Unknown Speaker 0:00:13.4: 

Good afternoon, everyone. Hopefully, everyone enjoyed their lunch and enjoyed the morning sessions. I'm really excited to present one of our domain experts, Brad Eckhart, to talk today about how AI is elevating end-to-end retail planning. So without further ado, let's give Brad a big welcome applause and we'll get started. 

 

Brad Eckhart 0:00:34.6: 

Thanks everyone. Thank you, so I know you've probably - you've been hearing a lot about AI up to this point, so this session is going to be no different; we're going to talk some more about AI. But in this case what we're going to talk specifically about is how AI is going to be elevating the end-to-end planning process. My name is Brad Eckhardt. As Josh said, I'm a supply chain domain advisor with Anaplan. I've been with Anaplan a little less than a year. Prior to that, I was in retail planning and allocation and inventory management for many years and for a number of specialty retailers, particularly in fashion and fashion apparel. A lot of what we're going to talk about, a lot of what - are things that I've experienced hands on and so I hopefully can add some color to some of the challenges that people in the planning and allocation world face, and inventory management and merchandizing, and how AI can improve that.  

 

Brad Eckhart 0:01:41.2: 

So without further ado, we will get right into it. What I profess to start with is to say that when we think about the suboptimal outcomes of retail and the anatomy of retail, I would profess that most bad decisions aren't made by bad - aren't made by people. But they're more the result of some broken systems that exist within retail and those come from things like data foundations being fragmented. How many times have we all experienced that? The fact that cross-functional areas may have different guardrails that they're working in. For instance, they - you may have… If you think about as an example in fragmented data: how many times have two planners come into a meeting on a Monday morning and started - or come into work on a Monday morning and started - reviewing their data? The two planners are looking at two different sets of information and the sales numbers are different. They just spend hours in the morning going through and trying to figure out whose number’s right. 

 

Brad Eckhart 0:03:06.5: 

At the end of the day, they find out because one person is working off POS sales data and another planner is working off of a dashboard that they've pulled information from. The end result is that they discover that the difference is the return rates for e-commerce was in one and it wasn't in the other. So you spend time trying to come up with what that solution was instead of making a decision. That's an example of the fragmented data. Guardrails being inconsistent across the company can also result in some suboptimal decisions. For instance, you have a planning team that's working off of a margin goal, but you have an inventory planner who's working off of a target that they're going - that they need to hit from an inventory perspective. As a result of that, you have an inventory planner who launches a promotion to discount their outerwear to get out of the inventory and then the margins are missed for that month. That's an example of how, when guardrails are not aligned, that it can result in some poor decision-making.  

 

Brad Eckhart 0:04:16.3: 

The third one is disconnected intelligence and workflow, so insights you might think about as an example, you might have an AI tool that predicts a spike in demand for a particular item. But the output comes in a PDF report and gets passed off to the merchant. The merchant is working in a different system to write the purchase order in order to chase the inventory. As a result, you end up with a delayed purchase. The inventory is delayed and you have missed sales. Then finally, the fourth is a gap in between the time of the insights and when the actions can actually be taken. Even though the insights are clear, they don't translate into a fast decision. So for instance, a planner sees a fast sell-through on an item on a Monday morning, but allocations are not created in the distribution center until Wednesday. So the stores are out of stock for two days and there are missed sales. Those are all things that take - to take into account when you think about how these misalignments can result in poor decisions. But it's not necessarily the decision-maker who's the result of that; it's the infrastructure that's in place that's leading up to that.  

 

Brad Eckhart 0:05:36.3: 

For those of you who attended the keynote this morning, you may’ve seen a slide similar to this that EJ presented. What I'm going to do is take a slightly different approach at it and say what we're… What I just talked about, about those different challenges in terms of decision-making, how that relates to these layers. When you think about we have a unified data layer and then we also have these functional integration layers and the integrated workflows. So basically, what we're looking at here is you… Everything, especially as it relates to AI, everything starts with accurate data, right? One of the biggest challenges that I think retailers are coming to terms with now is: in order to really have a strong AI strategy, they really have to start with the data. They have to make sure that the data are in one location and they’re accurate, and that those data roll up to a consistent source where that then the AI can be layered on top of that to make those kinds of decisions.  

 

Brad Eckhart 0:06:47.9: 

So with Anaplan, what we are offering is the data orchestration level which is going to cleanse those data, put it in place so that then the applications can layer on top of that. The applications all speak to one another in a way that you have true end-to-end planning. That is going to allow us to have the AI layered on top of that in order to make more - quicker decisions and more accurate decisions. When we think about the data layer, often data are fragmented or inconsistent and siloed. Those key data streams - like sales, inventory, demand signals, etc. - those are all the data signals that are critical to be in a consistent place, with consistent metrics that are aligned across the company and with multiple sources - with one source of truth.  

 

Brad Eckhart 0:07:51.7: 

Once we have all that strong data foundation, then the next question is: how do we turn those data into decisions? That's the functional intelligence layer and that's where we're looking here at the applications that are going to achieve that. The intelligence level is built on domain-specific expertise. It's built on being context aware and aligned to the business logic. That's really what… It turns not just into predictive, but into a prescriptive planning process. For an example, it's not just like - it's not just answering: I think I'm going to be able to sell 500 units, but it's where do I place that inventory? How do I flow that inventory and how do I optimize for instance the size curves and the timing of when that inventory is going to be in place in order to maximize those sales? Forecasting alone isn't enough; it really needs to be optimization. We need to optimize in addition to that forecasting. The optimization is really where the business rules come into play and how we're going to utilize those forecasts to get to the right business decisions. 

 

Brad Eckhart 0:09:16.3: 

That's really where the integrated workflow comes into play. The integrated workflow is really the key decisions that are aligned to the enterprise strategy. Again, the applications are what's going to help achieve that. So for instance, the top-down strategy and a bottom-up execution so that the plans are aligned both from the top down and the bottom up. A financial plan in turn drives an assortment. The assortment in turn drives allocation decisions. Those allocation decisions feed back into the financial performance. In the end, the outcome is the strategy and the execution stay continuously aligned throughout the organization. Then when we talk about decision acceleration and how we're going to make decisions faster, that's where the AI layer comes into play. Basically, we want to be able to interrogate the data instantaneously and be able to surface root causes of any results that we're reviewing, and then tech come up with recommended actions.  

 

Brad Eckhart 0:10:30.8: 

So for example, a planner may ask: why is the margin down? The instant diagnosis would be coming back and saying, ‘You missed that margin because there was a markdown taken in this product category that took your margin below plan.’ As a result of that, it would recommend an action to be taken in order to relieve that margin drain in the future and come back with a better margin performance. But the requirements to accomplish all this are, you have to have a strong foundation of the data. You have to have the intelligence. So you need that domain expertise that's built in that, that the individuals can build into the system. Then you have to have a strong workflow that brings all of those processes together. 

 

Brad Eckhart 0:11:32.2: 

When you look at the… If we do a little bit deeper dive into the applications and do an overview of the applications, this is how really Anaplan delivers on the framework. Within the connected applications we - you've heard about Merchandise Financial Planning as an application and how that relates to the financial… That sets the financial guardrails, which in turn leads into the assortment planning process, which builds the profitable collections or the assortments. That, in turn, will feed into the allocation and replenishment system in order to execute in the market, so they're not standalone tools. They operate as one connected system. That is in essence where we - where Anaplan can help with that end-to-end planning execution. Further enhancements, which would come in for size optimization, which is more prescriptive size level recommendations, and price optimization as well through optimizing promotions and markdowns. The really core message here is that end-to-end - true end-to-end AI-enabled retail planning is what AI is going to help to drive in the future. 

 

Unknown Speaker 0:13:03.5: 

Brad, we have a question. 

 

Brad Eckhart 0:13:05.2: 

Yes, sorry. 

 

Audience 0:13:06.3: 

This is a question. The vision is to have AI look at the models, but would it make sense to have it look at ADO as well? You just want to have a look at the underlying [inaudible 0:13:15.7] sometimes. Like when you're looking through using Analyst and things like that, you need to go and drill down into the accruals, which resides over that part of the vision? Or is that something… 

 

Brad Eckhart 0:13:30.2: 

Well, ADO drives into all of these applications, so the data are coming… 

 

Audience 0:13:38.5: 

As mentioned, if you're looking at a transactional level, it’s because we have that level of detail, so you're looking at more than two places - unless you do a low-level model, which is counterintuitive. 

 

Brad Eckhart 0:13:50.4: 

Yes, that is a little over my pay grade in terms of technical expertise, so I apologize for that. 

 

Unknown Speaker 0:13:57.4: 

That’s a bit that Ade can help out. 

 

Brad Eckhart 0:13:59.5: 

Okay, thanks Steve. 

 

Audience 0:14:00.0: 

The vision is to leverage ADO as that framework that we would go through, whether that's through the [unclear word 0:14:04.9] layer or through the UX experience. So you can drill on to any level of that model, down through all that transaction and by leveraging your ADO framework. You’re exactly right in terms of that's where that orchestration point comes in. 

 

Audience 0:14:21.3: 

So you'll be able to point to your transactional-level data on ADO, which isn't necessarily [inaudible 0:14:25.9].  

 

Unknown Speaker 0:14:27.0: 

Correct, yes, that's right.  

 

Audience 0:14:31.4: 

You probably can't tell me, but when the roadmap is there? 

 

Unknown Speaker 0:14:36.5:  

I don't know exactly within the roadmap but we [inaudible 0:14:39.4] on that one. 

 

Brad Eckhart 0:14:44.5  

Thank you. So let's think about planning, say, five years from now. We can cover the exciting innovation-related change first, but I also want to make sure that we include some reality checks to that. When we think about the future of planning, continuous planning is the theme where we have a current - where the current state, the planning process generally is more linear today. The evolution is from input, act, input, act - and that's the linear process that we have today. But we really want to be in the future - we want to have it be more of a continuous loop and real - having that integrated real-time process. What that looks like is, instead of that linear handoff, planning becomes a live integrated loop. A sudden sales spike for a new jacket, for instance, captured by allocation system. The signal instantly updates the assortment plan, which flags the item for a potential chase.  

 

Brad Eckhart 0:16:06.1: 

Simultaneously, the merchandise planning - the merchandise plan recalculating the open to buy in real time and models the margin impact of expediting a reorder versus pulling forward a different item. A system then presents a signal unified recommendation - a single unified recommendation and approves a $100,000 chase order for the jacket, funded by a 5 per cent reduction in sweater budget to capture an estimated $350,000 in upside. So that's just an example of how that continuous data movement can effect a quick response. 

 

Audience 0:16:43.1:  

The current work that we are helping data in real time. We have to run actions, which basically slows down the models. Are we talking about not having those actions and, still, the model works smoothly, it being just data real time? 

 

Brad Eckhart 0:17:00.5:  

It’s not real time. Not real time, but even on a day-by-day basis, that would be the activity that you'd still be able to react more in pseudo real time, if you will. So despite some movement towards automation, current process is still a one-to-one replacement of individual tasks. The future state would be where the technology handles the synchronization. Humans focus on the judgment, so the agents would unlock the next evolution of this; a system of a human in the loop, with human in the loop checkpoints. Cross-functional coordination in context. For instance, we would be - we wouldn't be having… In the future you wouldn't be having as many meetings. It would be current state. The information sharing and process alignment across teams is much improved compared to where we started, but fewer spreadsheets and more and more shared dashboards.  

 

Brad Eckhart 0:18:23.9: 

But here's the reality: why is this hard and why hasn't it been done today? Lacking the unified data model is one reason, and that additional foundation layer and organizing that inertia. So the lack of a unified data model is one of the challenges that we face today. The organizational inertia and getting the organization to… Sorry, oh, there we go. Getting the organization behind AI and having them to agree and having individuals start to understand and trust the AI forecast is another challenge that we face today, but we expect to improve in the future. But that won't arrive by default. The roles themselves in the organization will have to evolve. The deep domain expertise becomes more critical, not less, but it must be paired with a new skill set and the ability to interpret and trust AI-driven recommendations. It also requires significant training and upskilling. So moving teams from spreadsheet - what we call spreadsheet jockeys - to strategic portfolio managers. So the most successful planners of tomorrow will be those who can confidently manage the system, not just operate within it.  

 

Brad Eckhart 0:20:17.2: 

So closing things out, I'll share a visual that I think covers what we're building toward with the new foundation. It's a framework we call sense, decide and execute. First AI-powered systems since market changes and democratize those inputs across the business in real time. This allows individuals to confidently decide in the moment and armed with clear understanding of the financial and operational trade-offs thanks to AI recommendations. Then finally, the teams can execute that strategy in unison because their actions are guided by shared data and intelligent guardrails and, increasingly, automated by AI. So this continuous loop - sense, decide, execute - is how retailers can finally overcome the friction of disconnected systems and unlock the full potential of their teams. 

 

Audience 0:21:20.5:  

For example, what we have been hearing very well yesterday and today about this continuous planning and continuous updating, but still, we are going to use the Anaplan platform. I don't know how this works, but for example for Anaplan for the [unclear words 0:21:39.6] you only set by the year. For a specific year. Let's say we do continuous … and say: if I'm going to plan it 52 weeks all the time, it’s going to cross over the next two years. So how are we going to handle such cases? We have this scenario now where we get the data always a little [?old 0:22:01.8], 52 weeks. But our years, as I said, either one year or two years, ten years [inaudible 0:22:10.6] are going to be changed and re-changed. So how are we going to handle those - I don't call it limitation - but like design and how it is today, and how are we going to handle those things? 

 

Brad Eckhart 0:22:23.2: 

So if you didn't hear this: the question was about how to handle in a continuous planning environment, how to handle that planning process. If I understood your question correctly: in your environment, your timing is a set timeframe. So you're planning in 12-month increments or however many months of increments. I don't know if there's a technical restriction to that in the format that you're in, but I know that in other Anaplan environments that calendar can roll. So the planning horizon can be continuous, so you can add… It doesn't have to be set in 12-month increments. 

 

Audience 0:23:12.1:  

So for example, I have two years that I think - the current year and one [?ahead 0:23:18.4] year. For example, when I apply, if I set it to run at 52 weeks it will go to the next year at some point. In that case, I have to change it there. I still have to have the past three years, but also I need additional weeks from the next - the second year. 

 

Brad Eckhart 0:23:50.3:  

But I've seen planning environments within Anaplan where, on a set timeframe, you can roll; you can roll that planning timeframe. So you could add, say - I've seen it most frequently done on a quarterly basis, where a quarter - every quarter, you add another quarter to the back end and drop off a quarter from the front end.  

 

Audience 0:24:09.2: 

You can do that? 

 

Brad Eckhart 0:24:10.1: 

I've seen it done that way. Yes, we… 

 

Audience 0:24:15.1: 

What if you only had one year? It has to always be one year out? You can drop one year and add a year? 

 

Unknown Speaker 0:24:26.4:  

You can do it with custom planning. 

 

Audience 0:24:28.4:  

Oh, not with the calendar. Not the calendar? 

 

Unknown Speaker 0:24:32.6: 

Custom planner. The years are rolling currently. 

 

Unknown Speaker 0:24:38.9:  

Sorry, what is that? 

 

Unknown Speaker 0:24:38.6: 

I plan two years rolling, but I will keep the current year… 

 

Audience 0:24:42.5  

Yes, so let's say that one of my questions is that you are always planning with at least two weeks. [Aside discussion 0:24:50.4] That means next week it is going to be an additional one week.  

 

Unknown Speaker 0:24:53.4: 

That’s why we leave the second year open. 

 

Audience 0:24:54.2: 

That would be right? What is that? 

 

Unknown Speaker 0:24:57.2:  

That's why we leave the second year open. 

 

Audience 0:24:59.1:  

Right, but at some point, though, you would cross over. When you do that, I said you don’t have to add an additional one year, which you don't want. You want to keep two years, so what you are saying is we have to use this custom planning, not… 

 

Unknown Speaker 0:25:13.3:  

Yes, custom planning is the solution. When it comes to Anaplan’s basic time it is there, and yes, it can be rolled only for one year. We cannot just do it quarterly, but you can use of it, with it being done quarterly with this custom plan. Custom plan is simply creating a normal list. 

 

Audience 0:25:34.0: 

A list of planning, right. 

 

Unknown Speaker 0:25:35.1: 

That looks like a time list - which is very, very common, or maybe half of the models. I'm just making up the numbers, so maybe half of the models have custom plan but it's very convenient with the subsets of time and it just keeps your model sizes [inaudible/over speaking 0:25:53.2]. 

 

Audience 0:25:59.5: 

For example, some of the formulas, you cannot use with the custom planning. 

 

Unknown Speaker 0:26:05.2:  

Yes, but there are workarounds, but you can switch back and forth. 

 

Unknown Speaker 0:26:10.3:  

I always look for visual information as well. We actually have a lot of our retail experts to do demos, and so they're in the hallway and so they can work to help get that… 

 

Brad Eckhart 0:26:20.4:  

Yes, that's a great idea. 

 

Unknown Speaker 0:26:21.2:  

Some are going to dive deeper into some of these individual comments. 

 

Brad Eckhart 0:26:25.0:  

Thank you, Josh. I think I was just getting into the execution phase. We talked about sensing, which is the AI-powered system to sense the market changes. The individuals would then decide and then finally, the teams would execute that strategy in unison because their actions are guided by shared data and intelligent guardrails. Then this continuous loop - sense, decide, execute - is how retailers can finally overcome that friction of disconnected systems and unlock the full potential of the teams. So I wrapped through that very fast. Yes, Ali? 

 

Audience 0:27:18.4:  

Going back to the retail applications, the [aside discussion 0:27:23.6]. Is this on? 

 

 

Brad Eckhart 0:27:24.1:  

Oh, there you go. 

 

Audience 0:27:25.3:  

Yes, this slide. Where are you seeing retailers start on this journey? Is it you have to start with MFP? You should? You can? Where's the right place to start the journey for what you're seeing with the retailers? 

 

Brad Eckhart 0:27:41.2: 

Yes, good question. Really, there's no right answer to that; it really depends on the organization and where you are in your maturity of the planning process. Some retailers like to start at the beginning, so they start with Merchandise Financial Planning. Then they move from there into Assortment Planning and finally into Allocation and Replenishment. But other organizations who maybe have a much more stringent ROI expectation from their implementations, we’ve found that they tend to start in Allocation and Replenishment, because that's the easier place to prove the ROI. But really, it just depends on the organization. The beautiful thing about Anaplan and what you see in all of these applications, is that they can… Once the platform is in place, these applications can be implemented just like the models do today, in independence from each other and then they can be incorporated or integrated later. 

 

Audience 0:28:45.2:  

We've also started a DC balancing. 

 

Brad Eckhart 0:28:49.1: 

DC Balancing? 

 

Audience 0:28:50.3: 

Yes, [inaudible 0:28:51.5] model as well, in that case we will … and then move to the other direction. 

 

Unknown Speaker 0:29:01.2:  

Also, starting with value data is more structured and you get the most production-related data. This can be smart. 

 

Brad Eckhart 0:29:10.2:  

Data are always king, right? Everything starts with the data. Then what I mentioned earlier is very true. AI is - I think it's really bringing to light with many retailers. Any time you do start to implement a system, you oftentimes run into data issues. When I was in my consulting days, many of the implementations were delayed at the very beginning because hierarchies weren't aligned, or I thought we were going to be able to plan by attribute, and our attributes aren't consistent. There are a million different issues that you can run into. So reviewing those data and having your data in a good place is always the best place to start, for sure. Trying to think of what else we could cover, because I did go through these things very quickly. 

 

Unknown Speaker 0:30:12.4:  

Brad, do you want to [over speaking 0:30:10.1] the survey results or… The survey results. 

 

Brad Eckhart 0:30:16.3:  

Oh, yes. Thanks. 

 

Audience 0:30:18.2:  

I have a question: where are you seeing agents on this, because we're talking agents in all sorts of platforms everywhere. They're going to bring us together. Where are the Anaplan agent benefits that you see coming in and [inaudible 0:30:35.4]? 

 

Brad Eckhart 0:30:36.4:  

Yes, well, the agents have started with… In the platform there is, well, CoModeler is one of the best places to start because it really saves your modelers a lot of time. But then the CoPlanner, which is now morphing into the agents in individual - of the individual applications. So one of the things that… Actually, if you want to stop by one of the booths, they can provide… Some of the individuals from Anaplan can provide you with some examples of where those agents are starting to be incorporated into MFP, for instance. That's a lot of what we're seeing in the future in terms of that connected planning and that closing that loop and getting those speed - those decisions. When I was talking here, this bottom-right corner insight and actions are separated in time. So speeding up those decisions and having the ability to use agents to help make those decisions faster, is really what we're excited about in terms of where AI is leading and where we're headed with AI and in the applications. 

 

Audience 0:32:06.2:  

I was going to say, I can, I'm an Anaplanner, but what I've seen my customers do with agents, I think it's cool to talk about the why: it’s like why do I miss my forecast or what happened in this year? But what I'm finding is just a simple thing of on-boarding and change management. I used to do this in Excel over here and I went through the training; where is that thing again? Whereas this model, it’ll always take you there, so it seems really simple. But those teams are - it's getting them more comfortable with Anaplan because they can just interrogate it and ask it: where is this thing or where do I go with this in my forecast? The way their leadership is thinking about it at that highest level is that their CEO is about to go into a meeting. You can just go ask it real quick: where are we at with orders? Why are you in this and then worry about $30 million this quarter, which is a pretty good question. Maybe don't ask an agent that, but it'll generate graphs.  

 

Audience 0:33:00.6: 

Okay, can you tweak it to take this year out? Okay, boom, you're done. They're looking at it in terms of the executives who aren't in Anaplan. They’re not planning, but they may have a quick question going into an executive meeting that they can ask it instead of sending it to their team to go figure it out for two weeks to get it, so the executive winds up not using it in the meeting at all. So I think it depends on what level you're coming from, but I don't think it needs to be pie-in-the-sky, really big use cases at once. It can just be as simple as: where is this or where is that? 

 

Brad Eckhart 0:33:36.3:  

Yes, so also what I've seen is from a training tool also - and using MFP as an example would be a good one because there are a lot of measures in Merchandise Financial Planning. If you have new people who may not understand the calculation of open to buy, being able to interrogate through an agent and have it search on the formulas that are within that, is a really practical application of agents and a really easy way, to Ali’s point, of getting people comfortable with AI and comfortable with that. Using an agent is a little like ChatGPT, so it gets people comfortable with it. 

 

Audience 0:34:21.7: 

Yes they do. 

 

Brad Eckhart 0:34:22.5: 

Yes, exactly. 

 

Audience 0:34:26.2:  

[Inaudible 0:34:27.0], so yesterday I was also attending one of the demonstrations and that was so much deeper. But my question is, yes, I think the focus was mostly from the data side. I have [inaudible 0:34:42.3]. What about with regard to formulas and stuff that I can see. Users are hard to come to, if I want to analyze, I write the formula … do we have something that's in the pipeline or already available from the AI side? 

 

Brad Eckhart 0:35:03.3:  

I'm not sure I understand the question. In terms of the… 

 

Audience 0:35:08.1:  

For analyzing formulas [inaudible 0:35:06.9], where do we have AI that’s … currently? 

 

Unknown Speaker 0:35:14.9: 

Can you repeat the question? I think… 

 

Brad Eckhart 0:35:17.3:  

Yes. Do we have… You're asking if there is AI integrated into the system? Oh, thanks, Josh. 

 

Audience 0:35:25.3:  

Yes, for analyzing formulas. 

 

Brad Eckhart 0:35:26.4:  

Yes, go ahead and repeat the question, so Steve… 

 

Unknown Speaker 0:35:29.1:  

So if you're talking at the model level, so that's what - in CoModeler, so if you're one of the model builders you can use CoModeler to explain things like the formulas inside of the model. It's a good way actually to - for somebody new to come into a model they may not understand and ask it questions and give some explanations, including things down to the line item level as well as formulas. Was that your question?  

 

Audience 0:35:57.1: 

Yes. 

 

Unknown Speaker 0:36:01.5: 

I think we're just about out of time. Were there any more questions? 

 

Brad Eckhart 0:36:03.5:  

Oh, that was fast. The one thing I did just want to close on is - and I missed this in the presentation - but there was a study that Anaplan did along with the World Retail Congress that showed… I'm not sure how I missed this. That showed that about five per cent of profitability was effected through lost sales as a result of the different things that I spoke about in terms of on this slide, which is really what led… Sorry, yes, I don't know why this has hung up now [long pause]. Really, what led to these four misalignments is from that study that was done, where about five per cent of the profitability was lost as a result. They could be equated to these four things. That study is going to be out I think this - in April. You can find it either on our Anaplan website or you can find it on the World Retail Congress website. Okay. 

 

Unknown Speaker 0:37:47.4:  

Well, fantastic. I want to thank Brad so much. 

 

Brad Eckhart 0:37:48.2: 

Thank you all. 

 

Unknown Speaker 0:37:47.9: 

Thank you for attending this session. We're really excited as well for our next session, so we'll hope you'll join us here back at two o’clock. We'll have actually a customer panel session with Nvidia and so now there is a break for the next 20 minutes. We'll start again in this room at two o’clock. Thank you. 

 

Brad Eckhart 0:38:04.4:  

Thank you all.