View Recording: Concurrency Viral Topics: Whose Brain Is It Anyway? The Model Wars Behind Your Copilot Bill

0:0:6.432 –> 0:0:26.872 Admirals Conference Room what a week it has been. The last week, week and a half, there’s been a lot of stuff going on. Customers are asking us, you know, what does this mean? How do we deal with it? So the point of this little podcast that we kind of put together was we just never know when something’s going to creep up. And we actually caught lightning in a bottle and we’re going to be able to talk about that. 0:0:27.192 –> 0:0:49.192 Admirals Conference Room So I want to welcome everybody that’s been here. We got a lot of people joining. I am joined by Tabatha Frozena, our Director of Sales, and Alex Barr, who is our Director of Delivery. So let’s go ahead and kick it off. Let’s talk about Fable. So for those of you that don’t know, Fable is this 0:0:49.272 –> 0:1:8.432 Admirals Conference Room Awesome model that was released by Anthropic under the code name Methos, you know, originally, and we all got to use it for about two days before the government came and shut it down. And you know, was it awesome? Well, I mean, some people got to do some really cool stuff with it. I only got to play with it, you know, like… 0:1:9.112 –> 0:1:27.592 Admirals Conference Room maybe for like 1/2 hour, but like the bigger story is like, what does this mean, right? Like, is the government going to start controlling, you know, what we have access to, you know, et cetera. Like what, Tabatha, I’ll go ahead and start with you. What’s your perspective on what are you hearing from customers? What are they worried about? 0:1:28.272 –> 0:1:50.312 Admirals Conference Room Well, a lot of them didn’t even get a chance to play around with this, but definitely with the shutting it down, just got fear of what is the right model. Maybe a little bit of the protection side, but just the power behind the technology is freaky, scary. It was able to do some really good, really, really cool stuff. So I don’t think they would shut it down. 0:1:50.392 –> 0:2:9.72 Admirals Conference Room Because it generated an email, like it, it’s yeah, had to have done some pretty, yeah, I’ve heard some things. So, originally, the way that the reason why they didn’t go live with Meetos is because, as part of the red teaming activities, they were able to identify a ton of security vulnerabilities and… 0:2:9.152 –> 0:2:29.512 Admirals Conference Room Mythos was actually able to like act on it. And so they needed to draw that back a little bit in order to feel comfortable about it. And you know, some of the feedback, some of the things that I’ve been seeing on LinkedIn and Twitter is that, you know, the things that got reported after Fable Over Sleep was released are the same things that like OpenAI can do today and other of these, you know, 0:2:29.792 –> 0:2:48.912 Admirals Conference Room It’s just, it’s not an easy problem to solve. And is this really just an assertion on the government’s perspective to maintain some sort of control? And I know it’s kind of like conspiracy theory. Yeah, but like, but there was like this like argument between anthropic and… 0:2:48.992 –> 0:3:10.632 Admirals Conference Room the White House and there’s this IPO and also like a lot of politicians like just wanting to have more control over AI. What do you think? Do you think that comes to play in any of that? I, Alex, what do you think? Yeah, I mean, I hadn’t thought about it that way yet, Brian. My first reaction was, I think there’s Ben, you mentioned fear, like 0:3:11.112 –> 0:3:30.112 Admirals Conference Room I think fear has been a word in AI for a while, and this was the newest version of it. And like the first thing I thought about was, you know, there could companies have been using this stuff now for months and months. And if they built on a model and just someday the government’s like, hey, that one’s a national security risk and rips it out and they’ve like already. 0:3:30.552 –> 0:3:50.312 Admirals Conference Room you know, put people in different places to do other jobs and now this tool is just out, what are they going to do about it? I think we’re going to talk a little bit about that later and get into it, but that was my first thing was just like, I think it’s going to instill some fear of adoption, which might not necessarily be the worst thing. 0:3:50.392 –> 0:4:9.432 Admirals Conference Room because I think we’re gonna talk about like how to responsibly adopt and like where to put your focus. And so, yeah, I don’t know. It was a good segue. Let me ask you this question. So, like these models are getting really powerful and every one of these new generations. Actually, that was one thing that was interesting. Fable was like the first time you. 0:4:9.632 –> 0:4:29.672 Admirals Conference Room heard one of these frontier labs come out with a new iteration, a new version, a new naming convention, a new thing. But as these things get more powerful, Alex, do enterprises really need that kind of power? Or are we doing something like throwing a Ferrari to go get the… 0:4:30.72 –> 0:4:49.192 Admirals Conference Room go get the mail, you know. Yeah, I mean, that is the big question. I think that the people that can figure out how to effectively use these models are going to be the winners. And the ones that just kind of throw it out there and have every big crowd run to it are going to be the ones that end up with a pretty nasty bill. 0:4:49.272 –> 0:5:10.392 Admirals Conference Room at the end of the month. So the answer is no. Like, I don’t think people should be driving Ferraris to the Mailbox, but the people that can figure out which model to use for what problem are going to be the winners. One thing that I don’t think that people understood with this Fable release was… 0:5:10.792 –> 0:5:32.632 Admirals Conference Room In the short term, it was just part of your subscription. Like, you got to be able to use it as part of your subscription. But it was right there in the, like, right at the top of the chat window. It pretty much said, like, as of July 1st or July 15th or whatever, this is going to start to convert to API tokens. They weren’t going to include it even as part of the max plans. They were going to turn that to a consumption-based, you know, agent. 0:5:32.992 –> 0:5:54.392 Admirals Conference Room which, you know, to kind of segue to Cowork, which is I think we’re going to spend a lot of time talking today, is the other huge announcement is the pricing model on Cowork. And Tab, you know, we were talking a little bit about this yesterday, and it was like, well, with GA, that’s the big news, right? Oh, no, wait a second. Yeah, that’s great. That’s great news. 0:5:54.632 –> 0:6:12.792 Admirals Conference Room We should use it. Once we started talking about the economics about this, the conversation got really, really interesting. And I think we’ve got a ton of people on the, you know, on the podcast today are here because that is important. What are you hearing? And you know, what are you hearing? What are the signals telling you right now from customers? 0:6:13.992 –> 0:6:32.392 Admirals Conference Room Well, I just, well, one, it’s a new announcement with a two-week runway, right? So now what, right? I got all excited about this tool. Now I have to plan. Now I have to, so I think the now what, which is, I’m sure why a lot of people are joining today and want our perspective on the now what. 0:6:33.752 –> 0:6:50.152 Admirals Conference Room Because, yeah, do I shut off innovation? Do I have to get the CFO involved? What’s the budget planning look like? Right? There’s a lot of exciting momentum happening of what this can do, but how much is this going to cost at the operational scale becomes the biggest question mark. 0:6:52.392 –> 0:7:11.992 Admirals Conference Room Interesting thing to me was when Microsoft announced Cowork, I had already gotten used to using Anthropics Cowork. And it was like, why are they using that name? And then you start to read about it and underlying it was still just using, you know, Anthropics Cowork. But now we’re seeing signals that it’s not just that. 0:7:12.72 –> 0:7:31.992 Admirals Conference Room You can use GPT. The new Microsoft AI models are going to be a part of that. You know, I’m wondering from your perspective, like, how does how is your how is the delivery team going to prepare for like just not really understanding how to use this, what it’s really good for, how we’re going to advise our customers? 0:7:32.32 –> 0:7:51.872 Admirals Conference Room Yeah, I think the reality is, is we’re all learning. Like anyone that’s claiming to be an expert in this is lying to your face unless they work directly for these companies. And even those people probably don’t have their arms fully wrapped around it. But it’s a lot of a lot of knowledge sharing and just making sure that. 0:7:51.992 –> 0:8:12.952 Admirals Conference Room All of this news that’s coming out and how these things work together, people are staying on top of. The good thing is we’ve got a great group of people that are passionate about this stuff, and I think just naturally want to do that here. But it’s really forcing it into conversations, team meetings, organization meetings, like all company meetings, and making sure. 0:8:13.232 –> 0:8:32.472 Admirals Conference Room people are informed because there’s a lot of information coming out and what’s true today is not true tomorrow. So that’s the scary part of it. That’s a really good point. And you know, that actually is one of the things that just blew my mind is that Microsoft made this announcement like that. It was, hey, here’s this GA announcement. Oh, by the way. 0:8:32.712 –> 0:8:51.352 Admirals Conference Room Nobody has any warning about this whatsoever. Starting tomorrow, you’re going to start paying consumption. Now, we’ve, you know, I’ve been, I looked at the blog post and, you know, we can drop a link in the chat or we can, you know, get this to you afterwards. But like what I’m seeing is the spreadsheet that Microsoft provided. 0:8:51.472 –> 0:9:11.872 Admirals Conference Room breaks it down into four different personas, knowledge worker heavy, knowledge worker light, you know, sales management, you know, etc. And on the low end, they’re predicting somewhere around, I don’t know, like how to quantify it, maybe like 1 1/2 like cowork sessions per day, right, like light usage. That’s low end, that’s what translates to like. 0:9:12.232 –> 0:9:31.112 Admirals Conference Room $150 in token cost or like consumption cost. And on the higher end, like people that might be more heavily users that are going to do one or two heavy usage things a week, they’re predicting like $250, you know, a month, you know, for some of those. And I don’t think it’s unreasonable to say that some power users like me might. 0:9:31.672 –> 0:9:53.352 Admirals Conference Room like hit like 800 or 1000. So the math just changed very dramatically. And that, that’s to the, like the topic that I was, you know, really wondering what your perspective is and the way you framed it up is this is changing so fast, right? Like organizations can’t plan on that kind of change, like where the math changes so dramatically. 0:9:53.592 –> 0:10:14.752 Admirals Conference Room in a matter of two weeks, right? Right. Yeah, I think, so we’re going to get into it, but governance is a really broad word. Yeah. And I think that there’s a lot of specificity under that single word, and this is just one of those components. I’ve heard a lot of people talking about. 0:10:14.832 –> 0:10:36.912 Admirals Conference Room you know, cloud FinOps and this being somewhat relatable to cloud FinOps. And I think at the surface it is like you can put cost controls in place by groups, by people, by tenant level, but I think it’s a lot different than or a lot less simple than cloud governance or cloud FinOps because you’re taking. 0:10:37.432 –> 0:10:56.832 Admirals Conference Room what is this methodology for a small subset of engineers that build workloads on a platform, and you’re now expanding it to the entire company. Anyone with a seat could drive cost up. And it’s just a very different problem to solve. And I don’t think anyone’s got it figured out yet. 0:10:56.912 –> 0:11:16.112 Admirals Conference Room So that’s the fun part of this is we get to be at the cutting edge of figuring out how to solve that problem and bring it to our customers. Robert is asking in the chat that we’ve heard reports of Copilot clearly having issues and some reported Microsoft glitches happening. 0:11:16.592 –> 0:11:31.152 Admirals Conference Room And he said, you know, do we have any kind of speculation why this is happening? I’m going to ask each of you, this is not a question that we proud through anything, like what you’re, Pav, I’ll start with you, like why do you think Microsoft is ripping this bandit up so quickly and so quickly? 0:11:32.512 –> 0:11:52.152 Admirals Conference Room Well, I think this was always the plan, right? To your point, like there was always a plan. They’re a for-profit company, right? And I think it’s timing around their new fiscal year that’s happening as well. But also, I think this continues to make AI real and 0:11:52.232 –> 0:12:16.592 Admirals Conference Room forces, the right use cases and the right opportunities for businesses to make change. Yeah, I was going to say something similar, like just having a flat rate on this stuff to use it was not making them any money. And it seemed extreme to a lot of businesses. Like $30 per user per seat sounded like a lot, but when you actually realize what value you’re getting out of it, 0:12:16.632 –> 0:12:35.392 Admirals Conference Room It’s not expensive at all. And I think right now they’re just trying to reconcile that. The truth is nobody knows how to price it. And I think this is just kind of testing the market, the three or four pricing tiers you brought up of like co-pilot credits and how they’re using, you know, different types of tasks to estimate. 0:12:35.712 –> 0:12:54.272 Admirals Conference Room how much something’s going to cost. It’s somewhat exploratory. And if people pay for it and it goes well and their business is doing well, they’ll probably stick with it. If people stop, pull back significantly, they’re going to need to do what they did with the initial co-pilot seat license, where, you know, it was large organizations, 30 bucks, and then. 0:12:54.312 –> 0:13:13.152 Admirals Conference Room They rolled out a different tier for smaller ones under 300 seats, I think was like half of that. So I think this is going to be similar, but right now it’s just trying to figure out what is the right price for this tech. Yeah. So let me push back on that for a second. Like, all right, so you said you probably, Brian, you, well, I had a thought too. 0:13:13.392 –> 0:13:33.472 Admirals Conference Room But I want to push back a little bit because you bring up a really good point. Organizations are struggling with the $30 a month. I mean, we see this all the time. Microsoft has really worked with the account teams. They’re having a hard time convincing like this to roll this out to the organization. Some organizations aren’t. But now you have this conversation like. 0:13:33.792 –> 0:13:55.392 Admirals Conference Room It’s not just $30 a month. It’s like, that’s just the seat license. You’re going to pay $150 a month minimum for each one of these users that you might use to work. You better have a really strong value proposition. Microsoft only has about 1% of their M365 customers that have actually formally about to co-pilot seats. 0:13:55.672 –> 0:14:19.272 Admirals Conference Room That’s not a lot. That’s showing a lot of struggles in this market. You add this kind of cost on top of that. How do they get over that? How do they make this ubiquitous in organizations where they hit a 50% threshold? Yeah, I think it’s going to be being more intentional about what use cases are actually worth the cost of this. And it is going to cause organizations to pull back. 0:14:19.632 –> 0:14:38.432 Admirals Conference Room short term, every week workflow that was automated is now going to have a price tag associated with it. And that price tag probably outweighs the ROI that it actually generates. And I was joking around with a peer a couple of weeks ago, and he was like, at what point are they going to start? 0:14:38.592 –> 0:14:57.872 Admirals Conference Room bringing humans to replace agents because of the cost of it. And I think it’s a real conversation going on right now. But the one thing it forces is organizations to look at use cases with a little bit more intention and say, you know, is this going to be worth the cost of automating? Which before. 0:14:58.352 –> 0:15:18.192 Admirals Conference Room I think you’re kind of saying a conflicting point. It was expensive before, but you really look at what this tech is capable of, $30 if somebody’s using it the right way is not too much. Oh, yeah, I would agree with that statement 100%. You were going to. No, I was thinking just along the same. 0:15:18.432 –> 0:15:38.152 Admirals Conference Room route. So I’m just thinking of customers that have hundreds and 500 agents just sitting there, but also how many of them are actually doing anything. And now to this point, I just think the collaboration between technology, the technology teams managing these agents and setting up the governance and structure, but the business and the financials of it. 0:15:38.352 –> 0:15:56.992 Admirals Conference Room really owning the value that it does to these teams. Like I just think it’s the organizations actually doing it well are already having those conversations with the business. And it’s not just like IT owns this budget for agents and are the ones setting up the $30 a seat. Like this is going to accelerate and force those conversations to really happen. 0:15:57.712 –> 0:16:16.192 Admirals Conference Room Oddly, you and I had a conversation with a customer probably about 1/2 hour ago, 45 minutes ago, and that they said, we’re still going to use it, right? Yeah. So I think some are going to continue on this path, right? Some are going to control it with the governance controls. 0:16:16.352 –> 0:16:36.912 Admirals Conference Room Here’s my take on this. The question was like this abrupt, you know, kind of aspect of it. I think something changed. I think I think Microsoft either underestimated what these consumption costs were going to look like in the short term and like what the average user was going to like produce with this. 0:16:37.152 –> 0:16:58.712 Admirals Conference Room They wanted to get this GA, they’re ready to get a GA, and they, but they couldn’t go GA with the cost model that they had in place. And it was preview, right? So like they had that ability to make changes. It’s very, very rough. So I just think that something must have happened. Or it could have been the Microsoft side or it could have been the anthropic side. Right. They’re getting ready for the IPO. Yeah. 0:16:59.152 –> 0:17:17.992 Admirals Conference Room they need to show money. And maybe they went back to Microsoft and said, we can’t support you on this pricing model. We’re going to change it and you’re going to have to pass that on to your customers or absorb it. So I don’t know what happened, but the fact that this was not rolled out with any kind of like signals build, you know, we talked about. 0:17:18.72 –> 0:17:37.792 Admirals Conference Room About this, there was some stuff at build that was, you know, that alluded to this, none of that kind of came out, you know, it just seems like real sloppy from a Microsoft standpoint, and so I’ve got a couple of questions here talking about ROI and how can we like, you know, control some of this. 0:17:38.512 –> 0:17:56.992 Admirals Conference Room Tabatha, you’ve got a pro tip. Oh yeah. I’ve got one tip that I’ll share really quickly is if you are using Cowork today, you can use slash cost as like, that was your tip? Yes. Go ahead and explain it. Right now, if you’re using Cowork. 0:17:57.392 –> 0:18:15.792 Admirals Conference Room and you ask it to do a task or agent. When you’re at the end of that, do the backslash cost, and it will tell you how many tokens or credits that it’s used to do that action. So right now, between this time, and like Brian said, we’ll put the link in there if you haven’t seen it on the announcements. 0:18:16.112 –> 0:18:35.712 Admirals Conference Room of the buckets and then how much each of those tasks are going to cost you. That’s a good way to just start right now seeing the things that you’re asking it to do, how much it is generating. So with a backslash cost. Backslash cost. And to talk about. Only in cohort. It doesn’t show you that in co-pilot. 0:18:35.872 –> 0:18:55.152 Admirals Conference Room Yeah, and and there’s some co-pilot stuff that I read into this morning in prep for this that we should we should talk about as well. So, just to give people an understanding of the cost here, a here here’s some of the numbers that that will help you sort of like compartmentalize this. A light thing. 0:18:55.272 –> 0:19:19.792 Admirals Conference Room to do might cost 125 credits. Something that’s medium, they’re budgeting around 500 credits and a heavy workload. A long running task, you know, something that is going to happen is 2500 credits. So we’re talking about 1 cent per credit. So you do the math, this is on Opus, that, you know, 125 to $250 a month comes pretty quick. That’s using it twice a day. 0:19:20.512 –> 0:19:39.32 Admirals Conference Room to get that kind of some consumption for a task. And I think you need to start figuring out, people should actually start figuring out how to use these tools most effectively. A strategy that I have that I’ve developed personally over time is that when I’m using more advanced models or using cowork specifically, 0:19:39.312 –> 0:19:59.752 Admirals Conference Room I’ll use something lighter, low cost, to help me build a prompt that gives like a really explicit set of instructions so that it’s not, you know, just flailing about trying to figure it out on its own. As an example, I think one of our coworkers was talking about using it to update a statement of work. 0:20:0.272 –> 0:20:18.592 Admirals Conference Room template for just changing the client names, right? Like something really trivial. And he used Cowork to do that and it costs like $4 to do just that task. I mean, realistically, that saved like what, 2 minutes time? Not a really good use case. I wonder if he would have used like a different way, you know, if he could have just said. 0:20:18.752 –> 0:20:39.792 Admirals Conference Room Here’s the problem. I can get that down to 50 cents, you know, to really optimize that. So people just need to start thinking about that. I’m sloppy. I have access to these like really, really super high power tools. I’m always running Opus 4A at high or extra high, you know, but I think we need to start coming up with different strategies, not drive that Ferrari. 0:20:40.112 –> 0:21:2.272 Admirals Conference Room you know, to the Mailbox. As fun as it is, I probably shouldn’t do that. Okay, so anything else on? Well, what are some ways that, I mean, people that are like multi-models happening, what are some ways up front, Brian, that can help them know that versus doing it, finding out the cost, but figuring out an efficient way? Is there? 0:21:2.832 –> 0:21:21.872 Admirals Conference Room Well, okay, so Microsoft has got a couple of different ways they’re going to manage this. I don’t know what the token cost difference between Opus and GPT-5 is, but it’s different, right? Microsoft also is coming out with their co-work model, so it’s their own specific model. That’s going to be probably almost a 50% reduction in the token cost. 0:21:22.592 –> 0:21:40.832 Admirals Conference Room And then there’s other strategies that they’re rolling out to help people sort of mitigate these costs. The point is, and what I was trying to like articulate is use tools, you know, for the intended purpose. Use the light, low-cost tools to help you frame up your argument, frame up your instructions. We’re really, really good at that. 0:21:41.232 –> 0:22:2.112 Admirals Conference Room And then once you have a plan in place, yeah, throw it at the big co-work if you need to, if it’s going to be something that’s going to be helpful. And then use that cost, you know, to sort of say, is this worth $5? Is this worth $20? Right? Because you might like think about that and go, well, I’ll just spend another two minutes to do that myself. 0:22:2.992 –> 0:22:22.272 Admirals Conference Room So that’s the way I would look at it. I also just want to call out as we’re on this conversation that this isn’t just the cowork aspect of it as well either. So Copilot is moving into, Copilot is moving into a consumption model, not in a way that you would experience with the chat. 0:22:22.672 –> 0:22:44.112 Admirals Conference Room but more from the standpoint of if you’re using the work IQ components and you’re making API calls against that, they are changing that into setting your workflows that you’re using that use third parties to kind of interact inside of that whole agent, you know, framework. There’s going to be that same cost model going into play as well, and it’s the same. 0:22:44.512 –> 0:23:3.632 Admirals Conference Room pricing structure. So you’re going to be paying those coal pilot credits. Yeah. One thing just quick to tack on, I think when you’re talking about knowing which model to use when, it’s just interesting because not only is this technology itself changing fast and making everyone kind of scramble to figure it out, 0:23:4.32 –> 0:23:22.272 Admirals Conference Room But even the definition of something you think you might be doing as a company or you have already done. So take adoption and training, for example. Six months ago, adoption and training didn’t necessarily include thinking about which model to use when, because if there wasn’t a big problem that existed. 0:23:22.752 –> 0:23:28.192 Admirals Conference Room And so I just would say that I think it’s important that… 0:23:29.312 –> 0:23:48.992 Admirals Conference Room As this stuff evolves, you’re re-looking at everything that you have in place to make sure its impact all the way upstream is is accounted for downstream. Fairly to call out, ’cause yeah, when it’s a fixed price, use use any tool, but when there’s these might need. 0:23:49.152 –> 0:24:9.472 Admirals Conference Room to then make sure you are either adding new models to your organization that you maybe didn’t consider before or different ways to use the models that you have available and set those structured from training and adoption. Yeah, absolutely. So Cowork is 1 kind of like autonomous, like workflow agent that we’ve sort of discovered. Microsoft that build announced this. 0:24:9.632 –> 0:24:28.352 Admirals Conference Room scout, you know, agent as well that we haven’t gotten an opportunity to play with yet. You know, it requires a few heavy-handed steps. And I honestly don’t recommend any organization just like turn it on. I think you do this in a kind of a pilot. You got to turn open claw on. 0:24:28.672 –> 0:24:48.512 Admirals Conference Room You have to have consumption wired up to your GitHub Enterprise, you know, for the tokens and consumption. And we don’t know a lot about that. But strategically, do you think that this is a way for Microsoft to give different options? Or what do you think the use cases are for this Scout agent? 0:24:48.592 –> 0:25:9.312 Admirals Conference Room that’s coming out. I think it’s just kind of a natural evolution. Like even when you think about the simple chat bots where you would, when some of this stuff was first getting hot on the market, people were building chat bots and pointing it at one or two documents or anything about like an HR use case of pointing at your human resources handbook. 0:25:9.712 –> 0:25:30.352 Admirals Conference Room to let employees answer questions. And as this evolves, like people don’t even want to have to ask the questions. And I think the whole premise of Scout is that it’s just kind of continuously running and serving things up to you versus you having to ask questions. So it seems like a proactive play on it is what it feels like. 0:25:31.152 –> 0:25:50.752 Admirals Conference Room I agree with you. Like I would be a little bit cautious about just like flipping it on and letting it run because I think there’s a lot of information that still needs to come out about it. Specifically pricing. Like I haven’t seen any formal pricing model around how Scout’s going to work. I’m assuming it’s going to be somewhat consumption based. 0:25:50.832 –> 0:25:56.912 Admirals Conference Room All consumption, yeah, but what does that mean for something that runs without being prompted? 0:25:58.432 –> 0:26:5.232 Admirals Conference Room Dan? I don’t think I have too much to add on that. I do think it’s to serve different use cases, but also… 0:26:6.832 –> 0:26:25.672 Admirals Conference Room Yeah, continue to win at the AI game. Yeah, I think it’s a way for Microsoft to have an exit strategy out of this co-work thing if it doesn’t pan out the way that they want. That’s kind of my prediction is that, you know, it’s a, I still don’t see a… 0:26:26.112 –> 0:26:46.192 Admirals Conference Room a distinct value proposition for one or the other. I don’t see enough difference between them. But what I did see is that this is the first of these autopilots that Microsoft’s talking about. And so maybe this is an exit strategy for them to promote their own way of building these types of long running autopilot, autonomous tasks, something like that. 0:26:46.592 –> 0:27:5.792 Admirals Conference Room We got a couple minutes left. Any predictions? Any thoughts? Just one question for you. I think a lot of what we’ve been talking around is the governance word that I used before. And I think it, generally speaking, when it’s heard, people think data readiness and people think. 0:27:7.72 –> 0:27:30.752 Admirals Conference Room cost and I think it’s way more multifaceted than that. So I’m just your opinion on like what are the key pillars to really make sure you’re looking at when you hear a vendor or anyone in your organization talking about governance with AI. Yeah, I think you hit the nail on my head that data governance is really when people say governance, they’re thinking about I want to make sure that my data is protected. 0:27:31.232 –> 0:27:55.152 Admirals Conference Room Like, I’m not losing company secrets. And, you know, we’re seeing a very distinct shift right now in the cost of AI, how it’s going into more of a consumptive model. And I think that this Percy model and a fixed cost for people using these tools, it’s gone, like at this point, and we need to prepare for that. And so the governance things that we need to be start doing include some of the cost optimization. 0:27:55.352 –> 0:28:14.512 Admirals Conference Room and how we’re controlling costs. If we’re doing pro code projects, we’ve been doing that for years now, for a couple of years. But when it’s these, you know, tools that used to be per C, like nobody cared. And we got to start caring quite a bit about what that’s going to look like. And to the point, you know, when we sort of glossed over this in co-work, 0:28:14.912 –> 0:28:37.72 Admirals Conference Room you, you know, pretty easily can go in and say, I’m going to set this user group has access to it, this everybody else doesn’t, so you can kind of control it that way. And then you can set, you know, usage limits or consumption limits, you know, on a user basis, group basis, and the organization-wide. So if you’re not doing that today, I think you should do that. 0:28:37.392 –> 0:28:55.952 Admirals Conference Room I think every organization should figure out what is that maximum that you want to put. And that doesn’t mean that people can’t use it. It means that they’re going to have to click the button that says, like, can I use some more? And somebody’s going to have to prove it. So I think that’s the shift. I think that our governance work is going to include more cost modeling. 0:28:56.432 –> 0:29:20.672 Admirals Conference Room than before. And I think, you know, my takeaway for people that are on this call is start thinking about those, like, what is the value that I’m getting out of this agent, that this task? Because I think that’s what people just haven’t really cared about. You need to start thinking about it. Yeah, and I would just say like the role that may not have existed before within your organization that will need to exist to actually manage all of this. 0:29:20.872 –> 0:29:38.992 Admirals Conference Room Because another governance pillar that I think is going to start getting hot is just like this model maintenance, right? You build agents or you build workflows on a specific model and then it gets deprecated, then what? So people are going to have to manage these things. They’re going to have to understand the cost of using them. 0:29:40.272 –> 0:29:49.472 Admirals Conference Room And there’s some new roles that will exist in companies that don’t exist today. I thought there were going to be less jobs with AI. Pat, you got the final word. We’re right at the top. 0:29:51.152 –> 0:30:9.152 Admirals Conference Room I think this is just going to be a fun, exciting time. I really hope it doesn’t slow down the adoption because I think these are, it’s very powerful for organizations and we’re seeing just processes, governance, everyone sit up a little bit straighter and really think about what’s their secret sauce within their company and how they can use AI. 0:30:9.552 –> 0:30:20.512 Admirals Conference Room So I think it’s an exciting time. We’re all buckling up and happy to talk more. All right, well, fantastic. Thanks a lot for joining us. We’re out of time. See you next time. Thanks, everyone.