process, analyze and then predict all of those things in the organisation. So it has to come back to the human again. Uh, which I suppose is the bit that then you, you're able to accept into that organization and read those, those signals and hidden, hidden messages.

Sarah Heffron: Yeah. I think it's also you know, one of those things as people think about how to deploy AI in their organization to make sure they're thinking about the fact that they need to continuously make sure they're building More and more people who are still going to have all of that context and all of that, who will build that institutional knowledge, who will learn the institutional judgment. Um, because again you're always going to need people, you're always going to need humans. Um, and so I do get a little worried when people say I'm just going to eliminate this entire category of whatever type of role. And I go, you're going to lose things. If you just wholesale try to eliminate the human being, um, or at least as AI is today, you are, you are going to lose something. So I just think it's so important to really think about that all of those things that humans bring and what humans, you know, again it's I take a human centric, AI enabled, let's use AI to enable us to do more, to do better, to get better results, um, to extend our reach and our impact. Um, but you know, it can do everything.

Tim Flagg: Yeah, we're not redundant yet. Humans are still an important there. But I think going back to what the way you framed this I think is really important because it's about understanding the operational model of an organization, but within that, therefore you have to understand the role that only humans can play. Um, so whether that's the input, um, or whether that's the sort of relationships or that's the organizational knowledge we talked about, but also even in sort of the programming of the uh, of any AI powered tools that you want to bring in, you have to understand how that's going to impact on humans and how you're going to get the adoption in the organization. So so many of the challenges which we're facing in order to bring AI powered solutions into an organization rely upon the human. Um, so it really, you can't get away from that. Um, so just as we start to wrap things up, I wanted to ask you what's the sort of, the one thing that you're really excited about as you look ahead over the next six to 12 months? Um, are there particular things which are happening, um, either with technology or more broadly within your space that you're really excited about, you think are going to change how we view, um, how we view using AI?

Sarah Heffron: Well, I am first, um, of all enormously optimistic that we are going to find ways of using AI to address some of these operational and strategy execution problems that again, historically I would have needed to bring in 100 consultants and build a whole transformation office and it would be an enormous big thing. And now few AI agents, carefully crafted and thoughtfully deployed, can address the same need. But more Fundamentally, I think I am seeing a bit of a shift as more and more people talk about the operating model as the necessary foundation for AI. So I think more and more companies are recognizing, more and more advisors, um, and AI, uh, implementation firms are recognizing they're being asked to build a house on unstable ground and getting that operating model foundation right. And really then building on something solid is what's delivering the better outcomes. We're starting to see the differences between organizations who've done that and those that haven't.

Tim Flagg: Yeah, I think it's really interesting and um, I found it really valuable to um, understand more about the operating model here because actually it brings in the conversations we often have around data. Um, you know, we talk about the foundation role of getting good data, clean, ah, data, well structured data, etc. Yes, that is important. We also talk about the skills which people need to have the organization. But those are the sort of the component foundations, if you like it. I'm thinking about building a house here as an analogy. Those are the foundations literally that go into the ground and then the structure. But unless you have an architect's plan, um, for what the house looks like, then you don't know what you're doing, um, and the house will turn out as a complete mess. Um, and I suppose that's where the operating model comes in. In that you're sort of saying that you have to have that operating model, um, and you need to know what are you building here? So you wouldn't start. An architect wouldn't just start, uh, drafting something without knowing what kind of house it is, what kind of building it is. Um, so I guess that's the way to understand this as well. Um, so just as kind of last of the question then, in terms of how can we follow, uh, what you're doing and stay in touch and what have you got coming up over the next few months?

Sarah Heffron: Yeah, so, um, I, obviously I have my LinkedIn people can find me there. I do have a website as well. Um, and for the next several Fridays, I'm actually offering sort of free office hours. Bring me your operating model questions. So if you go to my LinkedIn, you'll find a link to book time with me. Just book time any of the next several Fridays. Um, bring your questions and let's have a conversation.

Tim Flagg: Fantastic. Well, what a, what a great offer. Thank you for sharing that with the, the audience and I hope they take you up on it because I, I can see already I've got a lot of value out of this conversation and I'm sure they would as well in terms of understanding how it could be relevant to their business. So thank you so much. We'll put those, um, link the in the comments there as well. Sarah, it's been really interesting. We've gone through many different topics, but I think we've come back to the importance of being human centered in everything we do. And that's reassuring. But also it's really sort of, it's a great sort of message that we can also take back to the organizations that we're working with. Um, so human centered AI enabled and the other phrase which kept on coming back to was that continuously adaptive. And I like the way that we got into quite early on talking around that as a paradigm rather than just transformation, which does kind of give the uh, perception that it's going to be done at some point. And actually you've made the point, you know, that actually doesn't really ever get done. Um, so it really has been very valuable to talk this through. Thank you for sharing all your knowledge and experience with us on this edition of the podcast. Thanks very much.

Sarah Heffron: Thank you very much. It's been.

Narrator: Don't miss the next episode. Subscribe now. UK AI is the trade association for businesses across the uk. Tech and non tech, large and small. AI is our business. Find out more at UKAI Code.