Luke: [00:00:00] You’re listening to a new episode of The Brave Technologist, and this one features Dr. Mohsen Ghasempour, who’s a distinguished expert in the field of artificial intelligence, currently serving as the Chief AI Officer at Kingfisher, where his leadership contributes to the advancement of artificial intelligence and machine learning technologies within the organization.

With over 15 years of experience as an accomplished data scientist and visionary leader, Mohsen has consistently harnessed the power of data to drive strategic decision-making. In this episode, we discussed focusing on solving problems with AI versus using AI for the sake of AI. Where personalization becomes helpful versus creepy, how AI as a tool is implemented across a global multi-brand company.

And now for this week’s episode of The Brave Technologist.

​

Luke: Dr. Mohsen, we- welcome to The Brave Technologist. How are you doing?

Dr. Mohsen: Good. Thank you for having me.

Luke: Yeah, yeah. So we’re here at the, AI Summit in London and you had a speaking, engagement this morning.what was the core message you were trying to [00:01:00] get, convey during your keynote today?

Dr. Mohsen: so I mean, I’ve always been lucky to be able to share the amazing work, our team do at Kingfisher on AI. But I think the core message was, the most of AI success that people are talking- Mm-hmm … is less about technology, it’s about execution.

Luke: Mm.

Dr. Mohsen: So I think today we share how you can actually bring, that technology to life.

and sometimes it’s not about how complex is that technology, it’s about focusing on solving the right problem.

Luke: Mm-hmm.

Dr. Mohsen: Uh, and I think, you know, for us, the philosophy at Kingfisher is AI is just a tool, not the mission. I think focusing on solving the right problem and the benefit is preparing some, some.

Luke: Excellent, excellent. Yeah, and I think, you know, you spent over 15 years kind of at the intersection of data and science and business transformation. Was there kind of an early moment in your career that convinced you around AI or these tools kind of fundamentally reshaping how companies operate?

Dr. Mohsen: I mean, yes, I, I had one interesting moment. It’s less about AI because you, I mean, AI is a new term these days, right? Right. [00:02:00] Right. Right. but I think around, uh, a couple… I mean, early age, its age, After my PhD. Mm-hmm. I did my PhD in computer science. Okay. and I did it at, Advanced Processor Technology Group at University of Manchester.

So at that time, I was working with a human brain project, so we’ve been working on a very, I mean, sophisticated, challenging- Of course … right, high-pressure type of, environment. after that, I joined, a kind of a company work on, special education.

Luke: Mm-hmm.

Dr. Mohsen: so the idea was how you can help with kids with severe disabilities- Hmm

uh, to learn.

Luke: Very cool.

Dr. Mohsen: which was interesting. However, for me, it wasn’t that complex. So for me, the algorithm we used there was a very big difference between what I’d used during my, during my PhD. However, one day we receive an email, from a mother, and the content of email was interesting.

it was, of course, thanking the technology team, but what she said was, “This was the first time after 20 years that I have evidence that my kid is learning.”

Luke: [00:03:00] Really?

Dr. Mohsen: So, and that kind of stuck with me because, before that, my focus was the more complex, more challenging of a problem to solve is better, right?

that moment is not about the technology, right? Mm-hmm. It’s about the impact you’re making. So fast-forward now, I… Of course, AI is the most promising, powerful technology we have. I still think it’s not about technology, right?

Luke: Right.

Dr. Mohsen: I think for organization, every organization, including us, exists to solve a problem for a cause sake.

the problem we solved at that time was for parent for kids with severe visibility, today we solving different problem, different organization solving different problem. I think the key here is focusing on solving the right problem because when you solve the problem, then people and customer feel it.

And I think, AI just make it much easier than before.

Luke: Yeah.

Dr. Mohsen: So, and I think that moment, is still referred back, and I think that’s behind our, uh, philosophy, that AI is just a tool- Yeah … like many other tools we have, but not the mission.

Luke: Yeah. No, I think that’s- that’s great, and it’s cool [00:04:00] that it kinda connects together with what you guys are doing today ‘cause, I mean, there’s so much talk about AI, and it’s like, okay, what are we actually doing here?

Sure. Right? and so I think a part connects to a lot to just a more practical use of like, okay, everybody knows it’s here now, what is- were we actually doing this for? Yes. where is it actually useful, you know? so Kingfisher, like, let’s kind of get into that a little bit.

They’re a home improvement retailer across multiple brands, right? Like, um, and markets. Like, did you have to make kind of a case internally for having a, um, a chief AI officer? Or like, what was that process like?

Dr. Mohsen: not really. I mean, to me, the chief AI officer role is less of a personal recognition.

Mm-hmm. I mean, of course I’m very humble to have that title, but I think it’s more of a reflection of, company maturity and commitment- Mm-hmm … of using AI across, uh, the operation. So that telling, everyone that AI is a business critical for us is not that we’re gonna run isolation PU- isolated PUC and experimentation.

This is [00:05:00] part of how we operate. And when you look at, at the moment, AI Kingfisher from supply chain, personal experience, on-site AI agent, market and commercial team is using it, customer service using it. So when you see AI embedded in your day-to-day operation, that kind of feel a need of somebody that help with the strategy.

as you mentioned, we are an international multi-brand, multi-country-

Luke: Right,

Dr. Mohsen: right … type of brand, and as a result of that, I think we just felt we need, the role that can glue everything together.

Luke: The Brave Technologist is brought to you by the Brave Search API. Access billions of indexed web results from a simple API call with the Brave Search API. Join the leading names in AI and tech using the Brave Search API to power agenic search, keep LLMs current with real-time data, train foundational models, and bring the best of the web directly to the leading edge.

Get started today at brave.com/api.

It’s a all-encompassing kind of thing, too, right? I would imagine you, you’re dealing with internal use and how this interfaces with the world, and like you said, like, uh, [00:06:00] international, so all these different, you know, types of compliance or whatnot, you know, in, in all these stages in there.

So I’d imagine it’s quite a lot, at hand. I mean, walk us through… I know you talked about kind of like solving problems, right? what’s the big strategy with Kingfisher with AI right now that you’re kind of focusing on? is there criteria you’re looking at, where AI can kind of move the needle?

I know we’re talking earlier a little bit about solutions. Let’s unpack all that a little bit more.

Dr. Mohsen: Yeah, of, of course. So if you look at our, our brands, we all on the home improvement, industry, but we have very diverse and unique brand. We have Screwfix, which they are Main cohort of the customer base are trade professional.

These people which are like speed is money for them, so they do it for a living. We have, B&Q and Castorama, which are just general DIYer, that they just, like, probably need a bit more support on how you get the job. But also we have Brico Depot, which are like our discounters, which is efficiency and the lean operation is the key.

So we don’t have a single AI strategy. Our strategy is to support our grand [00:07:00] strategy. We exist to empower our brand to reach, their strategic goals. and part of that is for us to, to go to different functions, making sure we identify the pinpoint. efficiency gain is the key. So, uh, I mean, retail is a complex o- or, you know, o- operation.

I like to describe it as a orchestra.

Luke: Mm.

Dr. Mohsen: Yeah. So every single function needs to play in harmony-

Luke: Yeah …

Dr. Mohsen: in order for us to deliver for our customer. so technically our strategy is to go around, try to identify dependency, try to identify the where we can add the most value-

Luke: Mm-hmm …

Dr. Mohsen: uh, which is either solving a problem for our customer, making a better customer experience, empowering our colleagues to deliver our good, the best, but also automating some of our operations, right?

So at, as a whole, as an organization, we can be leaner and we can move faster.

Luke: Such an interesting kind of use case, too, around home improvement and these types of things that have been around, and, um, people haven’t necessarily thought about them from kind of a technology perspective. But, like, you’re in there, right?

you’re kind of, like, having to w- see how this [00:08:00] can, you know, remove pain points or do these things and, and kind of have this holistic view. how is that? Like, working, like, where meets, you know, new and old worlds kind of meet on this front?

Dr. Mohsen: So my, my personal opinion in this is, I, I know, I mean, we are, we are a retailer, right?

Yeah. And when you ask us, we just, we don’t see us as a technology company. However, I don’t think we can, or retailers moving forward in the world of AI, can avoid behaving like a technology company, right? Right. So of course within Kingfisher we had, uh, a very supportive leadership.

Luke: Mm.

Dr. Mohsen: And then four, five years ago we had start establishing a central AI function, because we wanted to own, the competitive advantage that we’re having through this technology.

and now we have around over 50 AI engineers- Wow … that we can build in-house solution, right? of course we still don’t see us as a tech company, but that tell you that we are still, you are still that new to have that capabilities- to be able to own your destiny, right? So AI is moving so fast, that, you want to control at least part of that.

You have great partnership, with [00:09:00] different, different partners of giants in this space, but there’s still, there is part that we feel we need to be more reactive-

Luke: Yeah …

Dr. Mohsen: uh, and proactive, right, so in, in owning that space.

Luke: Interesting. Yeah, yeah, yeah. I know. It is moving fast. It’s also, like- Yeah, like you were saying though, like what problem are we trying to solve?

Like, I keep going back to that ‘cause I think it’s just really, relevant, you know? and I know it’s like part of like you know, bringing this strategy to life and inevitable questions start to come up around measuring ROI around these things. is it around like efficiency?

Is it around revenue? how are you guys thinking about that right now?

Dr. Mohsen: I genuinely believe, industry making, measurement of ROI around AI more complicated than it sounds, right? So it’s the same. AI is just piece of software. We measure it the same way we measure everything else we do in our investment, right?

Luke: Right.

Dr. Mohsen: So we’re measuring the impact on revenue, we’re measuring the impact on, margin improvement, so efficiency gain, cost saving. of course, it happens to be AI- Yeah … that help us to do that.

Luke: Right.

Dr. Mohsen: Uh, but again, coming back to the [00:10:00] first point, right, is there is nothing like AI ROI because it’s not about AI.

Luke: Yeah.

Dr. Mohsen: You try to solve something, and if you don’t know what problem you solve, what is gonna be the benefit of that, you probably shouldn’t solve that,

Luke: right? Right,

Dr. Mohsen: right, right, right. And that’s the first question, right? Doesn’t matter what you use to solve that problem. If you want to improve the customer experience, you should…

You just need to know, okay, what is your testing methodology? Is it AD test? is it the cost, reduction? What is your current cost? How do you do? So I think, fundamentally, we go back to the first problem. I think we just need to not to be distracted-

Luke: Mm-hmm …

Dr. Mohsen: by the, the actual technology, which today is AI, tomorrow maybe quantum, I don’t know, whatever.

Uh, the measuring ROI is exactly the same in the business.

Luke: Yeah.

Dr. Mohsen: You do- you’re just either improving revenue or your margin, or reducing costs, right? Or efficiency gain. So, and the same way you measure it using any other tool, you, you

Luke: just use the same point. Yeah. Makes sense. Like, don’t need to make new- Exactly, right

Dr. Mohsen: new

Luke: API or KPIs, reinvent the wheel. We’re at a conference, everything’s APIs. [00:11:00] it’s coming out. so I know that, like, kind of from a privacy perspective, like, there’s a lot of customer data, like, going on with AI, you know, at Brave, we, we think about this a lot.

Um, but obviously, like, we’re kind of on the, more on the, user end with browsers and things like that. what do you think about the tension between building really good personalization and kind of keeping that balance around being stewards over data or handling it, in a way that’s, like, both practical and kind of, preserving and,

Trying to avoid, like, a lot of the liability or pitfalls that come with, you know, having to custody data?

Dr. Mohsen: that’s interesting question. I think, there is a very fine line, between being helpful-

Luke: Mm-hmm …

Dr. Mohsen: and being creepy.

Luke: Right.

Dr. Mohsen: Okay? Right,

Luke: right.

Dr. Mohsen: So, customers appreciate a personalized experience when it’s helpful.

So when you go to DIY.com, buy a drill, and we just offer you DrillyBits, it’s just like, “Oh, okay, I need that.” That feels helpful.

Luke: Yeah.

Dr. Mohsen: customer feel uncomfortable when, you know, when they feel they are being watched rather than being served.

Luke: Mm-hmm.

Dr. Mohsen: and I think some- sometimes organization get it [00:12:00] wrong.

The fact that you have a lot of data from customer, it doesn’t mean you should use it.

Luke: Right,

Dr. Mohsen: right.

Luke: Right?

Dr. Mohsen: Right. So I think the idea here is how you can use the data you have, from customer who trusted us with that data. How you can use that to just add value, right? And give them something useful.

Uh, within Kingfisher, this is exactly about how we operate. Again, The question is, are you trying to be helpful for the customer? Is it the right recommendation that he’s showing? Is it the right time? and I think as simple as that. So technically, we have a very simple test. if you ask this question, if customer knows exactly how you provide that particular recommendation, are they comfortable?

If they comfortable. If they are not comfortable, that means you’re crossing the line, right?

Luke: Right,

Dr. Mohsen: right, right. I mean, again, coming back with the super rule, try to be helpful.

Luke: Yeah.

Dr. Mohsen: we just need to be careful we are not intrusive.

Luke: Well, and there’s ways, too, I mean, like, I think in the marketing side, people overthink that personalization thing- Yeah

a lot, right? Like, I mean, if you have a customer that’s buying a, a drill, like your example, right? Yes. [00:13:00] Like, oh, I know that the shelf life is, is usually gonna be this long. Like, you know, good recommendation is when they hit that point, you know, maybe I recommend a new model or something, right? You don’t have to, like, collect everything about them to, like, be helpful.

Dr. Mohsen: exactly, and that’s again being helpful, right? Yeah. So sometimes you, you try to oversell, right?

Luke: Yeah.

Dr. Mohsen: if customer feel, I mean, because we are all human, right? So we, we… I mean, you feel, you know, for example, I mean, it happened to me. I was trying to look for something else. You go to YouTube, and you see a, a recommendation, just like, wow.

Right? It just, just feel uncomfortable, right? Yeah. So but when you try to finish, in our case, we need to help our customer to finish their project. So there are some necessary things they need to finish their project. When you’re recommending those at the right time, so for example, at Screwfix, we have a very cool recommendation for our trade people that we can predict if they’re running out of a consumer, right?

As a trade person, you have a, a sealant.

Luke: Yeah.

Dr. Mohsen: And you keep buying them every three weeks.

Luke: Yeah.

Dr. Mohsen: the worst thing that can happen to you, once you go to the work, and then you don’t have a sealant, right? Right, right. Like, because it wastes time. So what [00:14:00] we do, we try to predict when you’re gonna run out of sealant, and then we try to proactively recommend, oh, we think you’re gonna be run out of sealant in the next 30 days.

So you see, you have to just think, are you being helpful-

Luke: Yeah …

Dr. Mohsen: for that cohort of customer, right?

Luke: I’d imagine for those types of consumables too, like with things that are happening that might be disruptive, right? with the supply chains and things like that. being able to kinda know the customer, but then also, like, look at that broader scope and say, “Okay,” like, “Hey, you know, maybe it’s good to buy a little more now or stock up or whatever,” right?

Like, when based off of what’s happening in the bigger picture.

Dr. Mohsen: E-exactly. The whole purpose of customer trusting us with their data- Mm-hmm … is to being helpful. Mm-hmm. Right? Yeah. If you use it for any other purpose, that’s crossing the line. That coming back to your question, that, that’s just like, that’s not the purpose, right?

Yeah. So I think we just need to be grounded. Is it the right thing for customer and then business?

Luke: Makes a lot of sense, I think. are there any other, ethical questions kinda coming up for you as AI’s hitting, aside from kinda this creepiness factor or anything like that?

Dr. Mohsen: In, in all the [00:15:00] organi– I mean, d- d- that can depends on the organization.

Yeah. But more and more that AI come into helping with decision-making- Mm-hmm … and it going everywhere, right? So you have to be mindful of, again, coming back to the basic, is it the right thing to do? Uh, ho- however, some, some organiz- if you are in credit risk that you’re making decision that impacting customer life, then you’ve gotta be in the completely different league-

Luke: Right

Dr. Mohsen: uh, than, for example, in our case that, you know, is a, is a, is a probably lower risk area. Uh, but again, just being mindful, have a governance across all, all the domains of what we do for customer service from, you know, customer experience on the website, from back in operation. I think the ethical question always is on top of the list, that for every application you have to ask Easiest of right things to do, and that’s all around putting AI governance around open system Mm-hmm.

Luke: Is it tough to keep track of all that, in this global space where you’ve got, like, Europe’s kinda having their own set of rules, and US is doing other things and, you know, trying to honor operate in this global environment?

Dr. Mohsen: Yeah, exactly. [00:16:00] So, and, and that’s a good, that’s a good point because AI Act is something that coming to play is already here.

So at within Kingfisher, we are, compliance with AI Act, and we try to… And also we have European brands- Yeah … so we have to make sure we actually comply across the region. but also one very important thing in this space is historically, especially, you know, working few years back, so we used to build solution, and then you go and take the compliance of legal approval.

But today you can’t work like that. Today our compliance and legal team are part of our development team. Right. So from initiation of AI project, they’re gonna be with us. Yeah. So they’re asking the right question that they saw before even developing a tool, right? And that make everything much easier because at the end of the day, you feel much more secure and, you know, comply with everything.

Your legal and compliance are also feel comfortable because they’ve been part of the journey, and as a result you can just scale it, right, so it’s- It’s

Luke: good to see that maturity in the space- Exactly. Yeah … right? Like, compared to where it was not that long ago even, you know. so I, I think, being a multi-brand, kind of [00:17:00] multi-country operation, there’s kinda this big debate right now around, kinda large centralized AI models, right?

Like, and, and whether the right architecture for enterprise. What’s your take on that, like, from your point of view?

Dr. Mohsen: So that was an interesting point that we covered also in the keynote this, in the morning, right? Uh, so I think it’s not a binary decision. many people talk about centralized or decentralized.

The simple answer is you want to centralize capability and decentralize innovation, especially for us, which are multi, you know, international, multi-brand with different focus. you want to make sure your brand have autonomy to innovate, right? Because they know their customer better than anyone. You want to make sure they’re empowering them.

at the same time, you don’t want to build the same thing 10 times- Right … across 10 different entities, right? So I think there’s a fine, finding that balance is the key, right? So in– at Kingfisher, as I mentioned, four, five years ago, we decided to build a central AI function-

Luke: Mm-hmm …

Dr. Mohsen: which are so focused to build capabilities.

So we built Athena framework that [00:18:00] help us to deploy AI agent at a scale in the safe and secure environment. And technically solve that problem of security and safeguarding for all of our brand. We didn’t need to actually go to individual brand to solve that. At the same time, our brand have full control on our AI agents, what they do, how they do that, what is the behavior, right?

Same for our personalization engine, which is again, we deliver 165 million attributed revenue last year. That’s across all the brand. Again, we have a central capability, which is fine-tuned for every brand. and we just discussed about the Screwfix example, which is very specific to trade customer for Screwfix.

Luke: Oh, yeah.

Dr. Mohsen: And B&Q may not make- need completely different, personalization algorithm. They both running on the same platform, right? So from a group, the cost, security, speed-

Luke: Yeah …

Dr. Mohsen: comes from centralizing the capability, but decentralizing innovation I think is essential.

Luke: Yeah, it’s like you have decent- you decentralize problem sets, right?

And sometimes one might have a certain approach, that works better for that specific [00:19:00] problem set than the other, right? Like- Exactly. Yeah, no, I love it, man. Like, this is, like, so much more practical than a lot of these, handwaviness things are in this space, you know. so what…

Looking in the future, what’s getting you excited around future capabilities that you might be able to unlock at Kingfisher?

Dr. Mohsen: So coming back to the point, as I say, I generally believe retail is an orchestra, right? Yeah. I think one of the, one of the thing we can unlock, when you go to the area of agentic, and I generally believe agentic AI is one of the most overused term.

Sure. But if you look at the true meaning of agentic capabilities, which means we have autonomous agent that can talk and interact, so in, in a community of an agent, I think that can revolutionize our operation. At the moment, we have supply chain, we have customer service, we have e-commerce, marketplace.

Luke: Mm-hmm.

Dr. Mohsen: and sometimes, we solving some e- problems in isolation, and I think what would be exciting, in the future when we are a bit more mature of using AI agents, those agent can talk, uh- Yeah … more freely on the background, and as a result [00:20:00] of that, you start optimizing the full end-to-end equation rather than individual functions.

Luke: Mm-hmm.

Dr. Mohsen: So I think agentic AI, if it’s done properly, with the right integration, right, I think that can revolutionize the way we operate.

and I generally think, uh, the biggest, uh, success of an

AI is when nobody call it AI anymore. Right.

Luke: Right. Right? Exactly.

Dr. Mohsen: So, I mean, these days you don’t wake up ev- every day and just, uh, excited about the internet.

Luke: Yeah,

Dr. Mohsen: right. You can’t operate without internet, right? So I think, uh, hopefully in couple of years when people just used to this technology, and they don’t…

Because at the moment, many, many distractions because we are overexcited-

Luke: Yeah …

Dr. Mohsen: about the technology. Yeah. But as soon as you accept this is part of your life, you can’t operate without that -

Luke: Yeah …

Dr. Mohsen: I think that’s where you see the actual, uh, transformation, right? Because people are less distracted, and they’re focusing to just embed it in, in their workflow.

So, and probably fingers crossed. Hopefully

Luke: it’s gonna happen. Fingers crossed. Yeah, of course. No, I think, yeah, that would be the biggest win is, is to kind of like abstract away the buzzwords, right? Yeah. And like, and we’re just using this stuff. No, I [00:21:00] think it makes sense. Um, where, where can people follow along with you, where, where you guys are doing…

Uh, uh, are you guys putting work out there? Can you, people find you on social or anything like that?

Dr. Mohsen: We are, we are, we are very active as a Kingfisher group, right, on LinkedIn. We share about all of our, kind of findings. So recently we had some consumer research we publish on, Kingfisher’s sort of feature.

If people find it, I think they can see what we talk about that. But I think mostly LinkedIn, we talk, and we are very active on that domain, sharing what we do across our brands.

Luke: Excellent, man. Well, really appreciate you making the time. It’s great conversation and, uh, yeah, love to, love to pick it back up and see how you guys are growing in the future.

Dr. Mohsen: Many thanks for having me, Liam. Take care.

Luke: Thanks very much.

Dr. Mohsen: Thank you.

Luke: Thanks for listening to the Brave Technologist Podcast. To never miss an episode, make sure you hit follow in your podcast app. If you haven’t already made the switch to the Brave browser, you can download it for free today at brave.com and start using Brave Search, which enables you to search the web privately.

Brave also shields you from the ads, trackers, and other creepy stuff following you across the web.