How Mirakl Is Powering the Future of AI Commerce with Chief Data & AI Officer Anne-Claire Baschet
Omni Talk sits down with Mirakl’s Chief Data & AI Officer, Anne-Claire Baschet, live from the Mirakl Summit in NYC to talk all things artificial intelligence and e-commerce. Anne-Claire explains how Mirakl is embedding AI across its platform—from catalog onboarding to media operations—and why the future lies in “agentic commerce,” where AI-powered agents help shoppers and sellers alike. With 91% of Mirakl employees already using AI daily, this is a behind-the-scenes look at how Mirakl is preparing for the next generation of e-commerce infrastructure.
#MiraklAI #agenticai #aiinretail #omnitalkretail #MiraklSummit2025 #ecommerceinnovation #retailai #productdiscovery
This podcast uses the following third-party services for analysis:
Podcorn - https://podcorn.com/privacy
Transcript
Hello, welcome back.
Speaker A:This is amitalk Retail.
Speaker A:I'm Anne Mazinga.
Speaker B:And I'm Chris Walton.
Speaker A:And we are here at the Miracle Summit in New York City.
Speaker A:It's hopping here, everybody's excited.
Speaker A:We're almost halfway through the day today.
Speaker A:And Chris, we have a really exciting guest here.
Speaker A:We get to start by welcoming Anne Claire Bechet, the Chief Data and AI Officer at Miracle.
Speaker A:Welcome, Anne Claire.
Speaker C:Hello.
Speaker C:Hello everyone.
Speaker B:Hello.
Speaker B:Hello.
Speaker B:So why don't we start by this.
Speaker B:This is a question I've always wanted to ask somebody.
Speaker B:What is it that a Chief Data and AI Officer actually does?
Speaker C:That's a question my kids asked to me and I don't know if they got the answer.
Speaker C:So I think my role is in two parties to really to help our customer to accelerate their growth with AI by identifying where AI can create gaps of value for them and try to embed that smoothly in our platform, both for sellers and operators in 1P and in 3P.
Speaker C:And I would say it's also a role for the internal miracle workers is really to help them to masterize AI, build their own agents.
Speaker C:And we shared in the keyNote that now 91% of miracle workers are leveraging AI on an everyday basis and 30% of them have already built their own agents.
Speaker C:So that's really both my role, it's really more I would say to empower people to leverage AI, not for the shiny part of it, sorry, but for the sake of it and create value.
Speaker A:Well, and how, how are you at Miracle?
Speaker A:Kind of helping the miracle workers kind of unlock new abilities as AI is constantly changing.
Speaker A:What are some examples of kind of how you're, you're staying innovative when it comes to a being part of the E commerce shopping experience.
Speaker C:So I think the first thing is to help them not to have the FOMO effect, to say it's cool, it's changing super fast.
Speaker C:And also try to demystify and share what really matters, but really concretely, I think my responsibility is to help them to solve the problem they have on an everyday basis with AI.
Speaker C:So first is demystify, give them the right tools to help them to accelerate, make people from the business build agent, use agency AI as easy as it should be.
Speaker C:And what we do concretely is in every keynote of the company, even in the XCOM and with all the exec of the company, we always share simple examples of people that have succeeded to crack a problem with AI and make a demo of this agent.
Speaker C:And sometimes it's really funny someone says before I had to take a coffee before doing this task because it was so horrible.
Speaker C:Now look, I asked my agent and he's able to solve it and I haven't got the time to start to drink my coffee.
Speaker C:So that's part of the success.
Speaker C:It's make people empower them because they are the one that have the problems.
Speaker C:They are the one that know they have the expertise and can reinvent their workflow.
Speaker B:I could just see all you in the home office in Paris sitting with your coffees comparing AI notes with each other.
Speaker B:That's great.
Speaker B:It's great.
Speaker B:I get, I have that visual picture in my head, believe it or not.
Speaker B:Maybe, maybe some French pastry alongside of it too.
Speaker B:All right, so you've done a lot of development on the platform in regards to AI.
Speaker B:So both recently and some, I imagine that's upcoming.
Speaker B:Which of you know what of that in that arena are you most excited about?
Speaker C:I would say what I'm the most excited about is all the change we have on the search markets and what is going to be really the agent e commerce experience.
Speaker C:And we strongly believe that miracle that we can build all provide all the infrastructure with this new area.
Speaker B:Wow.
Speaker C:Meaning that when you look at our product what we are doing is clearly accelerating and improving the quality on catalog management, catalog onboarding.
Speaker C:It's also powering with AI all the commerce and ads operation and really this is the value we can bring wherever you want to sell, wherever shoppers decide to sell.
Speaker C:So I would say overall it's that I would say the one I'm most excited about is all the potentiality we have with media and with multimodal because last year we announced Catalog Transformer was now we say more our agentic system where you can delegate to Catalog Transformer all the painful tasks of catalog onboarding.
Speaker C:Please categorize me this product map my attributes transform to the expected value list of this marketplace.
Speaker C:Then after maybe I have some information scrappy in my product description but you can structure that.
Speaker C:We have achieved that all over the past year.
Speaker C:We are not just building the Agentix system.
Speaker C:We have developed more than 10 specific E commerce models.
Speaker C:We have fine tuned to be better at solving specific commerce tasks than you can have with GPT4 or you can have just with Google's models really train them to be very good at commerce.
Speaker C:And now we are going to integrate in H2 multimodal capability in two ways.
Speaker C:The first one is consider image as the first source of product data nutrients and mix that with the text data so now you can have better product content.
Speaker C:And the second Thing is also enhance all the images because in this agent e commerce era, it's going to be also agents browsing your website looking at the images.
Speaker C:So if you are looking for a dark blue dress and it's only written blue in your description, they will use the images.
Speaker C:So they better be with good quality.
Speaker B:Right, Right.
Speaker B:Wow.
Speaker A:So it's like a double check or like multiple point check to help people find the products that they're looking for.
Speaker C:And I think the challenge we are facing is going even further in the quality of the product description and considering images as really an important part of it.
Speaker C:Important part to have more information about the occasion where you can use the product.
Speaker C:Because definitely on this agentic platform, 10 to 11 words to express your needs.
Speaker A:Right?
Speaker A:Exactly.
Speaker C:And you have more context so you can understand that you need technical shoes for a marathon or a 10K.
Speaker C:And if you don't have this information, that's more complicated that your product is going to show up.
Speaker A:Right.
Speaker A:Well, Claire, I'm wondering if you can give us.
Speaker A:You mentioned you like to tell stories of how AI is helping with these simple stories.
Speaker A:Do you have any customer success stories that you could share with us quickly that demonstrate how miracle AI is really helping do the things that you just talked about?
Speaker C:I have one example that was an example of last week.
Speaker B:Okay.
Speaker B:So.
Speaker C:Office sellers that was trying to face a difficulty because it was with an operator that changed element in his taxonomy.
Speaker C:It's decathlon that is well known in Europe a bit I think in the US and it was complicated because they had to reimport all their catalog.
Speaker C:But with catalog transformer they were able to reinborn in just one night in four countries for the marketplace and to start to sell again.
Speaker C:So we think about catalog onboarding and it was able to do that in one night for them.
Speaker C:They wake up, they say this is the best tech ever.
Speaker C:And I was thinking, okay, that's good because I didn't even thought about this use case, about solving the problem, changing in the taxonomy.
Speaker C:Sellers couldn't sell anymore.
Speaker C:But for both parts it's super great.
Speaker C:Wow.
Speaker B:All right, well, we're going to get you out on here on this.
Speaker B:We've been asking everyone this question to close us up.
Speaker B:You know this.
Speaker B:The theme of the summit is accelerate.
Speaker B:So Ann and I are curious of all the AI trends out there.
Speaker B:I think I have a feeling of how you might answer this based on what you've already said.
Speaker B:But of all the AI trends out there, which one do you think is going to accelerate E commerce?
Speaker B:The most.
Speaker C:I think definitely agentic.
Speaker C:And I would say all the fact that, you know, the everyday tool we are going to have is our individual agent we are going to build ourselves maybe or an agentic platform like Perplexity, Entropy, grok, whatever, Google, OpenAI.
Speaker C:And I think this will deeply change the way we will express our need of products and services and we are going to also reach to this kind of product.
Speaker C:So in a way it's a threat because you can say I built for over a decade century brand awareness.
Speaker C:So the risk is you no longer show up on this kind of platform.
Speaker C:But this is also a huge and massive opportunity to change the game and to be less monopolistic on the search market than we are today.
Speaker B:Right, right, right, right.
Speaker C:And I think this is, this is super exciting.
Speaker B:Yeah, wow.
Speaker B:Got my head spinning.
Speaker B:I got some questions for you after we're done too.
Speaker B:I want to get your thoughts and press your brain on that.
Speaker C:Yeah.
Speaker C:When you look at some metrics, if we want to share that with everyone.
Speaker C: look@chatgpt.com just in May: Speaker C: s in their product in October: Speaker C:In the US it's the same market share as being.
Speaker A:That's just one shit, one provider.
Speaker A:Yeah.
Speaker C:And when they launched the Discover product feature, I think it was one month, a little bit than one month ago, that allows you to directly explore products in their interface.
Speaker C:They shared that it was 1 billion of products and services search per week on their platform.
Speaker C:So new opportunities clearly, but the game is changing and accelerating.
Speaker B:Right, good, good, good stab there.
Speaker B:I like that at the end.
Speaker B:Good drop.
Speaker A:Well, thank you so much and Claire, we really appreciate you taking the time.
Speaker A:Thank you again to Miracle for helping us bring you all of the interviews here from the Miracle Summit today.
Speaker B:And until next time, be careful out there.
