Customerland

Advanced Data Management for Modern Retail Supply Chains

• Santiago Garcia Poveda • Season 2 • Episode 24

Share episode

Ever wondered how AI can completely transform the way retail and other industries operate? Join us for an eye-opening episode featuring Santiago Garcia Poveda from O9 Solutions as he reveals how their cutting-edge platform is shattering traditional, siloed models and bringing unprecedented integration and collaboration to the table. Santiago dives into the nitty-gritty of managing massive datasets across functions like merchandising, demand management, and manufacturing, and shares the secrets to overcoming cultural and technological roadblocks. Discover how O9’s patented GraphCube database provides unmatched scalability and flexibility for data aggregation and relationship building, and learn why aligning people and processes is crucial for a successful digital transformation.

In another fascinating segment, Santiago discusses the revolutionary "supply chain digital twin" technology powered by O9's digital brain platform. Imagine having a precise digital replica of your entire value chain, complete with distribution lanes, lead times, capacities, costs, and even CO2 emissions—all optimized for better decision-making. We also delve into a compelling case study featuring a major North American retailer, illustrating the collaborative efforts required to implement such advanced technologies. And for a lighter touch, Santiago shares his personal experience of uncovering hidden gems in his hometown using ChatGPT, showcasing AI's surprising potential to enrich our everyday lives. Don't miss out on this captivating conversation that's sure to broaden your understanding of AI’s transformative power.

Santiago Garcia Poveda:

You put in those rules and then we can run a solver that actually will peg through the entire chain, that demand that we have at the ending point on the different days, and provide visibility on the supportability of that demand and the optimal distribution, manufacturing and sourcing plans to actually support and service that demand that demand Santiago Garcia Poveda. Maria with O9 Solutions. Thank you for joining me. I really appreciate this. Thank you, Mike, for having me join you. It's a pleasure to be here with you today.

Mike Giambattista:

Thank you. So Santiago and I first met at NRF earlier this year. I got an introduction to 09's digital brain platform, which has been helping retailers of all sorts with their I'd say overall strategies and implementing them. Discuss something I barely know anything about, let's have Santiago talk about. If you wouldn't mind explain a little bit more about Onine, specifically the problems you're solving and also your role there.

Santiago Garcia Poveda:

Totally, totally so. Basically, onine Solutions is a leading AI-powered software-as-a-service platform to support end-to-end planning for companies across industries. So we touch definitely retail and also consumer products and also more industrial and supply manufacturing, and we cover all the space from merchandising, demand management, commercial management, distribution planning, manufacturing and sourcing, and my role at R9 is to lead our global retail and fashion sales team globally and it's a pleasure to kind of discuss a little bit the challenges that we're seeing in some of our clients and how they are actually addressing those today Challenges that we're seeing in some of our clients and how they are actually addressing those today.

Mike Giambattista:

Thank you for that beautiful introduction because really, where I'd love to focus this conversation are those challenges you provide, to see what the challenges are that retailers are facing and how they're going about making the decisions that will go into solving them. So let's just start with some of the challenges and the complexities that you see retailers facing right now.

Santiago Garcia Poveda:

Yeah, and I think the main challenge and it's not exclusive of retailers is that we have an extremely data-rich world today and we have tons of data all over us, but unfortunately, I think we haven't yet found necessarily the best way to extract the most value out of it.

Santiago Garcia Poveda:

Out of it, and what we in ONI have felt that is the main reason for that is that we, despite having data rich environments, we're still operating in traditional operating models very siloed, disconnected and not collaborative enough. So our motto being, and our objective, is to help companies transform their operating model from traditional to digital operating models where you can actually connect different functions over a single data model, end-to-end, and really making decisions for the better outcome of the company and not for the individual KPIs of your function. That, I think, is the most ingrained challenge that I'm seeing when people are now approaching digital transformation, and that's leading to a traditional best-of-breed logic, to looking more for a best-of-suite approach that actually allows to touch on different functional areas and integrate them on what it's called integrated planning integrated business planning, integrated retail planning.

Mike Giambattista:

So, to be fair, there are a lot of technology providers out there that will say that their solutions help to solve those problems, because every single retailer beyond retail would agree with you that the biggest challenge right now is moving from siloed data sets and the barriers that presents to genuine insights. And I have I've taken a look at the solution set that that 09 offers and it's vast. What you guys have is is touches on potentially every aspect of the decision-making chain across all of retail. It's vast and super impressive. But so much of those kinds of transformations go beyond the technology because you're talking about corporate cultures that are built to protect those silos often. So I guess I'm curious with a solution such as O9, which is so vast, so comprehensive and is designed to break down those silos, how do you approach the cultural hurdles to implement something as extensive as what O9 is bringing to the table?

Santiago Garcia Poveda:

Yeah, and I think it's a great question, mike. So the first thing is we are cognizant that we rarely are going to find an enterprise-grade company that is completely greenfield and it's looking for all the systems at the same time and there's no legacy to actually integrate with. So the way that we architected the platform is notably thinking about the end-to-end connectivity, but designed as Lego blocks that integrate between them seamlessly and you don't necessarily need all of them to be able to build the picture that operates for your company. So that's just one way to actually overcome the first hurdle, that is, not everyone might actually be on board in the transformation. The first hurdle, that is, not everyone might actually be on board in the transformation. The second one is there's a big technological hurdle on how to handle data, because every single function has their own data sources, their own data granularities, their own data hierarchies, and being able to harmonize all that and centralize it on a single data model is one of the key challenges where many technologies fail in this daunting task.

Santiago Garcia Poveda:

We have a patented proprietary GraphCube database that actually bridges the best of the all AppCube databases that provide the possibility of actually aggregating up now in the hierarchies, and the graph technology that actually allows to build relationships between different objects and through that we have created a database that is actually extremely scalable and extremely flexible to actually be able to load those different data sources and be able to create relationships between them and create a single optimized in-memory database that actually spans different functions and allows to really visualize the data and operate on the data at the granularity that is needed for each individual user.

Santiago Garcia Poveda:

So there is a technical element of it, and then I would say that the biggest hurdle of all and you were touching on it is the people aspect of it.

Santiago Garcia Poveda:

So it's people need to have a common view of change and an aligned and coordinated effort on both the selection process as well as then operationally, how to use the tool, and that actually, for us, is more as the part of the selection. So we are not aiming to work with a million companies in the world. We want to be strategic partners of our partners and customers and therefore what we're looking at is for people that have a vision to actually go into a journey of transformation and they are really at the ambition level and engagement of the top management to actually go after something that is transformative. And that's the people that we will work with and it's not always from the beginning on an end-to-end transformation. We often maybe start with one use cases, but we are actually have an understanding that the vision that the company is operating on it's going in that direction of a digital effective model and leveraging data and technology in a different way for making better decisions.

Mike Giambattista:

Picture your organization as a solitary island amidst a vast churning sea of opportunities. On the horizon lie other islands, relationships waiting to be built, partnerships waiting to be formed, but between your island and those distant shores stretches a turbulent expanse filled with uncertainty and risk. You're going to need a bridge or bridges plural to span this divide, connecting you to new horizons. Those bridges are strategic partnerships. Yet building bridges and partnerships is no simple task. They both require careful planning, precise engineering and a deep understanding of the terrain. Missteps can lead to collapse, leaving your business stranded on its island.

Mike Giambattista:

Tectonic builds strategic partnerships, helping organizations navigate the treacherous waters, ensuring that those bridges are strong and resilient and scalable. There's a reason why some of the largest, most progressive companies in the world are retooling their revenue models and investing in partnerships. Find out more at tectonicco. That's T-E-C-T-O-N-I-Qco, t-o-n-i-q dot C-O. Would it be fair to say that the way 09 approaches these kinds of new client engagements, that you've got a graph database application that can be plugged into from any source and layered on top of the organization to the point where other functions and use cases can then be plugged into that same database over time and the same functionality? Or am I misreading that?

Santiago Garcia Poveda:

No, I think you're reading that correctly. It's a very flexible architecture where we can one plug in data from other sources and eventually we will be able to also bring new functionalities on our platform, and to that point, it's also a very open platform. The entire analytics layer is an open architecture layer, so that our algorithms are not a black box like many other solutions are. We believe that if there's one thing that I I think we can all be certain of, is that the world in five years will look nothing like the world looks today. So trying to solve for today's problem will definitely not provide a solution from problems five years from now, and we hope that all customers will be looking at future-proving their technology decisions and for us, providing an open analytics layer. It's part of that. We see a growing trend of companies having their own data science teams and growing their own proprietary IP on that space, and we want to enable a platform that can provide that out of the box if customers want, but they can also eventually develop their own IP on the platform.

Mike Giambattista:

You mentioned earlier on that the starting point for working with O9 solutions could be any one of a number of use cases. What would? I don't know if this is even a fair question, but what is the most common starting point? Or would you say, if you're a retailer and you're thinking it's time, it's time to accelerate my digital transformation, I've got to start thinking in a more integrated way. Where would be the first place to apply 09's kinds of solutions?

Santiago Garcia Poveda:

Yeah, it's a good question, Mike. I always say it depends and it's a terrible answer, but it's okay. There is no really one place. Where is the best place to start it? It really is dependent on the company's needs and what's their starting situation, their current technological landscape and their strategic priorities. And, to be honest with many customers, the way that I tend to try to identify that best place to start and it might sound like a joke but it is not is to ask where do you have the most Excel files or the biggest and more complex Excel files?

Santiago Garcia Poveda:

That's the place that you actually have the direst need for our technology to help you manage that process better, and that might be a fair number of places. So we are seeing a lot of traction on supplier collaboration, which might seem like an area that is not inside the realm of your four walls, but we constantly keep sending excels over email to our suppliers, having no visibility, losing time back and forth. That's one space where and it's not probably where we would be starting usually transformations one year from now, but we're seeing a lot of traction there now and if you want a quicker return to value, because you need to fund the rest of your transformation, usually going onto the operational horizon, is the one that provides, I would say, a faster ROI, because you immediately impact on the bottom line. So you have a lot of people looking at forecasting replenishment and supply chain constraints. So it truly depends on what is the challenges that each company is facing at that moment in time.

Mike Giambattista:

You mentioned supply chain. We'd be remiss if we didn't start talking about it, because I know that's a big part of what O9 is solving for. But let's talk about the specific issues that O9 is addressing, because those are clearly hot buttons with a lot of people right now, and how O9 is solving for those.

Santiago Garcia Poveda:

Yeah, so basically, in the end it's online digital brain platform that you were referring to before.

Santiago Garcia Poveda:

It operates on a digital twin of the entire value chain, no-transcript.

Santiago Garcia Poveda:

So you have your primary distribution lanes, any alternative lanes that you have available, the lead times that those different distribution steps or manufacturing steps or processing steps take, the capacities of each one of those nodes and steps in terms of storage, loading, personnel whatever kind of capacity constraints we can load and the costs of every single one of those steps, and now even the CO2 emissions associated to every single one of those steps.

Santiago Garcia Poveda:

When you are able to load all that information on a supply chain digital twin, then you can run solvers that allow you to have a view of demand and having a prioritization of what demand is most relevant for the customer, whatever it is as margin, flagship stores, a certain channel, customers with a higher SLA, whatever it is that is relevant for you, and adding the planning policies what's your inventory levels, your SLAs, your allowance to actually pre-build the stock or postpone and deliver late, or your product substitution availability and then we can run a solver that actually will peg through the entire chain, that demand that we have at the ending point on the different days and provide visibility on the supportability of that demand and the optimal distribution, manufacturing and sourcing plans to actually support and service that demand. That's the main concept behind our supply chain solution and that's how we support distribution challenges, manufacturing challenges or sourcing challenges. Preston.

Mike Giambattista:

Pyshkoff 1.0. Interesting the phrase you just used. I talk to retail technology providers, technology providers from all across the horizon here, often many of whom are facing eyes in here, often many of whom are facing supply chain hurdles. But the concept, the phrase you just used, I think, is really an interesting one the supply chain digital twin idea, which there are other technologies out there that are kind of doing that. But I think somehow that phrase starts to bring some clarity into what you could actually do here.

Mike Giambattista:

So one, I think you should keep using that phrase because it's a really good one, but I can't think of anyone else. I could be missing somebody, but I can't think of anyone else who's kind of presenting their framework in that way. And I, you know, um, to be able to deploy solvers, as you call them, in different scenarios that uh can identify pain points, breakages, uh, you know, uh, cost and efficiencies and those kinds of things seems invaluable. And I know there are lots of other platforms that do those kinds of things, but they don't tend to present it in a kind of a digital twin perspective and I think that's really interesting. I don't know if that's part of the core of how you see this platform, or if that's just a phrase you came up with just now, but I think it's pretty good.

Santiago Garcia Poveda:

No, no, no, definitely. It's instrumental to O9's concept. I cannot take the credit for that sentence. It's definitely something that our founders have had in mind since the beginning of O9. And I think it's truly differential on the way that we think about operational decisions.

Santiago Garcia Poveda:

For retail In fact most of the market, for example, thinking of operational decisions like forecasting and replenishment, it's actually pretty much a constraint. It's very difficult to load on that replenishment logic all the upstream constraints of your supply chain, of the capacity of the employees that you have on the warehouse, the lead times, the track availability, adding all that into the replenishment. It's a complexity layer that not everyone is including on their existing solutions and one of our biggest customers actually in North America it's a very big retailer there that had actually tried to add that layer of constraining the replenishment for a very long time and struggled with many technologies because of the volume of data, the need of those solvers to be really performant and the scalability requirements. And what we actually were able to build with them is building those full level of complexity with all the constraints and running the replenishment being able to make decisions real time of three days.

Santiago Garcia Poveda:

From now you're going to have a peak that you will not be able to service. So we need to make decisions. Do we actually pre-build that stock? Do we postpone something else and run with a lower safety stock level for that week Because it's something that is less important? How do we prioritize what is important? Or do we actually invest in additional supply chain capacity for that week so that we are able to service that specific peak? It always happens before Christmas, before Labor Day, before kind of any of those moments and kind of handling that operationally linked with the supply chain perspective. I think it just enhances that value and that's the concept of integrating the commercial forecasting and replenishment distribution decision with the upper distribution view, with the upper distribution view of the supply chain.

Mike Giambattista:

I'm just curious with that large retailer you mentioned, how long did it take to go from zero to that level of integration where you could pull off that kind of analysis, I mean, and was it just a serious like a plug and play, we'll hook up to your APIs and we're ready to go? I can't imagine that being the case, especially with a large retailer. But you know, I think it could be instructive for people listening to this. You know what's involved in having that kind of operability.

Santiago Garcia Poveda:

Yeah, and the key challenge in this case was that the constraints on every node of the supply chain is not something that is available on any database that you say, let me just load it. So we actually have to work with them to go identify what's the capacity that we have on every single DC. We actually started trying to model everything from personnel to forklifts to picking space, to packet, to packing to storage on every single DC on their network and very soon they realized that the main constraint was personnel and personnel and loading base and we actually went into modeling those and not modeling the rest and in seven months the plan was six. We actually took seven to get life with that customer and with that solution, with those limitations of actually not loading every single constraint but loading just personnel and base. Aidan.

Mike Giambattista:

McCullen With that level of granularity that you mentioned earlier, jens Nielsen, yep, yeah.

Santiago Garcia Poveda:

Aidan McCullen that's remarkable.

Mike Giambattista:

I mean that's just remarkable to get that kind of complexity done in that period of time from scratch.

Santiago Garcia Poveda:

Yeah, that definitely was a big kind of challenge that we faced and we were able to actually deliver in time.

Mike Giambattista:

Yeah, that's massive. Just to preserve my credibility, we have to address AI, but I'd love to know how O9 is viewing AI, how you're deploying it. I would imagine it's a core feature set of some of the ways you're gathering and analyzing data. But let's hear from you, because everybody wants to know.

Santiago Garcia Poveda:

Yeah, it's the talk of the minute, so definitely, I mean AI to start with is a very broad field, so AI is central to the platform. Machine learning is part of AI and that's how we run forecasting in its majority. So basically, we have many AI use cases already plugged into a platform for years now, and when I say that we have an open analytics layer where we can run data science and machine learning algorithms, that's an AI layer that is open source right there. I guess that when you're asking about what we're thinking about AI today, we're thinking more about generative AI and large language models and what's now coming, and I think there are a couple elements here that are critical to us. One of the key changes that we believe that AI is going to bring to planning is that still around 80% of the knowledge on a company that is really important for planning is tribal knowledge on the minds of the planners, and that is something that, with this kind of large language models being able to process and structure data, we believe that over the next years, it's going to become the opposite. It's going to be 80% of it is going to be available and digitized and it will be a minority. That is tribal knowledge, and that will fundamentally change the way that planning occurs in companies, to the point where, eventually, we might be talking that companies are not competing on the street, like on the storefront, but on the power of their knowledge models that they have actually been able to build for their AI to be able to drive decisions. So we think that it's definitely central to everything that is going to happen in the next few years in the planning space, to everything that is going to happen in the next few years in the planning space.

Santiago Garcia Poveda:

The first steps that we've taken is to look at where can we add value today to our planning lifecycle with AI, and we identify several areas, I would say highlighting a few of them. I think one of them is on the configuration of the platform, sorry. So, basically, when we are deploying the platform and we're saying it takes seven months, it takes seven months because we still need to process data, you need to transform it, you need to load it, you need to structure it, make sense out of it. The minute that you can actually configure the platform through Gen AI, saying, hey, take this database, load it into the platform, merge it with this one and create whatever it is that we need to try to do or add an approve button that will actually jump to the senior manager and it will require this type of approval and we are developing those type of configuration steps that are not traditionally human-to-machine interface but more text-driven, and we believe that can be a game changer in the time it takes to deploy these tools and thinking on the value that is on the table in many of these transformations, accelerating the time to value by 50% would drive an amazing amount of value. The second one would be actually on the interface between planners and the tool to the human to machine interface.

Santiago Garcia Poveda:

Today is still very mouse driven and navigating through it and seeing the dashboards, but we see a wall in which this will be text or voice to text machine in a way where we will be able to say and we already have pilots on going with several customers and being able to say run in a scenario with a 10 increase of demand and tell me what's going to happen to my supply, and it automatically spits out well, you will have a problem in these products, in these locations, in this, and then it it's, and how can I solve it?

Santiago Garcia Poveda:

Would you be able to suggest any ways that we can resolve that say, well, well, if you actually bring this product by air and substitute this with this and allow this to be shipped from this location to this other store, you can actually solve 80% of the problems. There is these other three items that you don't really have an easy resolution for and that will again fundamentally transform the job of a planner in the way that the time it takes to really run a scenario, some interpreted scenarios, and be able to take decisions.

Mike Giambattista:

Can you even imagine a world where that could happen? So last question and you kind of alluded to it a minute ago but on a more personal level what is the most interesting thing you've ever experienced with chat, gpt?

Santiago Garcia Poveda:

That's. That's. That's a very good one. I, I, there are. There are many, I don't know if I would good enough. Interesting answer on this one. You hadn't sent me this one in advance.

Mike Giambattista:

That's okay, I asked somebody this question the other day and they said I can't tell you it's really good, but I can't say it in public, so that's fine too.

Santiago Garcia Poveda:

I would actually love that one. I could have come up with that one. No, I would say, one of the most interesting things that I actually did it's I asked Chad GPT to organize me a tour of three days on my hometown of where to go, what to see and where to eat, and, after many years living in my hometown, I actually was discovered a couple of very nice spots that I now can recommend to my friends. Amazing, completely. Things that I would have never thought about, and I wasn't expecting to be surprised by the output of a three-day recommendation. Amazing, all right.

Mike Giambattista:

That one will go on the record as a good one. Well good, santiago Garcia Pobeda. Maria, thank you so much for the time, the discussion, your ideas, the way that you and 09 Solutions are looking into the future and trying to bring it to the here and now. I really appreciate this and I'm looking forward to the time that we can do this again.

Santiago Garcia Poveda:

Definitely Anytime you want, Mike. It was a pleasure.

People on this episode

Head Shot

Mike Giambattista

Host