Hello, everyone, and thank you for joining us today. This is number six in the series of webinars that Adisseo is putting together this year to provide useful information for our customers and the market. I'm Mike Shearing, the global formulation manager with Adisseo, and we are very excited today to have both Dr. Mike Van Amburg and Dr. Andrew LaPierre from Cornell University here with us today to talk about CNCPS v7. So to begin, we could ask our speakers to share just a little bit of background, history, and maybe an overview of the process that led to the development of CNCPS v7, Mike or Andrew? I'll take a quick lead on that. It was suggested to me that I just do a brief overview of CNCPS period. I think I'll start there. So thank you very much, Mike and Adisseo, for the invitation to the webinar. You know, this is a long, long term project. I think some of us forget when this thing actually came into being and had to go remind myself here of the dates. It was first described in a symposium conducted at Oklahoma State in 1980. In two papers, one the lead author was Dr. Peter Van Soest, the other one was Danny Fox, and they described the rumen submodel and then meeting the protein requirements for growing cattle. So, you know, it's been 40 some years since this thing kind of showed up on the scene. You know where we're at now. You know, it's gone through continuous iteration. We you know, and I have to credit Adisseo for some of this, Brian Sloan especially. Always mentioned him before the new NASEM, roughly around 2010 we were working on version six. And Brian and a few others were pushing on us to say, how can you improve the amino acid supply and requirements in this model? And I had suggested to him that we think we know where some of those offsets are at. And what we need to do to make this work better, have more accountability, account for more of the information. Helene LaPierre was part of that conversation. And in the end, Adisseo stepped up to help support a graduate student, Ryan Higgs. And that was really the initiation of what ended up being version seven. That was not our intent. Our intent was to clean up version six as much as we could, and we realized that the way version six was constructed would not allow us to fix the offsets. And then we wouldn't have enough information. Actually, what we knew about where amino acids were coming from, the information from Helen and Daniel would suggest, there was endogenous protein that needed to be accounted for and version six was not designed for that. So that is really how we got off on the idea of just reconstructing the entire model and updating it as we went. So that's kind of where we're at. That's taken you know, we're 30 years in and think we're just now getting to something that will work. 30 years on updating version six, the iterations of version six into version seven, correct? Yes. But much of version seven in my understanding has already been released with the updated feed library, various updates to 40 system and so forth that have already been rolled out into the current version. Is that correct? Yeah, that's an interesting observation. So the concepts that will be necessary to make version seven run most effectively. In the field have kind of been, you know, put into the industry. uNDF thing was big, but that was needed whether we were running 6.55 or 7 , you know what we had come to learn is that, Peter's idea of lignin types 2.4, as a fixed factor for indigestible NDF, just we realized it wasn't working. Maybe for some alfalfa hay, but not for anything else. So that was a big step even to improve the precision around 6.55. Some of the other concepts around amino acids. Yeah, that's kind of filtered back and forth, right? Some of that's due to Andrew's work. But, you know, in seven. The we will use a lot of that information more fully. So think it's going to give you much different answers in the end for reasons that we don't fully appreciate. I'll comment on that please, Mike. Oftentimes we get asked what are the additional inputs that we're going to need to supply as end users of the model for version 7 that will be different than version 6.55 And I think by happenstance, we've done a good job in describing some of the additional inputs, as you've mentioned, out in the field. So that's already being done in terms of properly describing fiber degradability characteristics and then some of the amino acid stuff as well. So the hope is that by the time version seven does enter into a commercial space, there really isn't anything too novel that a progressive nutritionist isn't already collecting data on. So that we as Mike Van Amburg has pointed out, we are just making better use of those inputs in a different way, the goal being that we are describing. The animal and the system a little bit better than 6.55 is currently describing. The outputs won't be drastically different in terms of the results, but we'll have better understanding of, we think mechanistically, how that that will work within the room and submodel of version seven and then post absorptive how that works in version seven as well. So if I follow it sounds like. For the most part, the progressive user, the one who is taking the time and effort to get good information in to the current model and properly interpreting the results coming out of the current model. For the most part, there won't be dramatic changes in the predictions they receive. But it's been my experience and I think others as well, that with previous updates, many diets would be relatively unchanged, but a few diets would change quite dramatically or substantially, maybe because they're have different forage types or profiles or on the high or low end, say on carbohydrate or protein levels and so forth. Which or what types of diets might be? Shifting more as they're evaluated with version seven versus versus version 6.5. Is that a fair question or an impossible question? I think it's a great question. I'll see if Andrew wants to go first. Sure. I mean, there's probably two parts to that. In terms of the diets. You know, we you make tweaks over time and it's a multi-tiered approach. So the way in which the system operates is that from a Cornell perspective, we have the core model itself, the base model, if you will, that we make updates to, and then we license that technology out to software companies such as NDS, Amts, etc. And honestly, it can be their prerogative to continue making further predictions on those outcomes that the CNCPS provides. So we can't speak to all the changes that happen. So if you were to open a software program and notice that an update is being provided, we can't necessarily speak to all of that because some of that is under the influence of these software technologies. However, some of that is based on the updates that we provide within the base CNCPS model, within version seven, the expectation is because we're relying heavily on on better predictions on the fiber side, as we titrate more forage in just by happenstance, we would presumably get different predictions on not only the energy side, but also MP particularly from bacterial proliferation, just because we get better understanding of how that degradability works within the rumen, there's certainly opportunity to refine that on other nutrients as well. We can talk about starch and non-structural carbohydrates. I think that is a opportunity to continue describing further. I would also mention that on the protein side, particularly within the rumen Submodel, there would be some changes in higher protein diets in terms of partitioning what is degraded within the rumen and what escapes rumen and degradability as well. So those higher protein diets, you may expect differences in what actually the post absorptive sees. our understanding is that we are trying to get lower protein diets, so we're trying to titrate a little bit more on the precision nutrient side of things. So making sure the rumen protein is is where it needs to be. And then after we get past the rumen and how can we precisely get at some of those amino acids and other nutrients as well. So, those are my kind of understandings with version seven and how they would be be different relative to 6.55 or any other updates. Mike if you care to comment on future, or other things. I'll just pick up on what you said and refine a couple of things. so will they change? Maybe, maybe not. I think to Andrew's point, Mike, one of the things that I see is we're going to. Again, what's old is new in some of this. You know, years ago, Mertens gave us some guidelines for NDF intake. And and you're wondering why the heck am I talking about Mertens and NDF intake when I talk about version seven. Well, they've spent an enormous amount of time describing NDF intake as a percent of body weight. And we all got excited about that. Right. And that was in the early 2000, mid 2000, late 2000 up into 2010, 2011, 2012. And then it all kind of went away. Right. And people would say, well, that's really hard to use. Well, it's hard to use because it was ambiguous. And one of the things that, you know, the way seven's going to work, and especially with the two pool NDF digestibility approach or three pool, I guess we can actualize that and say, hey, you know how much should be in the rumen of a cow at steady state? And being able to quantify that and say, hey, it's not 1.3% of body weight, right? So that's ambiguous because we don't do body weights and so we don't know what 1.3% means. So the question is, well, should that cow have 8200g of NDF in the rumen and at any one time. And is that and then the question is if that's what she should have in the current diet situation, if you have good chemistry, is it going to be uNDF first limiting or total NDF first limiting on how that rumen fills and empties? Because then all of a sudden you can make different decisions if you have different forages and that might explain why you can't get the intakes you want, why you can't have the milk components you think you should have, or maybe why you've got some acidosis that you don't understand. Because instead of being at 8200g, you're actually running about 7700g. And that doesn't sound like a big difference, but it could be in some of these cows with time budgets that are a little bit off that that 500 gram difference in NDF May actually cause some acidosis, some acidotic conditions throughout the day as the rumen fills and empties. And we always thought that the cows, if they're that low, they would fill, you know, fill up that space. And what we're finding, you know, based on a bunch of the literature that we've been able to kind of run through the modeling from rumen emptying studies, they don't do that, right? They don't feel the way we think they should for various reasons. So think that'll be helpful. And that would cause you to say, okay, maybe I do need to make some modifications here. That's very interesting. And Mike, you bring up the idea of troubleshooting diets and want to ask about that. But first, I want to circle back to Andrew's comments, regarding potential changes to the nutritionist who you CNCPS might see as they graduate to version seven. And I think it's fair to say that certainly at least some would be a bit concerned that if their diets are not right down the middle necessarily with the average forage and average protein and so forth, that things will change and they won't know how to react and think. The concern in changing is that we have to do one or the other. We can't consider both and move from one system to another carefully or over time, a bit over a period of weeks or months, as opposed to flipping a switch and just going with a whole new set of numbers that are totally different and totally foreign. And will it be possible, do you think, depending on what the software companies who provide the commercial platforms to look at both sets of numbers at the same time without building two rations in two different programs? Oh, absolutely. you know, it's always been our intention that the functionality and opportunity to continue using version 6.55. We will support that even after it has version seven has been launched. Recognizing that there is a transitory period here where, user adaptability may take some time. So at least within our group, we still will welcome the use of version 6.55, knowing that it will take some time to switch over. Having had conversations with a few of the license holders. I believe that's their intention as well, is to have sets of numbers not probably not comprehensive, but sets of important outputs where you are able to view both the version six technology next to the version seven technology. As Mike Vandenberg has pointed out, we get, we get good numbers predictions coming out of version 655. It's just the intent with the modeling practice is that we're getting similar numbers but for more mechanistic or descriptive reasons so that when we run into troubleshooting issues, we're able to have a better tool set to try and understand where we think the opportunity lies to rectify the situation. So pointedly. Yes, that's the intent. I can't speak for everybody, but the hope is that for at least a period of time. Version six and version seven numbers will be available to a nutritionist or end user to compare the two. I don't think they'll need to create two separate rations. It's just that they would be able to interpret those numbers in in two different systems. I think that'll be very reassuring to the market to have that ability to ease into using a newer, different and hopefully better set of numbers to evaluate their diets. the troubleshooting component that to me sounds like it's going to be much more robust and more useful once we understand what all the inputs and outputs mean and how they're interrelated and interconnected. for the practicing nutritionist, what would you say are the, the biggest areas where they may be able to improve diets, Either improve diets or fix problems, troubleshoot? Is it fiber? Is it amino acids? Is it is it carbohydrates? Is something else that that you know, what should really entice them to want to jump into using version seven and taking advantage of it. Want to go first there. Yeah, go ahead, Mike. Well, I think there's several ways to look at that, Mike. And, you know, I'll come back to the troubleshooting. I think one of the big strengths of it is going to be the ability to troubleshoot some things that we don't understand, right. And ask some questions that we can't currently answer effectively. You know, we could it be the amino acid side? The answer is absolutely, because you're going to get all ten of essential amino acids and you're going to get their expected or predicted supply levels relative to requirements. You know, and some of that was in Andrew's own studies for his PhD, where we've run them back through version seven. And we were looking at Methionine and Lysine and come to find out that Isoleucine was predicted to be first limiting. Right. Who saw that coming? Right there's information like that. Again, I think the thing that to me so it could be amino acids that are intriguing to you. It could be that you want to figure out how to reduce your diet costs as much as possible. And if that's on the nitrogen side, you know, one of the strengths of seven is that it basically has an isolated rumen. Where we estimate ruminal requirements completely independently of the rest of the cow. In a way that as far as we know, no model has been able to do that. We've got full recycling, we've got the endogenous, we've got all the exchanges with the microbes, the protozoa, you know, getting into the weeds a little bit, that gives us a really robust model to predict, you know, ruminal nitrogen requirements to make sure that we never run out, but also disassociate it from this concept of crude protein. Right? It's all on a nitrogen basis. And then we can focus on the amino acids that are then supplied from the rumen, can go to meet the mammary requirements or any other productive function. So you know, I think that's going to help a lot of nutritionists, you know, in some markets. Right? If we were in Northern Europe right now, they'd be thrilled with that because those guys are being really challenged on how to reduce their nitrogen supply and reduce nitrogen excretion. And I think 7 is going to be much more powerful in being able to do that effectively. So that's that's one example, right? I could go on and I know Andrew's got some other ideas, so. Here's what I'll do. Visual learners out there. So Mike's right. There's a few things here that I think relative to the functionality of version six. I think there's some opportunity in what we think are some tools for helping to troubleshoot on some of these issues that are observed out in the field. recognizing that Cornell, Ithaca is in the Northeast. Traditionally, we think about higher foraged diets, but there are areas of the world where the CNCPS is used where that's not traditionally the case. There's some opportunity to use some tools such as what you see on this this slide, to try and understand maybe where the opportunity lies. So one of the things that version seven we're trying to develop here and of course it's not perfect, there's some opportunity for some improvement. But we try to describe a full gastrointestinal tract degradation here. So you can see we've got a higher forage diet. This is partitioned out for just the forage ingredients. But there's estimations of degradation, pool sizes and passage of the carbohydrate fractions that the CNCPS recognizes. So you have sugar starch, soluble fiber and NDF is broken out into its three pool system. Where the where we think the opportunity lies is, is that last row there where we get apparent total tract digestibility of those fractions and for brevity's sake we do the same on the concentrates or we can do it from a dietary perspective. But the opportunity here is Understanding that you have the right chemistry described here. There may be some opportunity to look at maximizing degradation of your diet by trying to optimize for the right nutrients, the right feed supply to get or maximize that TTNDF or NDFd or some other fractions. Now, this is one of the tools that we can use on the carbohydrate side. We're hopeful that it will be available to users through the software platforms. The other thing to keep in mind, too, is we do look at nitrogen as well. And this is very much in the weeds, but we have ruminal nitrogen transactions here. They're kind of broken out, by the way, in which the model describes them. We don't anticipate all nutritionists to use them, but think it's very insightful to try and, you know, look at your diets as they change over time. And what happens with the model predicts in terms of degradation, flows, absorption, those types of things. So you get ruminal nitrogen transactions here and then you also get post rumen nitrogen transactions as well. Right. And it's rather comprehensive list here. We don't expect everybody to make use of them all. But what I think is really insightful is that we have some of these outputs available. If nutritionists have the time and want to want to look through them. Could you go back one slide, Andrew real quick, want to insert just one comment to this. So a question, here's a practical question that I can't answer in 6.55 But I get asked all the time, Mike, why when I add sugars to my diet, do I not see big increases in microbial protein yield or bacterial yield? And my standard answer now is because we don't have the protozoa in 6.55 and they're the most sugar responsive pool of microorganisms in the rumen. So as you can see here in this slide, you know, down in the microbial section, you've got protozoal passage at 41g. I would expect this is back to more information as we put sugars into diets. Now with version seven, you're going to see changes in protozoal growth. And so now you're going to see the reward for putting the sugar in the diet. It's not just the energy. Might be some change in bacteria, but you're really probably going to see it in the protozoa. And that's something that we haven't been able to see in six because we just don't have that pool. Thanks. And then Mike Shearing the last thing I'll point on and Mike pointed this out as well is post absorptively we're looking at obviously milk production. So trying to understand where our amino acid balance is lie relative to what we think the metabolic demand is in the mammary as well as comprehensively throughout the cow. We do have apparent efficiencies of use. You'll see that in the fourth column here. So EoU is efficiency of use. That is our apparent efficiency of use based on the diet that has been described in version seven. And then some of the targets that we provide in that last column there. This is something, you know, isn't available necessarily in the version six platform. We are using a static efficiency of use. We know that changes over time. NASEM has described that in version seven will move towards a variable efficiency of use as well when it comes to amino acid balancing and on the MP side as well, because we don't want to neglect the non-essentials and their supplementation in a diet as well. So those are some of the tools. Think big picture that we are expecting nutritionists to use. There will be some other things that will be available in terms of fill and flux for NDF within the rumen. But for now, these are the these are the things that we are looking at. One of the other things that you see there, and we're actually using this information in a prospective way and I'll lead into that in a minute. But if you just look at the allowable milk. Right, and think this is to your question, Mike, about, you know, what are the users going to see and what might attract them to using this? You know, we get focused on lysine and methionine. So if we go down that allowable milk column you see some negatives in the balance. But then you see the, you know, the leucine I'm sorry, the lysine is at 40.4kg and the is at 41.3. So generally right now, in 6.55 or in most softwares, we'd say, hey, I'm going to add a few grams of lysine to make sure that it's not as limiting. But we would never look at valine, which is actually first limiting in this diet. Right. So all of a sudden what that would suggest is we can do whatever we want to lysine and methionine and we might see some component shifts because of their role in fat synthesis and protein synthesis. But we may their response to those additions may be muted by the fact that Valine was actually first limiting in that's where the power of this new version I think is going to come in and the ability to refine some of that. It does create another challenge. And that it's like, huh, wonder what feed I can feed that actually provides more. Valine. That's very interesting Mike and this is certainly new, different and better information than has been available in the past. We've got amino acid balance levels estimated down to the 10th of a gram and we're feeding cows with payloaders and scooping up hundreds or thousands of pounds of feed and dumping it in as quick as we can to get the job done. So my question is, we need good information going in to get good information out. What what would you say in terms of feed testing, feed chemistry? What information on that part of the diet is a minimum and what would be really optimal in terms of testing our feeds? Do we need fatty acid profiles? For example? Do we need amino acid profiles on forages or high protein supplements, for example, or can we use book values and just get the, the more basic numbers like 80 IP and NDIP and so forth? In other words, to make this work? What's the minimum that's really needed? and what would be the best case scenario to, to really get the information going in as accurate as possible to get the best results coming out. So I'll take a quick stab at that and then Andrew can can jump in. Just from a couple points. And, you know, because there's some labs right now starting to supply our analysis for amino acids. Right. And I appreciate the fact that we're going to have that information available to us. It's a bit troubling to me because I'm a little afraid we're going to get into the wild, wild West because there's a bunch of information missing when I see these values coming in but my quick take home on the amino acid thing is we won't for certain ingredients. It would probably be very useful to have some amino acid profiles. I think it'd be more important to have digestibility. All right. Because I think the profile of blood or the profile of fish or the profile of soybean or whatever isn't going to be remarkably different. But the digestibility could sure be remarkably different. So I think that's where we're probably going to get hung up on some of that. More than we are just on the profile. And the reason I take that stand is, you know, when we look at the work that Chuck Schwab has done, that, you know, if we look at anybody who's done amino acid work in the world, for the most part, we have pretty interesting data. And think about what we do here in house. As long as we have digestibility data, we can use the feed library for the most part to formulate diets and get nice responses. Right, which would say as long as you have described the feed appropriately and you have the digestibility characterized that the amino acid profile and the microbial profiles kind of follow along. Right. There's not these big outliers there where alfalfa or grass or corn silage or some other ingredients look so remarkably different that it doesn't look like anything in the feed library. Doesn't mean it's perfect to Andrew's point, but it gives me more satisfaction right than I used to have, because when I listened to people talk, well, I see these numbers, they're changing. This lab gives me these different numbers. Well, our own empirical data are formulating these diets would say, well, that's maybe true, but I'm not sure that it covers more of the variation than just understanding Digestibility. Which we tend to ignore. So that's kind of where I come in on something like that. That part of the information. Right. We could go on. There's a whole dissertation in this concept. Andrew. Do you want to? It's a very loaded topic. There's a lot of opportunity there. To Mike's point, I think. In addition to understanding, you know, let's take I think, think Feed Labs have been doing a good job in recognizing the nutrients that create what we call the mass balance for a given feed. So making sure that when all those mass balance values, protein, carbohydrates, fat, ash all add up to 100%, you know, you've got that feed properly described. I think we've done a good job with that. That is really the base, the minimum in understanding how these systems and I say systems abstractly from the CNCPS because it doesn't matter which model you're using, that's got to be the minimum requirement there. I think Digestibility is certainly important that captures a lot of the variation. But to be frank, I think it's kind of a chicken or an egg kind of scenario. you know, there's technology that's becoming more mainstream that allows us to procure this data and there's opportunity to use it so that that amino acid via NAR does allow us a little bit more, opportunity to understand what the variability of our feed ingredients are because quite honestly, we're never going to get around. feed ingredients being variable. The better approach is to understand what that variability is over time. And so you do need tools that are quick, that are responsive. Wet chemistry, amino acids, if you've ever done them, they're painstakingly long, and by the time you get numbers, you've already gone far beyond the feed that you were sampling in the first place. So there is opportunity with NAR to make things quicker the turnaround time more quick. But there really is to Mike Van Amburgh's point opportunity and making sure that the wet chemistry behind that is appropriate and those those equations are robust enough to handle multiple feeds that may be variable. You may get a truckload of soy that shows up that is out in left field relative to what book values may be. Are NAR equations going to be able to appropriately handle that? Because that's really where the troubleshooting comes in, is when we get something that doesn't fit the script. Our models equipped to handle that. And will our nutritionists understand that, you know, that variation may not be because the model is not predicting correctly, it may just be because the values of the feed chemistry side are not typical I guess. I think that is a very good explanation that you have. Stability, obviously is very important for all the major classes of nutrients, including fiber. Andrew, you had mentioned, I believe, fill and flux for NDF and that if I understood what was said, there's some information now in version seven, but there's more to be developed or that part of the model needs. Some extra work or additional work, is that correct? Or could you touch on the fill flux on the fiber ruminal part of things? where it might go. You know, basically the fill and flux kind of comes the ability to describe flux really comes from the fact that we are version seven would be iterating over a time series, a virtual day that is described with a time step that we have set that is not computationally excessive, but does allow for subtle description of changes in nutrient flows as the model has been described. but to be honest with you, I think where the opportunity lies is more on the behavioral side of things. So there are predictions out there. Nassim has them in terms of describing the animal and the animal and the feed inputs to get what the daily intake is of dry matter, which is great. I think that's a great step forward. I think the end goal is to try and describe and appropriate that intake throughout a day. And that's where the opportunity comes in to have some behavioral models. we do take a stab at that. There certainly is opportunity to refine and improve it. There always is. But I do think the saving grace here is that we have technologies that are coming online that do allow us to describe intake patterns of animals based on certain parameters, overcrowding, feed bunk space, times milked, all these different things that as the data becomes more available, we can procure it, we can parameterize equations and the hope is that we have some sort of fill and flux model on NDF that describes the intake of that animal over a day. because I think that's where the opportunity lies. As Mike has pointed out, when time budgets aren't quite right, perhaps there's opportunity to show the data, show the predictions where you could go to a producer and say, if we reduced overcrowding by 10% or 15%, we would expect intake to change by X percentage. and further, we would expect the fill and flux of NDF to change accordingly. really our mindset is that NDF, fiber intake would drive some of that intake. And if we appropriately describe the behavior of the animal, there may be some opportunity to refine what we think the fill and flux of fiber is and thereby understand what our dry matter intake equation is over a day. So it's a little bit cyclical. You've got to start somewhere. But I do think there is opportunity for refinement if we start on the behavioral side of things. So at this point, would you say that Just beginning to model the management. But the hope is that over time we can incorporate more of these external things into the system to improve and refine the results so we could develop with it. Is that fair to say? I mean, this system, when it comes to nutritional management is very comprehensive. It's more than just the feed supply to the animal. It's the management strategies, it's the financial considerations, it's the frequency in which we procure that data. I mean, it's very comprehensive in terms of making appropriate predictions for what the cow actually does at the end of the day. And so we have spent a lot of time on the nutritional side and will continue to do so. But it's, I think, appropriate to also start looking at things a little bit more comprehensively. And one area I think that could stand to be improved is understanding the behavior of those cows in a way in which we can quantify it. And there are groups, including ourselves, that are doing so, and starting to do so. And the hope is we would have inputs to allow producer to supply that information or nutritionist to supply that information. So what you're saying, Andrew, is the work is not done. The version seven isn't the end, that there's more that could or should be done. Yeah, absolutely. I mean, the whole concept of modeling is that it's never a complete picture. You'd never want it to be a complete picture. But as we have internal conversations about where we think the next opportunity lies, I think that's one of those opportunities. It's good job security saying that we're never, never really done right. So, I think it does a great job in what it's supposed to be doing, but there's more work. Where then do you see the CNCPS project going. Do you think we'll we'll have updates 7.1 or a 7.2. Do you see a more dramatic, longer term, a bigger update of version eight, for example, ten years down the road or after Mike retires or what big picture? Just without holding anybody to anything. what are your thoughts on where it might go at this time? I'll give you my answer and then Andrew can give you his. It probably won't be too remarkably different. I don't see a version eight. Mike. Don't think there's a need for a version eight at this point unless we learn something completely. Just something that we just don't understand. Or we learn to understand. Something that we can't model in the current format but don't see that happening. This structure is really good. And I don't see any modification of this structure. They might be different software might be different ways of putting it into, you know, from C Prime to R or whatever for our own internal use. But Now. But do I see updates? Absolutely. You know, I do see a lot of updates, potential, right. Things that Andrew said. Right. Just getting more robust intake models. More management models. Babino and I said was talking to Andrew, Babino and Ispent a bunch of time together last Friday. We have some ideas about how cows decide to synthesize, you know, de novo fatty acids versus amino acids based on some of the data that we currently have. You know, maybe that allows us to make better predictions about how we can influence milk composition. Right? Directly and say, to the user, hey, if you do this and this and this with these three amino acids and a little bit on the fatty acid side, you can get more de novos independent of any lactose or or milk protein. And if you do this, then you can get the milk protein. All right. So so we think that we're at a point with this information that we have as we refine this using version seven concepts that we might be able to tell you guys, hey, we know how to get, We can tell you which direction this is going to go if you do these following things. That's kind of cool. it would just emphasize that I don't see numerical. Improvements on versioning mean think the versioning is there just so somebody one of the end users understands that I'm using this technology and so the equation should behave like X, Y and Z. I think there's lots of work that can be done where you spend some time daydreaming about what an ideal model would be able to do. I think on the nutrition side, there's certainly opportunity to, at least within the CNCPS structure, to improve upon fatty acid nutrition and understanding. To Mike's point, how those fatty acids behave within the rumen are partitioned post absorptive and what sort of functionality they have within the local metabolism of the mammary gland. Starch is a big one too we spend a lot of time on fiber. We've feel like we've got that pretty well licked with the occasional edge case that makes us scratch our heads. But starch is very elusive and I know we've had a lot of conversations internally. Feed Labs have also engaged in that conversation. How do we want to start thinking about some of those starch degradation rates, developing some sort of commercial system that provides robust results but is also quick and cost effective? You know, those are the that's the trifecta there, if you will. And that can be rather difficult when it comes to starch. And, you know, it's think opportunity lies where we want to see it as an industry. I think one of the things that is. Becoming more and more apparent as technology improves is the feedback we get from the model itself. And so engaging with further engagement with feed companies, with feed labs, with end users, with the license holders, understanding what's where the capabilities are really good within version six or version seven and where there's opportunity for improvement. Because at the end of the day, there's a lot of livelihoods at stake when it comes to using these tools. Doesn't matter if it's CNCPS, NASEM, INRA, or Norfolk, people build diets for cows and make a livelihood off that. So we want to make sure that those groups are heard to kind of steer where the next opportunities exist. I'll just add one more comment there. The other thing, and we haven't talked about the environmental side of this, other than talking about nitrogen. You know, we've got some compounds, some feed additives coming that, you know, will alter rumen fermentation. We're going to reduce methane. Well, if we reduce methane, we're essentially saying we're going to retain carbon and hydrogen in the rumen in a form that no model currently predicts. I don't care who you are. And it also violates our current energy systems because the energy systems are developed on the idea that a certain amount of methane gas is lost. Right. That's part of our you know, DE to ME calculation. We all learned that in sophomore nutrition class. So what that leads to, Mike, and this is, I think where, you know, there will be incremental updates is let's say a product like Bovaer or 3NOP comes on the market, you know, but how are you going to formulate around that? Well, if it actually does reduce methane by 30%. What form is that carbon retained in? And then what use is it, if it's retained as acetate or butyrate, we shouldn't expect more milk volume, but we would have more energy for milk components. But if the diet doesn't change and if the microbes aren't killed in the process, which they're not, and that then we've got more energy, which means we're now amino acid limiting. But we don't have a way of we don't have a robust model for making those calculations. All right. So there's a bunch of subtle things there where all of a sudden we're going to we're also going to be forced into some more mechanistic modeling with some of these new technologies. To Andrew's point, the feedback from the industry is going to say, hey, we need to understand, you know, how do we model around, how do we formulate around some of these new ingredients that mitigate methane? And that's not actually being done right now effectively. And that certainly is going to be a bigger part of what we're all dealing with, sustainability and environmental impact. We are running up against our time limit, I think. So what I'd like to do is ask each of you, Mike and Andrew, to just share some final thoughts, some comments or insights that maybe we haven't discussed or covered, but that you think are important for the audience to hear. And then we'll we'll wrap up and conclude this webinar discussion. And anything we didn't cover or anything that noteworthy that we want to make sure the audience hears. Job security for Andrew. More funding for Mike's future research. What else do we. Did I miss something there? Yeah, you know, I do. The models used enough. Now I'll say this my way. You know, a couple things. Andrew alluded to the fatty acid side. You know, I will make an argument that cows have a fatty acid requirement similar to an amino acid requirement. But we've been describing fat mostly as an energy supply, not as a nutrient. So I think there's some, I think that's not low hanging fruit, but I think it's an opportunity for us to figure out how to make the cow more energetically efficient. So there's an opportunity. I do see this thing has been around a long time now. We're we would like to think we're pretty transparent about it. Um, and to Andrew's point, we do try to take, we take all the feedback from the industry. Sometimes we don't have the capacity to respond to it. But, you know, we do see this as a partnership. So there's, you know, we would like to think in whatever format the model will evolve over time to help the industry. Right. That's partly why we're in this. It's not this isn't a big moneymaker by any stretch of the imagination, right? I drive a truck with 230,000 miles on it. So, the point, though, is that intellectually it's a lot of fun for as a training tool for new students. It's brilliant as a tool in the field, it's been great. I think we just want to continue to build on that, right, and do what we can to help. And, you know, and maybe some day it becomes outdated. Who knows? But I don't know. Maybe someday it'll be in a cloud and the computer will be making the majority of the decisions and we'll be doing the refinement. You know, that's you know, we've talked about things like that to. Andrew, any final thoughts or comments? I mean. I'll just echo some of what Mike has said. I think when it comes to version seven, we're going to try and be as transparent and comprehensive when it comes to how the system behaves and what our expectations are when it comes to particular dietary parameters. But I'm going to imagine that even with the initial launch of version seven, there will be some edge cases that may have flown under the radar where there's some opportunity for further improvement that we haven't necessarily thought about because we do spend a lot of time thinking about this. And oftentimes when you take a step back, you kind of slap your head and say, well, of course that wasn't going to work that way. So there's opportunity there really on the feedback side. And I'm a big proponent of that, and I think it's going to be a collective effort regardless of the model that you're using to advance the industry forward. We all have some big jobs on our hands in terms of environmental sustainability and reducing methane. And that's all well and good. But in the meantime, we also have to make sure that these cows are fed healthy diets that improve productive efficiency here. And that's going to require a system that is circular, where we think we provide the best tools and those in the industry make use of them. And when they break, we are there to be able to help fix them when that time comes. So, education, that's a big part. And we really want to try and help where we can when it comes to that. Whether you're a new nutritionist or a seasoned one, there's always opportunity to improve and we want to be part of that. Well, very good. On behalf of Adisseo I'd like to thank both Dr. Van Amburgh and Dr. LaPierre for their time and their effort today. Their insights and comments and information are very helpful and very much appreciated, not only by us, certainly, but by our audience as well. So thank you and have a good day. Thanks, Mike.