What It Really Takes to Scale AI in Grocery Retail | Digital Grocer S7E3
Get More from AI for Your Grocery Business
AI doesn’t stall because the tech isn’t ready.
It stalls because organizations aren’t prepared.
We see this firsthand in the grocery industry, where too many retailers have invested in AI without seeing meaningful returns.
But that hasn’t been a problem for Southeastern Grocers (SEG).
In this episode of Digital Grocer, SEG CIO Todd Renaud joins Mercatus President & COO Sylvain Perrier to reveal how his company has been able to scale AI by solving two of the biggest—and most overlooked—blockers to its success: disconnected data and unprepared teams.
Watch the full episode to learn how you can overcome these hurdles to make AI work for your grocery business—and subscribe below to get notified whenever a new episode is released.
Prefer to listen to your podcasts? Digital Grocer is also available on Spotify.
Challenge #1: Fix Data First
Clean and connected data is essential to grocery retailers.
AI is only as effective as the data it’s built on. And in grocery—where speed, accuracy, and personalization matter—flawed data creates compounded risk.
That’s why grocers have to rethink how data flows through their organization, how often it updates, and how easily systems can use it. Retailers still relying on traditional processes for centralizing data risk being left behind.
To prepare for AI at scale, SEG rebuilt its infrastructure from the ground up: investing in real-time processing, rebuilding its architecture around cloud-based data storage, and minimizing the lag between data capture and decisioning.
These shifts have made personalization and automation possible, but more importantly, they’ve made AI outputs trustworthy and sustainable.
The Takeaway
If your data isn’t reliable, your AI won’t be either. Don’t heap more tools on your tech stack without building a better foundation first.
Challenge #2: Build a Culture that Can Keep Up
Even with the right data, AI adoption stalls when teams don’t understand the tools or don’t trust them.
To move fast without resistance, grocery retailers need a workforce that’s equipped, empowered, and engaged. That starts with these five mindset shifts:
- Prioritize experimentation over perfection: Launch fast, learn often. Early pilots should be treated as learning tools—not polished final products.
- Frame AI as an enabler, not a threat: Help teams see AI as a partner that enhances their work, rather than a force that replaces them.
- Make training practical and hands-on: Integrate AI into real tasks—like finding SOPs or streamlining workflows—to build confidence through use.
- Keep tools accessible and intuitive: Choose solutions that are easy to use and support everyday decision-making, not just IT-driven initiatives.
- Decentralize innovation with guardrails: Provide governance and data safety, but let departments explore, test, and drive adoption from the ground up.
The Takeaway
A culture that embraces AI is just as important as the tech itself. If your teams trust the tools and feel ownership, adoption (and success at scale) will follow naturally.
What You Can Learn from SEG’s AI Journey
Southeastern Grocers didn’t scale AI by forcing adoption from the top down.
They focused on connected data and a culture where teams felt confident using new tools.
That’s what allowed real value to emerge: faster execution, smarter decision-making, and stronger engagement across the organization.
For grocery retailers looking to move beyond pilots and turn AI potential into AI performance, the path is clear:
- Invest in connected data
- Equip your teams to succeed
- Create space for innovation to take root
That’s how AI becomes a practical driver of greater operational efficiency, more effective personalization, and profitable revenue growth.
Subscribe below for more practical insights that will help you get there.

Sylvain Perrier
Named as a Top 10 Influential in Retail 2020 and a 2019 Grocery Game Changer, Sylvain Perrier is a true digital retail trailblazer. As President and CEO of Mercatus, he is the driving force behind the leading digital commerce platform in grocery retail. As host of The Digital Grocer Podcast, he infuses these conversations with his vast understanding of retail, grocery operations and technology, as well as his quick wit and good humor.
Full Transcript
Perrier
Welcome to Digital Grocer season seven, episode three. This is the second half of our interview with Todd Renaud the CIO at Southeastern Grocers. If you missed the first half, there’s a card on the screen and a link in the description for our audio only listeners. Here’s the second half of our conversation. Enjoy.
Perrier
And, Todd, you’re going through this journey, and I really want to know from your perspective, what are the challenges that you’re seeing from implementing what you’re doing in AI with them within Southeastern Grocers?
Renaud
Yeah. Well, I think anyone in technology will tell you money, right? They haven’t enough money to do what you want to. But that really isn’t the biggest challenge per se. Not that we’re, you know, we could always use more. I think it’s just like any other technology. How do you innovate? What do you spend your time on?
Renaud
To me, and I didn’t invent this thought. I was actually listening to a conference one day and they were talking about data. Data has to be clean and it has to be connected. And when I say when they said it, and when I say it to you and to the listeners, a lot of people say, uh, no duh.
Renaud
Todd yeah. Ohoo, that sounds, you know, hope you didn’t use AI to come up with that. But if you really think about it, start to unpack it. I mean, yeah, you know, it’s true. But with AI, it’s actually times ten true or times 100 true is because it how fast it’s using information, how much people trust the response even though they know it lies or hallucinates or is incomplete because it’s so easy, it happens so quickly, it’s easy to just say, oh yeah, that’s accurate and it
Renaud
exacerbates, deficiencies, gaps, etc. in the data that you have. And you know, garbage in, garbage out actually becomes more true because you don’t have even a human really looking at it in a way to do a sanity check or something that kind of gets at your gut a little bit, because AI is not thinking contrary to any movie or magazine or book that people read.
Renaud
So data is the biggest challenge for us is how do we make sure I have the right data? It’s clean, it’s connected, it’s updated on a regular basis. So that’s number one. Number two is still going to come back to people and you know people and their skill sets. And being able to leverage this new technology. If you have people on your team who are curious they’re going to do well with AI.
Renaud
If you have those that aren’t, they’re not going to. It’s like any other technology. They’re not going to embrace and learn that. But we need to do upskilling because it’s a new set of skills and there’s nothing you can go to college for. Really, at the end of the day, you can read some books, but it changes so quickly.
00:02:49:15 – 00:03:07:11
Renaud
I mean, who would have thought a year ago when everyone said, AI prompt, engineers are getting hired for $250,000 a year. And now if you said, hey, what’s your title? And I said AI prompt engineer, they’d laugh. Why? Because AI is the prompt engineer. You don’t need a human to do that. You know, that’s how quickly it’s moved before it even got settled in.
Renaud
The job’s gone. The pushback still, to me, exists. This cultural shift of the AI empowers. It doesn’t replace. That’s how you ultimately need to review it. And so getting that cultural shift, I think, is there. So to me it comes down to data and, you know, people and their attitudes and the culture associated with it. And so what we’re really trying to do is to make it less scary, to make sure it’s governed.
Renaud
And again, to go back, I use the words, you know, awareness and engagement and education. That’s really, really important. And then just repeat that over and over again. But I think in our company, it will best serve us if we let it grow organically. I can’t own it all. I don’t want to be the department. That’s the Oracle on the hill and they have to get approval.
Renaud
No, I’m going to give them the tools and let them go. I just got to make sure the data is safe, which it is. And we do prevent a few things from being accessed and off our network because we just still have to do that. People sometimes make mistakes and we want to prevent that from happening.
Perrier
I can imagine in the day it’s like when people shifted from the typewriter to WordPerfect. Yeah, it’s like, here’s a better tool, here’s a better option. Try it. And eventually you got you get enough adoption, enough scale, and then you’re like, okay, we’re done. This is how we’re going to do things moving forward. And I’m probably dating myself using the word typewriter in any case.
Perrier
And you’re right about you’re right about data. Get data in flight and data at rest. And when we did our snowflake implementation, I think this is like going on 4 or 5 years now and, and our real time processing for our personalization engine, we had to reimagine how we’re curating our data and reviewing it. Gone were the days gone or I wouldn’t say they’re completely gone.
Perrier
You know, minimizing the whole ETL process for us was a bit of a mind shift where you could take a snapshot easily, QA and fix it. That’s completely change. So when you look into your crystal ball.
Renaud
Sylvain just for a sec, let me give you a different, another perspective, because I’ve been talking a lot more about an IT team and the corporate office. But let’s go out to the stores and and talk about your question there. And I think this example is going to be it was incredibly interesting to me. So we created a store assistant and simply we took some standard operating procedures and we enabled it through.
Renaud
It’s not even really generational. It’s not creating new content. It’s basically for certain departments. How do I do X? And before they might have a 300 page manual, we took it, they turned it, trained the AI. And you could say how do I do X? And it will come back with the SOP. It’ll show it. Initially it was still the PDF.
Renaud
So you know, it’s not really made for a tablet or a handheld like we want to eventually. But it took it and digitized it, put it there and it showed reference. So it said if I want to be able to do X, it would say, here’s how you do it, and it would give you the link to the actual SOP, and you could pull that up.
Renaud
And so we marketed, educated, piloted and rolled it out as a store assistant for very specific departments. And, you know, what was happening is, a majority of our questions coming back, because we logged any question that didn’t have an answer, were not related to the store operating procedures. They were related to generic company questions. Hey, you know, can I got a question on my benefits?
Renaud
How do I do this in a different department? Asking other questions. And everybody, not everybody. We had those that were viewed that as I don’t call it a failure, but like oh my goodness. You know, we’re going to have to figure out how to do this. And I got excited because what I said was, wait a second. Within literally a week or two of people rolling out, they were comfortable with the technology and they’re asking it to do more.
Renaud
And guess what? We can do more. I can give those answers. We just need to start thinking about it differently to be able to do that. So I think about changing the people and the culture. I also think that people are ready for this. They’re using it today. And you referred earlier to the and I can’t remember the term we used about, you know, making it easy available to everyone.
Renaud
I call it democratizing brute computing power to for 20 bucks a month and now free, you can get ChatGPT and the world was your oyster from processing power for that. So I think the biggest challenge also might be we’re going to be behind our users, we’re going to be behind our customers and what they want to do. So the companies that can figure this out ahead of time or be like, I always like to say capability had a need.
Renaud
Those are the ones that are going to be leveraging this technology for success. And so we’re we’re trying to shift that thinking here on our team here to say, hey, think differently about this. So anyway.
Perrier
Bigger concept of decentralizing. You have your loyal team but decentralizing. And a lot of people try it and letting it go organically, which I think is code for they’re going to do what they’re going to do. It’s the right is the right approach because at their fingertips, they have it in Instagram, they have it in Facebook, and they have it in Gemini.
Perrier
That’s now implemented by Google. So they’re testing it now. Today I think it’s a matter of time before somebody just kind of says, okay, well in my day to day I should be doing this Air Canada or Delta. I can’t remember which airline, but they’ve decided to replace their entire IVR tree with, as opposed to training a voice recognition on the proper enunciation for all the various ethnic groups that are calling in.
Perrier
They just boldly turned on AI, and it is the I think it’s it’s 40% cost savings out of the gate is being realized just by doing that, which is pretty phenomenal. Example, if anyone.
Renaud
We’re dabbling in that area as well. Nothing like what you just described. You know, as a technologist, I’d be very interested in understanding how long it took them to actually turn that on. And how did they do it, you know, how long did they have to train? Did they do a lot of listening? Did they use AI to do that knowledge, build and translation and transfer to kind of get to that point?
Renaud
And I think that’s a good example of when they say, look, AI is going to take some jobs. Yeah. I mean, you just described that long term. I mean that’s a reality. Call center reps are probably not going to be a job in five years or much less that that’s going to be an area that you’re going to have that, but you still need someone to train, someone to teach the response, someone to be empathetic, someone to be less than empathetic.
Renaud
Depending. There’s all of that. So you could actually take a lot of those individuals and say, well, you’re now going to help train the model, you know? And so I think it’s a job shift, but that’s incredibly interesting examples of of AI usage that has real business value.
Perrier
So when you look into your crystal ball and you think 3 to 5 years out, what excites you about AI in grocery?
Renaud
So I came from outside grocery, I was in retail, did finance consulting, energy product. So I like I’ve always been a technologist. So I like to say that I have a lot a few points to bring and any place that I’m, I’m working, but, you know, grocery is an area that excited me because the opportunity for innovation, although I would also say outside looking in again, I’ve only been here for years in grocery and, and many more people in this industry have been here for a long time.
Renaud
The more things change, the more they stay the same. When it comes to grocery, we still have checkout stands, we still have baskets. You still got to go down the aisles to get food. What’s interesting is my kids, who are in their early 20s, they love grocery shopping and not because I work at the grocery store. I mean, they have a relationship with food.
Renaud
They have a relationship with what they’re putting in their basket. They actually like the process of it. Not everyone’s like that at any age, but I find that very interesting. So there’s something deep and primal with shopping, I think, for food with humans. And I don’t know that that’s going to change in our lifetime. But when you think about how we can make the process more enjoyable and how we can make sure out of the 30,000 products that are in a store, that they’re getting the best ones for what they want and when they want it, I think the customer side of AI is important.
Renaud
So I think of things more with online. For example, AI agents. And look, that’s a simple term now like how does that grow long term? I think about being able to go into a store as a customer and, and really think about the vision piece. How can vision work for personalization? How can it work for just finding the products, taking advantage of all of that?
Renaud
I think that’s really exciting for grocery, you know, meta glasses. You think about the movie with Tom Cruise, and I’m having a complete brain fade now as we’re talking here, the Minority Report. Whereas he’s walking in the through the aisle and they do his eyeballs and say, hey, you bought pants from us in a while or something of that nature, right?
Renaud
I mean, being able to see what’s in your basket. Hey, you’ve got something. If you add rice, you’ll be able to do X, Y, and Z. So anything to do with vision and some kind of feedback loop through an app, through your meta glasses, through earbuds, whatever you have in the future. For that, I know I’m jumping around if I go back to the agents, is the automation preparing your grocery shop, preparing the recipes, the meals, and probably some areas that I can’t even think about.
Renaud
But it’s based on your history. And if you’re willing to give all that information or access to it, it can really make your life better. No different than social media did in its infancy. For for some of that, I think the vision and voice also from just general corporate standpoint, how am I represented from a digital self?
Renaud
I mean, that’s not something a lot of people are comfortable with now because of the security piece. You know, I think is there there is a huge efficiency play. And that’s not exciting per se, especially not for the customer. But I think there’s a lot of opportunity in there to really make these processes so much better. And then I’d also say, how can we make it such that the people and the associates that are in our stores are able to spend time focusing on the customer and not on getting a product onto a shelf at the right time, or knowing when fruits going bad or when expiration dates are happening.
Renaud
And so anything associated with that and instead letting it be more of an interactive experience between our customers and our associates to me is pretty exciting. You know, to do any of this, though, retailers are going to have to invest in data data Foundation and everything associated with it. I could say data under, you know, blank line, and there could just be 10 or 15 other terms in there.
Renaud
And today we talk about AI literacy. I think that’s the infant way of looking at it. It’s really thinking beyond what is what makes AI interesting and what skill sets do you need. And that’s interoperability. Thinking about leveraging pieces, it’s almost like everyone needs to learn how to code, right? It’s the code mentality of so they’re not going to code, but they have to learn to put pieces together into to to create agents.
Renaud
And so how are we going to, you know, deliver that type of skill set so people can do that into the future or you’re just going to create a faster way for failure to happen as well. So I think it’s, on this one, I would tell you, it’s hard for me to imagine where we’d be in a couple of years.
Renaud
If it keeps growing at this, my question would be, and maybe you could come up with a blank thought for you, you know, are we going to see a plateau and even a drop? Right? I mean, the cost of this processing, the number of these offerings and tools and AI becoming so ubiquitous and nomenclature as well as being embedded, do we see to where it really doesn’t change as much from now into the next five years, because it now becomes something everybody has, and therefore you’re back to where you are today.
3
Perrier
I so I think this cycle innovation is peaked. Yeah. Interesting. Right. And it’s the last few weeks and the trades are inundated with news on everything else. Right. And I say it’s peaked because we had a much smaller language, model come out of China. That’s the less costly to operate. But the efficiency is as is good enough.
Perrier
And, and I think that stop the hype cycle for OpenAI and Nvidia and other chipmakers. And I think Microsoft’s what they’re doing with Copilot. It’s a bit lost is my sense. I don’t think they know. And I think a lot of that is driven because this technology straddles the consumer market, straddles the enterprise space, and no one, no one has found anything yet concrete for it, not concrete for it to do.
Perrier
But what is a true ROI? It reminds me in the day when we had the 98 bubble, the internet bubble, and everyone said, oh my, the Nortels, the Quests, the AT&Ts of the world overestimated their the bandwidth requirement by 16X too high. And so I think there needs to be a bit of a retreat. And our business leaders and you being one of them will stand back and say, okay guys, great, thanks for the tech.
Perrier
Let me figure out what to do with it. Now, once we figure out what we can do with it, then that’s going to be the marching order where this tech will go for retail. At that point, that’s my view. But I don’t have triplets sitting in a pool of water behind me. The precogs here that are kind of determining where we go, right?
Perrier
Yeah. So, Todd, if I was a retailer and I was, you know, grocery retailer, and I’m flirting in this space, I don’t know what to do. And I’m for early, early stages. What would you advise me?
Renaud
Okay, a couple things come to mind because, you know, again, we’ve been doing this for a bit. I would say under-promise and overdeliver only because you’re going to go up against some of the more difficult challenges, such as people, culture, some pushback, the governance required security, and to do those appropriately takes time. I think that’s, you know, number one for that.
Renaud
Number two, don’t underestimate change management and education required and awareness. It’s really making sure people understand what’s in it for them. Why should they do this and not to be scared of it but to leverage it? I already referenced early on tackle the governance and security head on. Do it first and you know, look, your company culture is going to dictate some of this.
Renaud
Do you have to even do more or less. But go ahead and get that out of the way, get that understood. And that goes back with the education and the awareness. And if you can get those pieces out of the way, you’re going to end up being able to to go faster when you get there. I would also move very, very quickly.
Renaud
I would suggest you move very quickly and be comfortable with breakage. And again, someone might say, listen to this. No data that you know or you know, that’s what Facebook and Amazon always said go fast, break things. So forth and so on. But the reality is the technology is moving so fast. The solutions are moving so fast. If you take 6 to 9 months to do something, by the time you get done, you’re two versions behind and it’s so how can you architect your process to to move very quickly and, and then do it in an agile and a fast feedback loop?
Renaud
The final thing I’d say is you need people who are going to be curious to to push this forward. You need those who are going to be passionate because that’s what it’s going to take. There’s no roadmap today. Maybe in 2 to 3 years there will be, but there’s no roadmap to this. There’s no user manual for the technology.
Renaud
It’s not like a Microsoft office or an Excel where yeah, it maturates over time. But at the end of the day, you learn it, you get certified, you move to the net, it’s moving too quickly and there’s no roadmap. So how are you going to navigate that? You’ve got to have people who are curious, ask questions, move quickly and do so in a way that doesn’t put the company at risk.
Renaud
That’d be my advice for today. Ask again. Ask me again. In a week or two. I’ll be that much further along. And that might change. But it’s it’s it’s it’s an exciting time because it is moving so quickly. And you’ll get that immediate feedback.
Perrier
Absolutely. Sir. I want to say thank you for joining us on Digital Grocer. Well you’re welcome. And for those of you want to get ahold of Todd, you can find him on LinkedIn, his profile, and I’m sure you just DM any of your questions.
Renaud
Right. Well, thanks for having me. I really appreciate the opportunity to to be on the show.
Perrier
Thank you sir.
Perrier
Don’t forget what you have to say. Oh, yes. Subscribe. And like. That’s right. So you can get, all the episodes, even our back catalog pushed into your into your feed, so. Oh, the classics. I feel like this is a 70s revival band, and we’re we’re coming. We’re talking to a casino near you, and, we’re going to be playing the hits.
Perrier
There you go.