AI Readiness in Supply Chain Operations: Insights from Industry Leaders

Alan Pelz-Sharpe: Well, hello everyone, and welcome to today’s Hyperscience webinar on AI readiness in Supply Chain operations. I’m Alan Pelz-Sharpe, the founder and principal analyst of Deep Analysis and also your moderator for today’s discussion. I’m joined by three industry leaders who bring some seriously deep expertise in supply chain operations and AI-driven automation. I want to highlight where this all comes from. This was a groundbreaking new piece of research. We worked together with Hyperscience who sponsored this and underwrote all of this research. And we did it also in partnership with CSCMP, the Council of Supply Chain Management Professionals. This research provides first of their kind insights into the potential impact on supply chain operations, with a particular focus on back office operations.

Alan Pelz-Sharpe: You can all get hold of this study. It’s downloadable and free. It’s a pretty hefty study. The survey is a good survey. This was done seriously with a qualified panel of 300 senior leaders in supply chain across the United States of America from organizations that had a thousand or more employees. There was quite a heavy weighting in this one towards people who are involved in IT operations and senior leadership. We set out to challenge some preconceived notions. There is definitely a preconceived notion that supply chain operations today are behind the times when it comes to office paperwork. Office paperwork isn’t the sexiest topic in the world, but it’s really important across supply chain operations, be that in warehousing, transportation, and logistics. Anybody who works in this sector knows full well that we’re sort of drowning in paperwork and Excel spreadsheets.

Alan Pelz-Sharpe: I’d like to hear from David Teeple from CLAC and also very much involved with CSCMP. David, tell me a little bit about yourself and what were you thinking when this study started? What do you think about the role and opportunity for AI in supply chain?

David Teeple: Well, Alan, thank you for the opportunity. I think it’s such a relevant topic these days. You certainly see it on every post or every other post you see on LinkedIn. It’s on top of everybody’s mind. So it’s something that we’re all very interested in understanding. I think there’s a lot of confusion there about what it really is, but I think everybody’s very interested in what it can do for them and their business.

Alan Pelz-Sharpe: Excellent. Ivan Ramirez from Hirschbach, you are actually in the early stages of doing some of this stuff. So what opportunity do you see here with AI in supply chain?

Ivan Ramirez: We definitely are. I’ve been with the company for about 11 months. I come from an e-commerce supply chain background where obviously we’ve done a lot in supply chain automation. The opportunities here in transportation are big. While I’m not a fan of just implementing AI just for the sake of implementing AI, I wanna be very intentional and purposeful on how we implement it. We’re still in the early stages. That said, we’ve identified some compelling problems where I think AI is gonna be a strong fit. One of them is actually the billing side of our business, which still has a tremendous amount of paperwork. That’s an area where we’ve gone out and looked for strong partnerships because it’s a difficult problem to solve. It’s many types of documents and you’re looking for many unique characteristics and information in those documents at the customer level. So AI is definitely something that we wanna deploy into solving this problem.

Alan Pelz-Sharpe: Awesome. Thank you. And last, but certainly not least, Chip VonBurg from Hyperscience. Chip, so you are coming out from the perspective of the people who actually supply this technology. You’ve heard Dave mention there’s a lot of confusion and that it’s early days. I’m assuming you would agree with that, but what are you seeing happening and what do you see the opportunity being?

Chip VonBurg: It’s interesting. I’d echo a little bit of what both David and Ivan said. For sure it’s on the top of everybody’s mind. But I think Ivan’s point is really interesting that it’s sort of reinvigorated a lot of folks to think there could actually be an answer to this. So much of what I’ve seen over the years is difficult manual processes. Because of that, folks have sort of kicked the can and said, “well, we’re gonna solve this other stuff first. We’ll come back for that.” This resurgence of belief that technology can actually help you hit your goals is probably one of the biggest benefits that we have.

Alan Pelz-Sharpe: Talking about AI in general and everybody’s excited, I think this is insight number one. 98% of respondents whose organization use AI see it as useful, important, or vital. I mean, you just don’t typically see statistics like that come out of a survey. That to me was a surprise. But Ivan, you sounded quite bullish on this. I’m guessing you would’ve been one of these who said it’s very important. What do you make of this?

Ivan Ramirez: I think it is. It can’t be ignored. Like I mentioned earlier, you have to be very intentional and purposeful of where you actually deploy AI. And not to confuse that some of these problems can actually be solved with software and not necessarily AI. That’s the most important part is identifying the root cause of the problem and saying, “okay, is this a software problem or an AI problem?” and then making the right decision there. We’ve got some POCs that we’re obviously working on. I can tell you categories. One of them is customer service, and then dispatching, and then billing. Customer service is a big one for us where we think AI could be very beneficial. It’s one of those where the components, meaning a customer service ticketing system and knowledge base, have existed for years. But AI allows you to extract that information in a very simple way.

Ivan Ramirez: Billing is a big one for us. I still can’t believe the level of complexity for billing in transportation. That is an area where that’s definitely not a software problem. Part of it is software, but most of it is AI. You’ve got the intelligence you gotta have to scan those documents to apply business rules to those documents, which today gets done by humans. They’re actually pretty good at doing those things, but it doesn’t scale. When you’re talking about doubling, tripling the amount of business that you do, that doesn’t scale. And to train people for that, it’s super complex. These are people that have been doing this stuff for 15 years.

Alan Pelz-Sharpe: And Dave, when it comes to companies that are cutting edge, like Hirschbach who are actually starting that journey, what separates companies like that? What separates those companies that are willing to take that first step and those who are not?

David Teeple: No. I mean, certainly you’ve got concerns about what’s the right thing to do. There are companies that are coming outta Covid concerned about resilience and making sure that their business stays relevant. But the real opportunity here is for those companies that see that technology is vital. They’re industry leaders. They don’t see it as a might need. They see it as a must have. It’s part of their long-term growth. They see it as a competitive advantage. It’s a key driver in innovation. Versus the companies that are kind of toeing the line, they’re risk averse. They’re a little concerned about being kind of bleeding edge. They’re more concerned about making sure that they’re solving the problems that are in front of them.

David Teeple: We’re in a very interesting time coming out of Covid. Many organizations were caught with low infrastructure. They weren’t able to meet the demands. They cranked up their infrastructure, they cranked up their processing. And then the bull whip kind of came into the down cycle where there’s softness in the market. And so now everybody’s focused on really what do I do to make myself more efficient? The industry leaders are really focusing on AI and digitization, which I think go hand in hand here. Those industry leaders are getting out in front of it. They’re using this time to focus on making themselves stronger and making themselves more competitive in the market.

Alan Pelz-Sharpe: Chip, who are the people within those companies that are hungry for change? Who are the ones that want to listen to Hyperscience’s message, for example?

Chip VonBurg: I think the teams that we’re mostly seeing lead the charge on this are the technical teams. It’s the IT teams, it’s the centers of excellence. Those are the folks that are owning the technology and bringing the technology into the organizations. But I think that is actually driven by leadership. This is exactly what you came up with in your paper. There’s a strong sense from the leadership that we need to adopt these technologies to do things better, faster, more efficient, and IT are the folks that are being charged to make it happen. The good thing though is business is in there as well. What I’m also seeing is business is very quick to raise their hand and say, yes, we want this. We also want to do things better. So although IT is the one actually leading the charge, I think you see a little bit of all the major pieces of the business involved.

Alan Pelz-Sharpe: So insight number two. Confidence in the data provided by these systems—SCM, WMS, ERP—was at 81% thinking that it’s really good. Now, this was a surprise to me. I’ve done other surveys where we’ve asked similar questions to this one, definitely around ERP. And I can tell you with absolute assurance, when you talk to the people who actually use those systems on a day-to-day basis, they don’t have much trust in the data. So there’s that. And I also think it’s just really picking up on the point that you all sort of made: you can’t always fix the past, but you can do it better from now on.

Chip VonBurg: So I just I gotta step back a little bit too and say this was one of the ones that surprised me as well. As somebody that used to own ground truth systems, it was always kind of a common understanding of, “well, the data is sort of our issue, and that’s really difficult to work through.” So I was surprised to see this as high as it was. Let’s just take this at face value. If it is truly 81%, that means there really is room for growth here. What about the other folks that really don’t trust in the data? I think the thing that you really have to take away when it comes to the data is garbage in, garbage out. It’s the oldest saying in the world, but it’s true. And I think it becomes so much more true now that you’re employing these AI systems.

Chip VonBurg: The new terminology that I’m using is data is the fuel for AI. AI is this wonderful set of algorithms, but without data, it does nothing. So you really have to put thoughts into what is the data that I’m using. Ivan mentioned knowledge bases. If you’re putting a chatbot or a phone system hooking up to that, and somebody goes into the chatbot and says, “you know, I need to expedite this order,” and it gives them a completely wrong response, what does that do for you? I think that’s really the big takeaway: the importance of good, clean, reliable data can’t be understated. Luckily from our standpoint, this has always been something that we’ve strived for. And this is the reason that we have harnesses and guardrails around the data that comes out of Hyperscience. So that when we start to talk about things like redirecting that back office data into those AI systems, you know that you are a hundred percent confident in how clean that data is.

Alan Pelz-Sharpe: Dave, in your day job and your work with CSCMP, how do you guide people? Do you trust the data in their systems? Do they trust the data in their system? Where do you advise people where to start?

David Teeple: Well, certainly data has been historically bad in supply chain. It has been a paper-based, human-based process that is subject to error. Once you get that human error involved, then that makes the data suspect, and then the decision making that potentially comes out of AI-based systems then becomes questionable. So it becomes super important to migrate into the 21st century regarding the gathering of data and getting a much more reliable system in place to clean up that data going forward. You can spend a lot of time cleaning up old data, but I think you gotta start by fixing your processes to get the data going forward to be clean and trustworthy. Then you can start to adopt AI as a tool that enhances your operations. It makes people more productive. It doesn’t replace people, but it becomes an extremely powerful tool to streamline, make people more effective, and reduce the labor involved. It doesn’t scale well if you have to add people because there just is no labor out there. There’s such a competitive environment going after the same fixed resource.

Alan Pelz-Sharpe: And Ivan, I’m not gonna put you on the spot and say how accurate is the data in your company? But how do you approach this? What’s sort of an acceptable level of tolerance for accuracy?

Ivan Ramirez: Yeah, these numbers really, I was quite shocked to see these numbers, specifically the TMS one, given that that’s the world that I live in. I think I’m a strong advocate for data accuracy. But I will tell you that achieving it is very difficult. Even as digital native organizations, it’s still very difficult. And many applications have been developed in ways where they unintentionally have compromised data quality. We went through an example of this with Chip when we were hashing out through some of this stuff. We had a specific area where our data quality was off. We were using comment fields to drive billing in a certain area. And the Hyperscience team was like, “oh, we can solve this with AI.” But my immediate reaction was like, “no, let’s solve it upstream.” It was a simple solve upstream of identifying the top 10 types of accessorial billings and having a dropdown in the application instead of a free comment field that would allow the user to select that specific accessorial type and then have a conditional field that would say, “oh, this requires an actual value in it because it’s value driven.” And then have that data come downstream.

Ivan Ramirez: So I think that the thing that I would caution us on is that we can get lazy upstream. We can get really lazy with upstream systems knowing that AI can do a lot of this data cleansing and data cleaning. But in my opinion, if you’re having to clean up data downstream, you’re introducing a lot of complexity. So I’m a big fan of cleaning it upstream. Now, in reality, sometimes cleaning data upstream is really challenging ’cause it could be months of development or weeks of development and you don’t have that time. So I’m not saying that’s the only way to do it. It’s the ideal place to go and fix the problem. But in some cases, because of resource constraints or legacy system complexities, you can’t do that. So you’re gonna have to do it downstream with AI.

Alan Pelz-Sharpe: So this next one sort of ties into everything we’ve been talking about really. 82% of respondents reported that manual document processing has a heavy to extreme impact on operational efficiency. Now, anybody who works or has been associated with supply chain probably is not a surprise at all at this. I have shared this statistic with people outside of supply chain and they don’t believe it. But this is the real world of supply chain: lots and lots of documents. And if we throw Excel spreadsheets into that bucket too, then we’ve pretty much got what the supply chain runs on. But David, this is a chasm. Why are we at this point again? Are there misconceptions that they don’t understand that the technology’s moved on so far?

David Teeple: Certainly there’s a lot of confusion as to what it is, what it isn’t. But there’s also the traditional issues associated with any new technology. There’s always that fear that the technology is not mature enough, that they wanna wait and see and make sure that it’s a proven technology. People worry about issues with change management within their organizations. There’s fear of job displacement. There’s fear of that being too expensive and too complex. People don’t necessarily have the right skillset within their organization to really kind of take this on. They also worry about the cost, that it’s something for the big corporations but doesn’t work with my company. So again, that lack of understanding and their resistance to change has been a big barrier for many companies to kind of get over in really embracing it and running towards getting it implemented.

Alan Pelz-Sharpe: Chip, is improving operational efficiency enough? Or is there a possibility here to actually essentially unlock new business opportunities?

Chip VonBurg: I think there are other opportunities that come out of it. Most companies come at this with the efficiency play: “we could do this better and faster.” But think about all of the other pieces of the business and customer touch points that could ultimately be affected. AI is a tool. It’s just like any other tool. If you look at the handheld devices that shipping companies have now, it’s a tool. Before that they did things on paper clipboards. So when they got the handheld devices, they said, “ah, this is gonna be faster and easier.” But then think about how that changed other pieces of that delivery service. As a consumer, now I can sign for packages easier, they can scan my ID to validate that it’s me. They can take a picture at my doorstep. They changed the business, frankly.

Chip VonBurg: Customer service is a big one. Anything that you can do better with your data, you can start to affect customer service. And the great thing that comes out of great customer service is customer satisfaction. That would be the big thing that I would sort of underline here, is don’t forget about customer satisfaction. Because if you are able to deal with your data better and faster and more efficient, and you can make it easier for your customers to deal with their data through you, how do you now represent your company? You’re now really looked at by that customer very differently. So I think efficiencies are definitely a big piece of this, but there really is so much more that touches the customer that can come out of this.

David Teeple: That’s absolutely such an important point, Chip. Customer service, customer expectations. If we’ve learned nothing from COVID, it should be that there are very high customer expectations. So anything that we can do to give visibility to the customer, enhance that experience, will have a direct impact to the bottom line.

Chip VonBurg: And customers want to be served in different ways. It used to be just email. Now it’s email, portal, chat, text, you name it. It’s multi-channel.

Alan Pelz-Sharpe: Ivan, I want to come to you on this next one. Insight number four: the absence of standardized document formats is a pervasive issue. Nobody’s invoices are alike. Do you believe that by moving to more of an AI driven approach, that this can be managed more effectively in the sense that AI can actually make sense of all of these different document formats?

Ivan Ramirez: Yeah, I do. I’ve seen a couple of these things throughout my career. Standardization of EDI doesn’t exist. I was in PropTech for a bit, and they try to standardize MLS data, that was really hard to achieve. I know there’s a lot of initiatives around standardizing the digital aspect of freight document management. I’m very supportive of that, but who knows how long that’s gonna take? It’s gonna be really challenging to get everyone to agree on a standardization. So I think what AI has done for us is it’s allowed us to be a lot more agile around those things. I honestly don’t think that it’s going to happen anytime soon. In our example, we’ve identified probably around 26 plus document formats so far. There may be 30, 40, 50 different types of document types. There’s no way I’m gonna go back to Coca-Cola or Tyson or Walmart and say, “Hey, I need you to standardize the way you handle this.” They’re gonna be like, “no, thank you.” And so I’ve gotta deal with these complexities.

Ivan Ramirez: That is a perfect use case of AI. And then on top of that, you’ve got the complexity within the document. “I want a BOL.” And by the way, there’s 10 different types of BOLs, but I am able to bill this customer when there’s two stamps, two signatures, and a little wrinkle on the corner and a little smudge of Coca-Cola on it. It’s literally that level of requirements that these folks are looking for. Unless you look at every single one of those documents, which is what we’ve been doing for the last 20 years, or you apply AI to it, you’re not gonna be able to scale that. You can scale it, but you have to add a bunch of people. And now all of a sudden you’ve gotta train everyone to say, “Hey, these are the 20, 30, 40 different types of documents. And by the way, these are the business rules for every single customer.” It’s just complexity. So this is a perfect, very strong area where AI is a must because try to do this with software, it gets really, really complex. You can have workflows with software, but you need the identification, the categorization, the annotation, and the execution of the business logic. And that all comes with the AI module.

Alan Pelz-Sharpe: Chip, a lot of people in supply chain used capture software back in the day and it didn’t work. One of the key reasons it didn’t work was because of this lack of standardization of formats. Do you still encounter that when you are talking to potential customers, that their perception of what document processing software is is that it hasn’t changed that much?

Chip VonBurg: First off, I have to say, I was in the industry when it was called capture software. And I think that actually is sort of part of the problem. Some people have lived through this and would really struggle with “well, how could you deal with this?” I think the lesson that we all have now is AI has opened people’s eyes a little bit to say, “well, wait a minute. I would’ve thought that would’ve been difficult before. But now I do believe that with machine learning and AI technologies, maybe this can be done.” And I do think you see a little bit more of an acceptance that this is a possibility. Now, that doesn’t mean that we still don’t have to go through proof of concepts with customers to show them that it actually works on their documents. But I do think folks are at least more open to it.

Chip VonBurg: I do want to plug just a couple of things Ivan said a moment ago. The only way that you can feasibly solve this stuff is with AI and ML. The only other option you have is to bring people on board and train those people. And so, if you think about it, you’re doing the same thing with a Hyperscience type system. With an IDP system, you’re gonna train the system how to read those documents. The difference is the system scales with you instead of you having to retrain the next set of people that you bring in. And so that’s really an important way to think about it. The other thing that he said too is the decision that’s associated with it. So often these documents, they’re not just read by the employees, but they’re interpreted by the employees. And so being able to build that decision and interpretation into the system, that’s where the magic really actually starts to happen.

Alan Pelz-Sharpe: In my experience, reducing the number of errors as early as possible in a process is often the thing that brings the most value to any kind of project. If you start cutting down the number of errors, then by default you’re not sending out the wrong things to the wrong people and spending a lot of money fixing them. Is that something that you think should be stressed more? That it’s not just about doing things cheaper and faster, it’s about doing things and not having to have a reverse supply chain essentially.

David Teeple: Yeah. And I want to go back to what Chip said just a minute ago about bringing in people and trying to train them. That’s one of the challenges we have right now is an extraordinarily difficult labor force that’s out there. They’re high expectations. They don’t want to do a lot of detailed work. And the turnover rate is extremely high. Living on tribal knowledge and people that have been there for years has been the way many organizations have survived and overcome the challenges of bad data. But these tools are extremely important not only for making it more efficient, but it really becomes an important recruiting tool when you talk about dealing with high turnover. Making these things much more user friendly, easier to use kind of shortens that learning curve. So when you bring people in, when you’re dealing with that turnover, AI becomes almost an imperative for organizations to be able to survive related to their most important asset, which is their people. The alternative is to live with this bad data, live with bad customer service, live with the costs of redoing things or paying for canceled orders. So AI becomes again something that is critical for organizations to be able to survive.

Alan Pelz-Sharpe: This is our last insight: for organizations that are still essentially reliant on paper-based systems, and I’d hazard that’s a majority of companies, what do you see that just as high risk in this day and age? How do you talk to customers and get them to realize that “we’ve always done it this way” has to change?

David Teeple: Well, a lot of it has to do with just basic discovery. One of the things that we focus on from a consulting perspective is listening to the customer, understanding their current environment, what keeps them up at night. Understanding their current pain points, digging into each of the key stakeholders’ worlds—how they view the company, how they view their jobs, how they view their teams—that really provides you with the ammunition to try to really build the business case for these types of systems. Extremely important for change management because understanding all the various stakeholders (IT, operations, logistics, merchant group) and understanding the world from the lens that they view it helps them understand, “Hey, I really need this.”

Alan Pelz-Sharpe: And Chip, you have worked with many organizations who want to make that transition from paper dependent to fully digital. What successfully worked for you? What is it that triggers a successful change?

Chip VonBurg: I think Dave said this a little earlier on: think about your data today in one bucket and your data moving forward in a different bucket. They’re both things that you eventually want to try to tackle, but think about day forward separately. And I think that that’ll change your mindset. One of the things that I’ve seen be very successful here is think about capturing these documents as far out on the edge as possible. Years not that many years ago, folks would put scanners in distribution centers and they would have drivers capturing documents at the end of their days. Think about the power of mobile now, and you see more and more companies doing this. If you can capture those documents right up front and actually make them into a digital asset for you right then and there, just think about how much more time you’re actually buying your business to actually work on ’em. Sometimes a day, sometimes even multiple days of additional time there. So as you make that transition, how do you move it from paper into digital sooner? And then you can start to think about all of the downstream things that you’re gonna do with that data.

Alan Pelz-Sharpe: Ivan, what to you is the most important thing for this shift to happen successfully? Is it change management? Is it focusing on KPIs?

Ivan Ramirez: I think what we have found is we started very much with the emphasis on understanding the root problem that we were fixing. It first came as like, “Hey, I want to make billing more efficient.” And then some other group of people said, “well, we need to change out our imaging system.” So two very different problems. So we implemented a product management function which was non-existent here, very much around interacting with the stakeholder to really understand what is the actual problem we’re trying to solve here. Set those objectives, define those key outcomes that we wanted to achieve off of the back of this. In our case, financial benefits was a big one. The faster we bill, the faster we get money in the door. It was also accuracy of billing. A customer being billed correctly is an unhappy customer.

Ivan Ramirez: And then once we did that, we got all of our business stakeholders to align that that was the problem that we were going to go after. And that we had two different things: the root of the problem was extracting the data from the documents and running the business logic. And then our technology team went off and said, “okay, now we’re gonna go off and solution this.” And we will come back and give you some options on how we’re thinking about solving this. And they were part of the proof of concept that we did with several companies. Now, what was really challenging is it was really hard for a non-technical person to understand how this was going to work because they’re looking at Chip and his team demoing some AI models. It’s like, “well, I just want my interface where I can go in and review my documents and hit approve and move it on to Bill.” They want the end-to-end user experience. So we invested a lot of time in UX/UI designing prototypes so that we could then show them, because at the end of the day, the work that Hyperscience does is the magic that happens behind the scenes. But we then still have to create that user experience for the user to actually say, “oh, wow, okay, so you’re telling me now I don’t have to go touch every single customer in every single document. But where do I go see though? ‘Cause I still wanna see the stuff that was actually auto billed and go and click on it and just verify that the system did it correctly.” So there is a way in the user interface where they can go in and look at things that have been billed. All of that process, what we found, it really eased their concern and their enthusiasm grew about the technology because it’s really hard for a non-technical individual to understand the power of Hyperscience. And so demonstrating it in a practical end-to-end flow is what makes this potential real exciting. So involving your stakeholders early on organically handles the whole change management thing because then they are going to be your biggest advocate for the technology.

Alan Pelz-Sharpe: I’m hearing all the time about the AI bubble is about to burst, and that could well be true, but not in this part, not in the back office, not in document processing. This is growing at pace, and so it should. It’s about time for change. So get away from the headlines about generic AI. This is happening and we’ve got the data to support it as well. Chip, any parting sort of tips or advice for anyone?

Chip VonBurg: I agree. Great conversation. If you haven’t looked at Deep Analysis document on this, take a look. There’s lots of other really good insights there. The only other thing that I would say is I think what we’re all trying to do is fix the manual and exception based processes that we all deal with. My advice would be look at the documents that fuel your business and I’d be willing to bet that you’re gonna find some of those manual exception based processes there and that can really make a big impact to your organization.

Alan Pelz-Sharpe: And Dave, maybe with your CSCMP hat on, what’s your sort of advice to the members of CSCMP if they aren’t currently exploring this kind of thing?

David Teeple: We talked earlier about many organizations are dependent on an ERP or a warehouse management system. Those are massive systems. And you’re really beholden to them kind of running your operation. But if you think about AI as you would with many types of automation projects these days, just ripping and replacing those systems is a very scary thought. And it’s not necessarily something that you have to do to kind of dip your toe into the AI space. So there are software packages out like Hyperscience that can enhance an ERP or a WMS to get you some of that focus on kind of the major reasons. It becomes a much bigger business case to kind of build out to replace your ERP or WMS, but there are systems that can enhance or magnify the power of your current systems adding that functionality in there. So think about looking at enhancing your systems versus doing a massive rip and replace.

Alan Pelz-Sharpe: Excellent. Thank you everybody. And just to pick up on Dave’s point, you absolutely should download a copy of the report because you’ll find out in there that actually there is a temptation by quite a lot of organizations to say, “let’s rip and replace,” and that’s a bit scary. So make sure you download your free copy of that if you’re interested in the technology market as a whole, where intelligent document processing (IDP) and business automation is going. If you are a supply chain professional and you are not a member of CSCMP, you absolutely have to check that out and hopefully get to meet with your peers. The networking value of these kind of organizations is just incredible. So you don’t have to do this alone. And obviously Hyperscience, you’ve got the technology here, and that’s what we’ve been talking about, and I think it’s great. And thank you so much to Hyperscience for underwriting and supporting this research. And finally, thank you, Ivan. Thank you, Chip. Thank you Dave. And thank you to everybody who’s joined us today.