The Moos Room™

 Welcome to the Moos Room. It's actually finally raining in Western Minnesota for once. We haven't had rain, I don't know, maybe a few shots here and there, but the last two months it hasn't rained much, and everything is burning up. I was actually in Ohio last week, and the grass was green. Everything was nice and green.

I talked to somebody there, and it was like the grass is summer grass or spring grass. It's so lush and green. Not here in Minnesota. It's dry. It's brown. We're chopping corn silage. It's a little bit dry. Some of the fields are spot on for moisture. Some are starting to get dry. Some farmers, I'm afraid sometimes the corn silage is just gonna be too dry.

So it's kind of spotty all over, but we've been really dry here in the Midwest, so it's kinda crazy. So really it's nice to have the rain. That'll help at least for pastures going into the next year. I'm not sure that it's gonna do anything this year, but hopefully they'll recover for next year Anyways, I was talking to some extension people and we've had some meetings and we're trying to figure out this whole AI thing and how you can use it on your farm, either dairy farm, beef farm, whatever kind of farm you have.

So it's kind of interesting to look. So we were at an extension livestock meeting about three weeks ago, uh, where all of us extension livestock educators get together and we heard a presentation by a faculty member at Colorado State. Now this was about horses. I'm not gonna talk a lot about it, but it was interesting that they asked AI what sort of knowledge they have about equine extension and lots of different things about Issues related to equine.

And it was interesting. Basically, it showed that AI tools have potential as resources, but they really fall short of the expertise and knowledge that extension specialists can offer. Some of them were good. They gave vague responses, you name it. So it's kind of interesting, and it's got a lot of us thinking in the extension world, how can we use AI to help farmers, and are we doing any of it ourselves?

And I admit that I'm using it quite a bit, at least to summarize data and summarize data from our dairy farm. And maybe I wanted to talk a little bit about how some of the easy things that you can do that might be able to summarize your data. So I'll talk about summarizing production data, so basically bulk tank data.

You can download that from your milk collar word DFA. So I downloaded the DFA. I also downloaded our organic valley data from our farms, and we can kind of see what it is and if there's trends that we have. And I also summarized some DHI from our last test, and I have milking parlor data as well, and we'll maybe talk about some of these all in brief and, and how I summarized them.

So basically, I summarized our conventional dairy, downloaded bulk tank pickups from twenty fifteen to twenty twenty-six. So I have ten years worth of data on our conventional herd, downloaded it all from the web into a spreadsheet, and I went to ChatGPT and I said, "You know, summarize this conventional pickups to me.

What, what trends can I see? Anything that I'm looking for." Well, it was kind of interesting. If you look back across history, it tells me my average pickup was about eighty-four hundred pounds of milk. Obviously, it ranged quite a bit from a thousand to seventeen thousand. Our average butterfat, four point oh six.

Average protein, three point three. Average somatic cell count, two hundred and thirty-five thousand. MUN, nine point nine. So it really, it kind of averaged all those across. Now that we could probably figure out ourselves. We might not need AI to do that. But it was interesting because it found a hundred and five pickups across basically the ten years had somatic cell count that were four hundred thousand or more.

So there's three thousand one hundred and eighty-five pickups and a hundred of those, so about three percent had SCC over four hundred thousand. nothing exceeded 750,000. So it's kind of interesting to see that across time. It can actually look at year-to-date and production statistics. So it tells us that We're up about 8% in milk production from this time last year.

So we've had more milk this year in our conventional herd. Our average pickup is up about three and a half percent. Butterfat, we're down a little bit this year, so 4.3 last year, 4.26 this year. Protein has increased, so we were 3.36, now we're a 347 protein. So protein has gone up. Somatic cell count maybe has inched up a little bit this year.

Now, last year it was two hundred and sixteen thousand, this year two hundred and thirty-seven thousand. So it's gone up. Our MUNs have gone down from thirteen to 9.9. So it's really good. You can also summarize-- It kind of gives you some recent things to watch for. So it gave me some somatic cell count things to watch for.

Our-- Within the last month, it showed somatic cell count was a little bit high. We're maybe about three hundred and fifty thousand in July of this year. Last year, three hundred and twenty-eight. So we're up a little bit during this dry weather. And our com- production components were the same, kind of this year is what it was last year.

So you can have it make graphs and, and look at all that trends if you're a visual person and want to see graphs. I, I told it, I said, "Make a graph," and it graphed components across the last two years, and it's kind of interesting to see, uh, what's happening. So organic dairy, what did it tell us? So we're up 10% milk production this time from last year, which is good.

There's a increased demand for organic production. Our average pickup is up about 10%. Butterfat, we're down. So as production goes up, fat goes down. We're running a 4.2% fat now. Last year at this time, 4.5% fat. But our protein has increased 3.38 to a 343 protein. So our protein has gone up, and somatic cell count has gone down.

We're averaging about two hundred and eighty-four thousand in our organic herd. Our pickup's over four hundred thousand, and yes, we do have some of those, especially when it gets hot and muddy. We're only averaging about 10% this year, so we're down quite a bit, and our MUNs are down quite a bit as well. So it's a good thing.

We can kind of look at some of these things and, you know, talk through these with our nutritionists or with our management to see, uh, what's actually happening

So you can s- you know, summarize these however many years you want. You know, I summarized them from 2023, and it gives you averages for both of our herds for fat and protein and somatic cell count. So you can kind of look at the trends. We've been trending down actually quite a bit in somatic cell count in both herds.

If we look at 2023, our organic cell count average was 361, and we're averaging about a 280 now. So really good. So these are some things that, you know, it's hard to glean when you're looking at just kind of data on a spreadsheet, but I've used these AI models to really show what's going on. And it, it gives you some main findings.

So it says protein has improved with both of our herds. It says organic has a higher butterfat. The conventional has consistently lower somatic cells count, right? We, we know that. It tells us that Our somatic cell count was low. It does give us, you know, for above four hundred thousand, it kinda tells us where the break points are.

Uh, but it does tell us that our organic somatic cell count has substantially improved over the long haul. So it's kind of interesting to see that, and you get these trends. So you can kind of go back and look and see and, you know, what, what did we change in management and how that might relate to everything going on.

So that's our bulk tank information. You know, so I basically used a couple easy prompts and said, "Summarize this and make a graph," and it kind of pulls all this stuff together. So it's really easy, takes a few minutes to do, and you can look at it and move on with your day and kind of get some of these basics so you can see what's happening.

Now, I also took our DHI summary and, and summarized it as well and asked AI to summarize our August 10th DHI summary, and it basically tells us the herd is a seasonal dry period, so we have a lot of dry cows now. We're calving a lot here in the fall, but it also shows that somatic cell count is the clearest immediate concern, particularly in string one, which is our organic herd, and with a relatively small group of high somatic cell count cows.

So it summarizes the total number of cows, so we have about two hundred and thirty milking cows, which was down a little bit from last month, down about 50, 60 cows, 'cause we're drying off a lot of cows. It tells us our milk per cow was, uh, 48 pounds, which was up from 42 pounds the last month. So it's kind of interesting to see that.

Our component's about a 4% fat and a 3.3 protein. So it pulls together a lot of different production stuff. It also looks at udder health, so it tells us that Some of the cows are high for somatic cell. It says we have fourteen percent new infections, thirty percent chronic infections, which we know we have some cows that are kind of chronic, and we're trying to deal with that, and that's always a challenge.

And it really pulls out some of these cows that are really high somatic cell count. They could have had mastitis during their time, but it says that our somatic cell count is really concentrated. So it tells us that there's five cows that generated about thirty percent of the estimated cow-level somatic cell count load while only producing three percent of the test day milk.

So it tells us who these cows are. So we have these five troublemaker cows that we can go look at and decide what to do with. And it's hard. You might have to s- you would sift through these reports a lot, you know, in, in our DHI summaries, and I've summarized it, asked AI to summarize it, and it's pulled out these cows.

Now, I could find all of these. I could probably find this stuff myself, but it's gonna take me a long time. So it's really about trying to use AI to save time and effort and find these cows. Now, you probably wanna go back to the DHI sheets and double-check to make sure that it's right. But You can look at a lot of these things.

It gives you lots of information about reproduction. It tells us our first service conception rate is 42%. Our services per conception is 2.1. It tells us there's 20 cows that were flagged for more than 250 days open And across the previous year, it says we had a twenty-four percent turnover rate, and the leading reasons were repro, mastitis, other reasons, death, and feet and leg issues.

And it really, I'm interested because it really talks about cows that we should review and really look at it. So it's combined cows that have low production and high somatic cell count, and we should probably be looking at those. So it pulled out one cow that had eighteen pounds, two million cell count, and it said she aborted to start her lactation.

So I think that's a cow that we should probably decide to call into the future. So really, I think we can address some of these issues. It tells us that the priorities where we should focus are really looking at the small group of cows that are driving our somatic cell count. We should prepare for a really fall freshening period and review the organic herd's extended dry period and lower performance.

Well, they're gonna be a little bit lower. We know this, uh, as we go across time Now probably the most fascinating thing that I have found is our parlor information. So we have an AfiMilk parlor, so we, uh, it records, uh, all the daily production stuff, fat, protein, lactose, uh, conductivity, you name it. But it also reports parlor performance, so based on, you know, how many cows per hour, milk per hour, uh, stall per hour.

It looks at milking curves, so it looks at flow rates of cows, how milking speed, irregular takeoffs, bimodal milking. It looks at milking irregularly, so multiple attaches of units, unnecessary attachments, irregular takeoffs, kickoffs, cows that have long milking times. It also looks at our load attachment, so how fast we can load a parlor, and our load efficiency, so times between ID and first attachment and lots of different things.

And so there's about three or... Well, there's about five different spreadsheets, and I also combine it with milk weights to actually put into AI. There's no way I could summarize this. It would take me a long time. It'd take me hours to really summarize this stuff myself in spreadsheets and sort things and try and really figure out what's going on.

So I put them into AI, and it takes about four or five minutes for it to analyze all of this data and really look at it. So what does it really tell us? So I basically asked it to summarize it and give me a weekly parlor KPI report or key performance indicators. And so I tried to summarize, uh, this on a weekly basis, and so it gives you an executive read.

So I've been doing this for, since early July, so about two months, 'cause I'm curious. I want to summarize this data and really use it from a management perspective and how we can improve our herd. It, it tells me, compared with the week before, our cow flow fell sixteen percent in the morning milking and fourteen percent in the evening milking.

Milk per cow increased, but we had longer milking times, more low-flow times, and more irregular takeoffs, which reduced parlor throughput. So a lot of this we can figure out what's going on as well. It tells us our milking time, so some of these are a little crazy, and I don't know, sometimes we have to go back and look about what's going on.

If we look at morning milking, one time it took seven hours and forty minutes, and the next time it took five hours and twenty minutes. So I don't know what was happening for those two hours, but it's interesting to look at some of those even in the milking, evening milking. One time it took four hours, the next time it took five hours.

So I don't know. It's one of those things we have to revisit with employees and milking time and, and how things are going. Uh, but it is interesting. You know, it tells us milk per hour, our load efficiency, what our milking efficiency is, and kind of, you know, it pinpoints where were the slow sessions and what might have been going on.

So it looks at our throughput. And so our throughput kind of declines really aligned with our low flow periods, our higher milking times, and kind of this more irregular takeoff. So I'm not sure what was going on. The evening milking of that weeks were kind of a little bit higher, so we need to go and check what's going on.

But it tells us our multiple attaches. So in the morning for this week time period, we had 22 multiple attaches or 22% in the morning milking, 40% multiple attaches. And I don't know if, you know, sometimes maybe they're moving milkers and swing, because we have a swing parlor. So sometimes if you swing a unit over and there's no cow there, well, that accounts for an attachment when really shouldn't.

So yeah, I'm not sure. We need to really revisit those and see what's going on. It tells us we had 1% kickoffs in the evening, a very few in the morning.

So it basically looks at cow flow, um, maybe we have some cows that are taking too long to milk. We have a multiple attachment problem, too many reattachments on cows afterwards. And so we can really look at these across time and see, you know, what, what's actually happening in our herd Now, if I look at it just recently within the last week, basically our cow flow declined 7% and 16% in, in the evening milking.

Our milk has declined by about 8%. Well, that's because we're drying off cows, so we have a lot of late lactation cows. So some of those things we can kind of figure out based on our management But it still says that the evening milking has longer milking times, higher low flow times, and attachment irregularities.

And so we need to maybe revisit with our evening milkers to kind of look at what's going on, and basically it's trying to give us benchmarks as far as what's happening and where we can actually look at to see that. So it is, uh, interesting to summarize all of this data, and I would never be able to do that probably without AI.

And I think it's been great, uh, to be able to summarize information, and I know some people don't like AI. You know, what-- where is my data going? A, a lot of those questions. Um, I think it's a useful tool, and I think that it's only gonna expand into the future, and I'm using it for summarizing some of the data from our herd.

It's quite interesting. It's fascinating. I would never be able to do this if I didn't have AI, and it would take me too long. So it really helps us in our management and trying to figure out some of the issues and challenges that we have. So with that, hopefully you learned a little bit today and on this Labor Day and trying to figure out how to maybe manage, uh, your dairy and effectively use some of that data.

So with that, if you have any comments, questions, or scathing rebuttals, feel free to contact me at the Moose Room. That's T-H-E-M-O-O-S-R-O-O-M @umn.edu, or find us on the web at University of Minnesota Livestock Extension or UMN WC ROC Dairy. And with that, I hope you have a great week. Bye.