MediaMarktSaturn’s Bastian Sehnert on Scaling AI, Governance & Real Retail Value | Live from Paris

Hello everyone, this is Omnidog Retail and I am Ben Miller and I'm coming to you live from the Vuzion podcast booth at NRF's Retail Big Show Europe in Paris.

Joining me now is Bastien Sienna.

Bastien is head of Data and AI Initiative Office at mediamarkt Saturn.

Bastian, very warm welcome.

Thank you for having me.

Thank you, thank you for joining.

I guess two things, Bastien, please can you introduce yourself and your role and also look, we've got a global audience.

Some people will be very familiar with MediaMark and Saturn, others less so.

So maybe if you could introduce the company as well.

Of course.

Thank you very much.

So mediamark Saturn is the biggest electronic retailer in Europe.

We are in 11 countries, we have over 50,000 employees and over 1,000 store.

We are an omnichannel electronic retailer and I am as mentioned head of Data and AI Initiative office.

So basically my responsibility is for the retail group, for the holding organization, to take the data in AI strategy and make it out of it a portfolio that basically can create value and scale.

Okay, fantastic.

So it's been no surprise that AI has been the talk of the show here in Paris last few day.

So let's start with the big picture.

Mediamonk Saturn's strategic view on AI.

But really what do you see as the potential of AI for your operations?

In my opinion it's twofolded.

So you have on the one side obviously all the customer faced AI that we also know and see in everyday life.

So if you're looking into customer care, you do have the voice bots, the chatbots that are commonly, commonly known.

But also in terms of recommendation, how do you do advertisements in the future is something where AI is coming more and more to it.

The other side of the coin is basically into our internal process.

How you handle things, how you are moving things around.

It can be supply chain, it can be product data, it can be anything where AI can be helpful either in qualifying the data or in taking over whole processes.

Yeah, that's a really nice way to look at it.

That's similar to how we look at it.

You've got the internal how does it help your organization?

And the external how does it help influence customers and how can you help communicate to customers?

So let's take that principle and let's get practical.

You spoke today in the session, I've got it written over here is called using AI to solve real operational and customer challenges.

So go on Bastien, on the spot, tell me the real operational and Customer challenges that you're solving through AI?

Well, one thing is I can take as an example and master product data is a very good example to it.

So if you're looking at it, it's taken a lot of effort to create it, to maintain it, to bring it up into a quality.

And for every retailer, the product data is the most important asset besides the customer data.

So AI can help in classify it and quantify it and making sure that the quality is good even faster than the human.

But it frees up the human to take accountability and ownership on the process.

So the employee stays responsible in the ownership for creating the standard.

You this is something that we as human can do while repetitive tasks like entering something into an Excel sheet, this is something that we don't need to do anymore.

Okay, okay.

There's two elements to that.

One is around the governance.

If we're applying AI to consumer data, I'm going to come back to governance, kind of that's an area of your responsibility.

Before we do that, I want to talk about scaling, scaling AI.

So one of the challenges that we know and conversations we've been having all week is that taking AI from pilots to operational scaling is really hard.

So how do you think about it and how are you achieving that at Media?

Monk Saturn well, I think the two topics that you addressed are connected basically.

So you need to have a good AI governance, I would call it the road that the use cases are walking into that giving a small example.

So early on we bring in the governance into every AI use case and identify whether it's a low risk or a high risk scenario.

If it's a low risk, then we're creating a fast path to scale the use case.

If it's a high risk, then some scrutiny needs to be put in.

Why are we doing that?

Because for a retailer, fragmentation is always a huge issue.

So if you're starting building up an AI use case and down the road when you build something, then you come to the situation that you have data privacy approvals or other issues you might need to look into, rework things and that will create additional effort and slow down the road to get to go live.

And also what we look into, and this is the second part of your questions, what we are also looking into.

Every AI use case, does it have an owner?

So the business needs to be needs to have the ownership into it.

If I'm coming around and I try to push an AI use case to someone, that will not fly.

So the owner needs to be there.

Why?

Because also the Adaption is very important.

That's the second part of the things.

An AI agent that's been built and never been used is not worth it.

It's just cost, it's not a saving in there.

The third thing what I feel is important is it is it entered into our system.

So is this working with our workflows, with our processes?

Because most of the nice demos that you see are in environments where everything work out smoothly.

Once you need to connect backend system, legacy systems, data assets, then it will getting really interesting whether the AI is still usable or not.

And the fourth pillar, the important pillar for me is do you have a plan to scale?

Because having an AI pilot without a plan to scale is just.

Yeah, I completely agree.

So when you're scaling, you have, you've already mentioned you've got operations across multiple European countries, deal stores in different countries, run off different operating models or do you have the same systems?

Can AI help scale in that environment?

Definitely.

Having the same system in place help for AI use cases to scale in not every area we have that sometimes on purpose because you know, different countries also have a different perspective.

Especially when you're looking into customer face situation.

Then you have a different customer care strategy in one country, then into another country to factor in cultural topics, but the system is still the same.

So all of our countries are using the same customer care technology.

But then how the agent is then be tweaked into or localized.

That is something that we do with the countries on the ground to make sure that we are making sure that the Spanish chatbot is using this certain cultural flavor or the Dutch chatbot is using that cultural flavor.

And then when you talk about use cases and making sure the businesses integrated, where does the start point for an AI use case come from?

Is it from you?

Are you in your office and your team sharing ideas with the business?

Or is it the business bringing you?

Is there any way you could help me with xyz?

How does that work?

Well, it started out with us as an organization to push into all the business units to bring in some ideas.

Having workshop obviously, but more and more and also help to the overall development of AI.

So having accessibility to AI bots all around, people are getting more interested into AI and now we see a certain pull.

I have a great idea, but I don't know how to bring it to life.

This is the most conversation that I have right now.

Fantastic.

So looking out 2027 year ahead, what excites you?

What's most exciting for you?

Most interesting for you?

About what AI can do in the year ahead.

Maybe something you're working on or something that you hope to see happen?

Well definitely what we currently see in the US when it comes to agentic commerce is quite interesting also in China even more and I'm also looking ahead to have that in Europe available soon and also I think when you look at the recent announcement of OpenAI from the latest model you see that orchestration will become the next real way moving forward.

How do you make sure that you have all these AI capability woven into one seamless end to end process?

That is something that will be really interesting in the coming years and seeing the time it takes for those frontier companies to develop new AI models I'm not sure if we have to wait a year for that.

Every six weeks something new comes up.

Yeah they're really interesting.

Bastian thank you so much for joining us.

Thank you to Viewsion and NRS Big Show Europe for sporting their coverage this week and stay tuned for more.

Thank you.