DataTopics: All Things Data, AI & Tech

#90 How To Make AI Deliver Business Impact: Gaelle Helsmoortel on Turning Strategy Into Tangible Results

• DataTopics

Share episode

Send us a text

In this episode, Gaelle Helsmoortel joins us to discuss how to make AI truly deliver business impact, not just proof of concept.

With over 25 years of experience spanning L’Oréal, startup leadership, and her current role at Dataroots, Gaelle shares her approach to turning business challenges into measurable value. She breaks down her proven 5Ps framework (Purpose, People, Process, Platform, and Performance) and explains how companies can bridge the gap between strategy and execution to generate real results.

🎧 You’ll learn:

  • Why most AI projects fail (and how to prevent it)
  • How to move from proof of concept to proof of value
  • How to align business purpose, data, and people for maximum impact
  • Why “purpose before platform” is key to successful AI adoption

Whether you’re a business leader, strategist, or data professional, this episode will help you understand how to make AI work for business and deliver tangible results.

🔗 Connect with Gaëlle:

SPEAKER_00:

Today I'm joined by my new colleague Gail Helsmortel. She recently joined Data Roots, and I'm excited to dive into her career story, her mission here, and how we're shaping Data Roots end-to-end approach together. Welcome, Gail.

SPEAKER_01:

Thank you, Ben, for inviting me. I'm very happy to be here.

SPEAKER_00:

Very much looking forward to getting to know you a bit better. Can you start by telling us a bit about your career journey so far?

SPEAKER_01:

Yes, of course. So I started my career more than 25 years ago, and the first 13 years of my career were in a big corporate enterprise, L'Oreal. So I did uh 10 years marketing management there and three years recruitment director. But that's really where I focus on business, sales, marketing, selling, you know, how do you make money? But after that, I entered the advisory world, the consulting world, and uh during seven years I guided major uh companies as well, but in their digital transformation. Digital transformation at that time it was content marketing, social selling, because it was from 2012 to 2018. And after that, I I founded and uh led my own uh tech startup with two partners of mine. It was in the tech retail, so it was really technological software uh company, and uh I did that during five years, and now it's a little more than three years that I am doing uh consultancy advisory to help companies integrate AI successfully in their organization.

SPEAKER_00:

Looking forward to all three parts, I would say. Let's maybe start at the L'Oreal side because you mentioned that you went into consulting advisory. What was the trigger for you?

SPEAKER_01:

Well, we know when I start my career and when I was young, uh I always said, okay, I will work five years in a major company and then I will start my own company. And I need challenges. As we all do. And I need challenges. And you know, I was uh at L'Oreal, and at after 13 years, at that time, as well, I was uh 37 years old, I really thought, wow, either I stay here my whole life or I quit now. And I said, no, I need to quit. I need L'Oreal, it's a fantastic company, but I needed to see something else, to see other markets or challenges. And so that's what I do. And advisory was fantastic for that because I had a major background that I could use in order to help other companies, other markets to uh evolve.

SPEAKER_00:

And so the advisory part was more in the space of digital transformation. Yes. Did you already got some first signs of those bigger digital transformation projects at L'Oréal, or was it really a first step into that world?

SPEAKER_01:

It was not a first step into that world. Um I had the click of digital data when I was at L'Oreal in the HR. So in the recruit, in my recruitment role. Very strange for a lot of people, but why? Um, because at that time, as a recruiter, we were using LinkedIn. What was LinkedIn at that time? So it was 2008, 2009. It was a major Bible of CVs. And when I have seen that, I said, this will change the world of sales and marketing. I was sure of that. Um, and next to that, L'Oreal launched at that time the first e-gaming for recruitment, meaning that um, well, everybody could play the game in order to eventually have a job at L'Oreal. And I found that fantastic because it was a huge opportunity to discover not only, you know, the people who are the perfect CV that we wanted in order to get an interview, it was much more like that. So I found it was fantastic, and that was okay, business data, digital. So that's where I had the kick. And I said, okay, now I will do something regarding digital transformation. Um, and I was very um, very uh tightly following what was happening in the United States and in Canada, and that's how I entered the digital world.

SPEAKER_00:

It's cool to see when you look back at some big companies, the types of games and so on they developed back then. They were ahead of their time. And there are many stories of that, and I always like to hear it. Um, so you went to the advisory role, and then after some years, you started your own company, right? What's it called again?

SPEAKER_01:

Genius.

SPEAKER_00:

Genius. And what does the company do? Could you elaborate on that?

SPEAKER_01:

Yes, we um we provide our customer our retail uh shops. So they are they are um chains of shops, whatever the shops was selling, but it was chains of shops because what was the big issue for those big chains is that when you have 100, 200, um, 500 shops, you have a lot of pounds of sales, so a lot of data. And it's very complicated for those companies in real time to collect all those data in order to very closely adapt their marketing, their sales, and make business decisions. And that is the technology we came up with. So the the possibility to collect all those data in real time, and on the other end, the business people at the company had a front end in order to make in real time all the analysis they could do in order to make the right business decisions.

SPEAKER_00:

And so today the company still exists yes. Okay, cool. Um looking back at that time, do you have particular achievements or cool aha moments that you can share with us?

SPEAKER_01:

Yes, I had um a genius. So when I was uh CEO of this startup, I had uh two major haha moments. Um because I did two fundraising. Uh the first one I uh funded 1 million euros. It was in um March 2020, and I don't know if March 2020 rings a bell.

SPEAKER_00:

Always rings a bell, unfortunately.

SPEAKER_01:

Exactly. But it was in the middle of the lockdown, and our audience or our customers or prospects or retails of shops, meaning that they were all closed. Uh, but eventually I found uh two um professional investors who accept to um invest in our company, and in June we could uh set up the deal. So that was amazing because nobody believed in it. Uh, even my two partners said you will never succeed in raising money now. And I said, okay, maybe not, but uh so that was the first major aha moment. And the second one is the second fundraising round. It was uh then I I raised 1.5 million euro, and then it was in March 2023. This rings another bell, perhaps. It was the start of the war between Russia and Ukraine. So, meaning that it was a time where all investors were very cautious and a lot of investments were frozen. But there again, I I could raise uh another 1.5 million. So to be honest, I'm particularly proud of those.

SPEAKER_00:

I can imagine. Never waste a good crisis, right?

SPEAKER_01:

Exactly.

SPEAKER_00:

And so next to that, I can imagine, because that was very challenging. Next to that, that you also have challenging on the side of, for example, uh user adoption in or getting clients or anything else related to the software?

SPEAKER_01:

Of course. You know, it's hard, uh, such a startup when you start from zero, it's it's hard to um to get new customers to create this trust because of course it's a lot of trust and there is data, a lot of data. So um our main um challenge was not the technology. Okay, of course, technology was managed, but my two partners were very good in technology, they were top-class technologists. Um, but you know, we had this interface for the business persons, and this was our main weakness because it was too complicated, still not user-friendly enough. And um and you know, if it's not user-friendly enough, it's always a break. So, of course, we improved it, but it's not it's not improved in one night or one day, so it tooks it took time, but this was really um our first uh our main challenge to make it easier to use.

SPEAKER_00:

A key lesson to remember for today's podcast episode.

SPEAKER_01:

That's right.

SPEAKER_00:

Uh and with all that experience uh uh behind you, um, you've now joined Data Roots, you're sitting next to me. Um why did you join Data Roots?

SPEAKER_01:

Well, I joined Data Roots because I I wanted to come back to advisory, uh to um the consultancy. Um I like to sell by bringing value. What is important for me is bringing value. And um uh data roots um has super experts and super technical professionals. And uh when they ask me to help them to develop the consultancy advisory part in order to address you know top companies and how can we bring value to them, of course, using technology and particularly artificial intelligence. So that's that's what interests me. It's bringing value to our customers.

SPEAKER_00:

You saw a lot of potential and you were thinking we should leverage this more.

SPEAKER_01:

Exactly.

SPEAKER_00:

What stood out to you about data roots?

SPEAKER_01:

Excuse me?

SPEAKER_00:

What stood out to you at Data Roots?

SPEAKER_01:

I was really amazed by the the quality of the people. Um, because you you hear a lot of people doing uh data scientists, data engineers, all kind of technological experts, and um the people I met here, um young people but very um experts in their field and um and very motivated, want to have an impact also to bring value, and that I think it's quite uh quite important for me.

SPEAKER_00:

And so you joined Data Roots and you already mentioned the part about value. What does that look like in practice? What's what's your mission here, let's say?

SPEAKER_01:

Well, my mission is really because you know, today more than ever, all companies they have uh challenges. Um challenges usually are not technological challenges, not at first. It's okay, business challenges. How are we going to increase our sales, to decrease our costs, to do a particular thing or that? So they have challenges, and for a lot of those challenges, technological technology can be is a part of the answer. And what I like to do is starting from this challenge, you know, this pain, this ooh, ooch, that's that's pain, uh identifying with the customers what are the pains and finding the solution, the solution, sorry, including technology, but the right technology at the right time for the good people, bringing tangible business value.

SPEAKER_00:

So you want to be part of that early phase where we identify the right use cases, make sure there is business IT alignment, and that the companies that we help are focusing on the right uh business goals and right translation towards uh technological solutions.

SPEAKER_01:

Exactly, because and okay, we heard a lot of things regarding the fact does AI uh proof of concept fail or not? The reality is a lot of artificial intelligence, proof of concept fail. The main reason why, it's because at the start there was not clear business purpose. And the second reason, the people were not engaged. And that's always the case with technology. You know, a couple of years ago, uh when you were uh implementing a new ERP or a new CRM in a company, a lot of people think the tool will solve all our problems. It's never the case. The problems were present will stay there if you're not challenging, if you're not challenge them. So it's the same with AI. So starting with a very clear business purpose, involving the right people, integrating it in the right processes from the start, identifying okay, what platform, what AI model, and directly also identifying what are the metrics that we will follow in order to evaluate is this a success, yes or no. This is what I um I like to do, and with the people at Data Roots, I think we can do that very good.

SPEAKER_00:

I like how you make it measurable and tangible. Um Data Roots is always mentioning that they want to be end-to-end. I heard you talking about the five Ps already. Can you elaborate on those five Ps?

SPEAKER_01:

Well, end-to-end, a lot of organizations speak about end-to-end. What is it concretely? And my way to put it very clearly is the five P. Um and the first P, according to what we said, won't be a surprise for you. But the first one is business purpose. What is the objective of um of this use case? Actually, where does it spain? So, what is the business goal? So the first is the purpose, the second, the people. Who is involved, who should be involved, who shouldn't be involved. The third one is the process. What is the process today? How does this process need to involve to evolve? Uh sorry. The fourth one, the platform. It's only there that we start to speak about what tool, what technology, what artificial intelligence model, how to how to train this artificial intelligence model. And the fifth P is the performance. From the start, what are the metrics, the key's we will follow in order to assess ongoing the project? Uh is this use case successful? Yes or no?

SPEAKER_00:

I very much like these dimensions. To make it a bit more tangible, maybe very concretely, um there's a potential customer or a customer that is seeking support from us. Where do we start?

SPEAKER_01:

Okay. We start with the use case identification. What is use case identification that is trying to identify with the customers what are the business goals? What are the issues? And for that we use different uh techniques or different tactics. And um one of my favorite, because it's working very good, it's the user stories. So it's reflecting to the people in front of you. If you can say, as a CEO, I need to have monthly report on everything in order to inform the board. If you can say, as I a role, I need to a task, in order to a result, you have a use case.

SPEAKER_00:

It's a good structure and you can like challenge a bit too, I assume, because you can have a challenge on the scope already, uh, but also on end-user adoption, how will you use it? So already from the beginning, you ask the right questions and you trigger the right reflexes, let's say.

SPEAKER_01:

Indeed. And you know, when you when we do this exercise with uh our customers, um very often um use cases that come up are not at all artificial intelligence. And this is not a problem because artificial intelligence is not the solution for everything. Um, but at least it comes up. So and we take it, we map it, and um uh so for the customers it's also something that they need to take into account. But when we have this use case list, of course, we need to go much further in the description and all the aspects of these use cases.

SPEAKER_00:

And then you assess value and feasibility, that's the typical approach, I assume.

SPEAKER_01:

Yes, yes, not only. So it's um the objective of the use case description is also to know okay, but what data do you know? Do we need? Where is the data? Is the data available? Is the data qualitative? But not only, it's also what is the target audience for this use case? Because you can have um only an internal target audience, uh, external target audience, very limited one, very large one. Why is this important? Because the objective of the description of this use case is to get info, of course, but also to prioritize, because let's say we come up with 50 use cases. We of course won't implement 50 use cases at the same time. We will prioritize them. And in order to prioritize the use cases, we do that based on two types of criteria. The first one is data, as I said, where is the data? Is the data available? And the second one is business priority. And we do that, of course, with the customers. And then we set up the roadmap, and then of course we said, okay, now uh we do the detail of the roadmap, we we start with the first use case, etc.

SPEAKER_00:

Maybe to add to that, a client of mine, as a rule of thumb, they also focus on uh having the end-user integration. So to focus on the end-user adoption at the end, in the beginning, so when the use case is being defined, they already require the business team to put it on the roadmap. So, for example, if we're building an AI that should be used in the future, then do we already foresee that on the roadmap in the coming craters? Yes or no? If no, then we will not build anything AI or data related because beforehand we already know that it will not get adopted. And it seems very straightforward, but in practice we often see that something gets built, it was not put on the roadmap, and then it might be one year or even longer before it actually starts getting used.

SPEAKER_01:

Definitely, I guess based on what you told me, it's the people that didn't identify all the people involved. And according to the the size of the companies, there could be a lot of people involved, but it has to be mapped indeed. Otherwise, and when you start an artificial intelligence project or whatever project, by the way, but um you start such a project and it stays, it lasts months, years sometimes, costs a lot of money, investment of people, and then it fails. Well, it's really um it's really dematchable. So this is very important to start with the clear purpose, identify the people, the process, what what we need. This is really the key for success.

SPEAKER_00:

And then next to that, as you mentioned, you have the data and business priority axis or um dimensions. How do you assess the potential value of a use case?

SPEAKER_01:

Of course, that depends on the use cases, but um we see that um time saving for people is one of the main uh metrics, and of course, we detail this metric, uh, how how how many hours or many days for this typical use case, but time saving is a clear metric, and the second one could be FTE. So uh, and this is sure today we won't need for everything as many FTE as we did before. It doesn't mean that we have to fire all those persons, at least this is definitely not my way of doing, but but of course, it means upskilling those people, making them more valuable things. So the value chain in every company is changing enormously. And the the value that was generated before and now is completely different as the one that is ongoing by generating. And this is also what we need to evaluate and make the people having these new valuable jobs. This is the role of the of the C level of at uh of the board at higher at higher level in the company. And of course, um, next to time savings, FTE, there are a lot of others possible metrics, is increase of sales, um more visibility, supply chain-wise, you have a lot of metrics. This is definitely uh a part of the evaluation from the very start of uh before implementing the use case.

SPEAKER_00:

Okay, okay, a lot of theory. Do we also have clear examples of how we at Data Roots have applied such an end-to-end approach to a customer?

SPEAKER_01:

Yes, of course, because uh we did more than 50 uh very concrete uh AI implementations um till now. Well, I think of one of our um customers because you know we organized in June this our Gene AI event. So we organized our Gen AI event, and there two uh two of our clients came uh testimony. The first one it was EasyFares, so easy fairs is the event organizers, and there very completely their teams receive a huge amount of um how do you call that feedback. Thank you. They received um hundreds of satisfaction um of satisfaction service very regularly. You can imagine that it's huge to handle, and so we create with them what we called with them easy summary. So that's a solution that gives intelligent summary. So two main um great optimization, time of course, but also quality of the results, because artificial intelligence, generative AI with language model can go much more in the details of the wording of what the people are saying. So we could bring with their teams, uh it's a collaboration, uh save time and higher quality of the results in order again to make uh better decision making afterwards.

SPEAKER_00:

And so instead of having to scroll through all these survey results or feedback inputs, now they get a summary, they can take a look at it, understand what's behind that summary, and make more actionable recommendations for those events.

SPEAKER_01:

Exactly. And not only it's positive, it's negative, it's a five, it's a seven out of ten. No, much more uh qualitative than that.

SPEAKER_00:

Have a deep understanding of the experience of the customers.

SPEAKER_01:

Exactly.

SPEAKER_00:

Cool. And then I think you brought another example.

SPEAKER_01:

Yes. Uh the second one is the national lottery, and they're actually they have in their um they have in uh they have huge uh as you can imagine, regulatory documents. So it's quite long for those people to find the right information at the right moment. And um we worked with them on a solution. I go, I won't go into the detail, but it's a rag system, and it's we we we definitely um with them create a custom build solution. To be honest, in the beginning, they were not completely uh um convinced that AI could be the solution because of the reliability of the potential uh results. But together we build this system, we make all the steps in order to prevent the risks, because definitely with AI you could have risks, so it's important technically to limit the risks, but also to train the people, to bring to the people with these critical aspects, and um and we succeed with them to have this uh this uh system that they use. So it's um a nice collaboration.

SPEAKER_00:

If you look across our portfolio of clients today, what kind of teams are emerging in terms of client needs?

SPEAKER_01:

Well, definitely operational efficiency, optimization. Um, this is clear. It's sometimes funny because a lot of um customers come to say we want to innovate, we want AI help us innovate. And it's a nice goal, but um not every company has the maturity to uh directly start with innovation. That's also a part of when we prioritize, we generally start with reuse case, Gen AI use cases which uh that have an internal target and optimization. We rarely start with uh you know external innovation, um because risks, etc. in maturity is not yet there. So um, but usually they came with optimization, operational efficiency, and um and sometimes and as soon as use cases come to roll out, then we can go uh into uh innovation, new markets, new products, etc.

SPEAKER_00:

Looking ahead, what use cases are you most excited to explore together with clients?

SPEAKER_01:

Well, I what I like the most in is when we can implement a first use case where the value comes very quickly, you know, when this is something what I would perhaps call a quick win. And this is uh why, because everybody sees that okay, it's it's feasible. We we have the result, and everybody is uh is is is happy. So you definitely have this tangible value very fast. We did it with um, I have one example uh in mind of the company Easy Fairs again, because we did uh some uh AI use cases with them, and it's the one called Easy Score. I think it's a nice example. And um it's when artificial intelligence can assist the sales team and and say to the C and tell the sales team, okay, focus on this 100 prospects because they are 80% more likely to sign than the others. Although it asks work, of course, data in order for the AI tool to the AI model to do that, external data as well. But this is something great because you can really have the sales, the perfect sales assistant who analyze a lot of internal data, historic data, external data, in order to tell the sales team, okay, don't bother those ones, start with this one. Because for this, this and this reason, the probability to get the to get the deal is much higher.

SPEAKER_00:

Cool. Um to wrap up the episode, if people want to follow your work or connect with you because you share tons of content, I'm following you. Thank you. Where's the best place to do that?

SPEAKER_01:

Well, LinkedIn, of course. So I have a LinkedIn uh account, so this is a nice uh place to uh to follow me. I also have a YouTube uh channel, so it's also a possibility. And uh what's it called? Uh generative booster.

SPEAKER_00:

So it's very much focused on Gen AI.

SPEAKER_01:

Yes, it's uh purely, it's only focused on uh on Gen AI. And um I also have a weekly newsletter uh which is called the Game Changer list. Um so this is also but on my LinkedIn, you've got every information of what I do.

SPEAKER_00:

Okay, and I think we can drop the links in the description of the episodes to the newsletter and the YouTube channel too. Uh Gail, thank you so much for joining me today.

SPEAKER_01:

Thank you, Ben. It was a nice conversation. Thank you for that.

SPEAKER_00:

Thank you to share your story, your career history, your lessons learned, and of course the use cases that are coming. I'm very much looking forward to collaborating and seeing the business value in action.

SPEAKER_01:

That's the most important. Thank you, Ben. Bye.

SPEAKER_00:

You have taste in a way that's meaningful to software people. Hello, I'm Bill Gates.

SPEAKER_01:

I would I would recommend uh TypeScript. Yeah, it writes a lot of code for me.

Head Shot

Ben Mellaerts

Host
Head Shot

Murilo Cunha

Host