Whitney Lee: So, I think the easiest job in tech has got to be a salesperson for cost saving software. Yeah? Because it's just a simple math equation. It's like, how much does it cost? How much do I save? Is it, you know, is it good? Let's do it. Right, right. As far 

Coté New: as convincing people that it's something that you should be interested in and, uh, and buy.

It's maybe even easier than selling eggs. Thanks. It's just like, I don't know, what kind of eggs are they? Do I like them or not? Do I want eggs? Yeah. Is there 

Whitney Lee: even much opinion in there? It's numbers. Right? 

Coté New: That's right. It's just a matter of if it, uh, what's, what funny little symbols in front of the numbers, uh, that, that, that you care about, maybe, maybe if you use decimal places.

Have you encountered this in Europe where they, they use a comma instead of a decimal place? And in their numbers, I don't really, I tried to look this up once, like what the theory was. And I think like so many weird things here on the continent, it has something to do with the French Revolution. Um, I'm not really sure, but I feel like it originates there and people just got, they've missed, Napoleon probably screwed it up.

I don't know. 

Whitney Lee: Coté, before you derail us further, let's introduce our guest, Phil Andrews. Phil, will you introduce yourself, please? 

Phil Andrews: Sure. Uh, so I'm Phil Andrews. I'm field CTO at Cast AI. I've been here a little over three years now. Uh, we're selling, uh, you know, we work on cost optimization and resource optimization software.

Um, my prior life, I was a director of engineering at Oracle cloud. Uh, so a lot of background in security products before that, you know, we had built substantial amounts of, um, you know, net new products to go into the cloud there, so decent cloud background. And, you know, now we're, we're selling into the resource optimization space, which is interesting.

Whitney Lee: So Phil, is it the easiest job ever? Cloud software. Do I have cost saving software? Do I have it right? 

Phil Andrews: So I'm not a sales guy. I'm an engineer. Um, I have to sell as part of my job, but, um, but I'm not at my heart, a sales guy. This is the first sales job I've had in my career. Um, however, talking with our account executives, talking to our actual sales people for them.

It is an incredibly easy value proposition for exactly the reason you said, right when you say, look, we're gonna save you a million dollars, it's gonna cost you $300,000 a year to save that million dollars a year. Most companies are like, wait, that's it. I get to net out 700 KA year, , where do I, where do I sign?

Can I do this twice? Like ? Most, most people are excited about it. The problem comes when you start, you know, um, everybody has an idea in their mind about what something is worth. Like, well, you're saving me a million dollars, but I like, I think I only want, I want a 10 X ROI and you go on. But can't you just say, 

Whitney Lee: Oh, walk away.

Like, like, what do they really have to negotiate with? 

Phil Andrews: Yeah, that's a big part of it. I mean, there's not a lot of players in this space right now. So yeah, that is part of the 

Whitney Lee: negotiation. Call us when you're ready. Yeah. 

Phil Andrews: Yeah. And there, there have been a few scenarios like that, uh, where, where that was the end of the discussion where we said, okay, well, good luck.

We hope, we wish you the best, you know, come back and let us know if you need our help. So 

Coté New: would you say that like you're, you're in the FinOps business? I mean, is that, is that an accurate usage of that phrase? 

Phil Andrews: No, we're in the automation business. 

Coté New: Right. But, but so, so what, what would be the difference between how you describe yourself and then what a FinOps would be?

Phil Andrews: I view 

Coté New: FinOps 

Phil Andrews: as, uh, managing, organizing, aligning financials around the operations and technical space. Uh, FinOps teams are rarely in interested in how you get to the end result. They have it as a result of, we need to spend less money, we need to get to a certain point on our spending, or we need to be able to allocate our money to who's spending it.

Like, they're very concerned about how much money and where is it going and where is it coming from. 

Coté New: Right, it's a lot more, a lot more reporting and gathering versus like, To use a, uh, an analogy, there's a lot more reading than writing that goes on. And, and so, so, so do y'all limit yourselves just to writing from a financial angle, or is there other stuff that you automate?

Phil Andrews: So, um, we are moving into two other spaces. So a lot of the stuff we're doing right now is financially based, uh, because to your point, it's a lot easier to sell money in a down economy than it is to sell like DevOps tooling or other growth. Yeah. Yeah. Um, you know, oh, this will, this will make your developers happier.

That's not a good sales pitch these days. Executives don't care about making their developers happier. They care about keeping their company alive. Um, you know, so from that side of things, that was our first go to market. is the resource optimization, cost optimization. Uh, we're spanning out into our goal is always going to be net neutral costs.

So some of the areas that we're spending out into other cloud costs, uh, we're actually going to be launching some, some additional products. I can't mention them yet, uh, but they're going to be more around performance optimization, but the performance gains will result in a cost reduction that will have cast AI being a net zero cost.

So we'll basically increase performance. At a net zero cost to the customers. And so we are going to be moving into more performance management in some of our new products. 

Whitney Lee: So I see it. I was joking at the beginning that it's a simple math equation, but I don't think it's quite as simple. Like I can think of a few other factors besides, so there's automation, there's right sizing, but then there's like literal cost.

Like I understand some cost saving software is actually monitoring cloud APIs and whatever else to make sure you have a great instance at the moment. And then there's also the cost of running the cost saving software. So whether it's training or, uh, the compute itself of the, or the time cost or so, and then of course the literal cost.

So how much do those other factors play in? To our not so simple after all math equation. 

Phil Andrews: So those factors are a lot more present in some of the more reporting tools out there, the tools that are giving recommendations, because any tool that's going to give you recommendation, you need an engineer clicking refresh on some cadence and then going and doing the work to apply those recommendations.

So we tend to see customers getting exhausted with those types of scenarios where they're, um, constantly chasing. The recommendations, and they've got an engineer basically dedicated full time to doing that. When we move more towards full automation, there is a ramp in the beginning, and we see customers, you know, one to two months of kind of ramp up time in the beginning doing the implementation across their platform, but it's much more of like a set it and forget it, uh, once that's in place and as it's kind of going forward from that point.

Whitney Lee: That gives you, that makes me have a couple of questions. One is, um, What are you monitoring constantly? Like, is it a one and done? Like, I have recommendations. I apply the recommendations. Then why do I need the software at all? Or are you constantly monitoring it? 

Phil Andrews: So we do get that question a lot too. Like, well, once you fix our environment, aren't we done?

Well, sure. If you have an extremely static environment where nothing ever changes, but if that's the case, then you're doing kubernetes wrong in the first place because you're paying for whatever your peak usage is. 24 7, 365, you're paying for whatever that flight usage is using kubernetes and cloud native correctly, we'll say, uh, being able to scale based on load, you know, um, just in time capacity for being able to manage your load, which was the whole point of the cloud in the first place was that you, you weren't designing for that Amazon black Friday scenario anymore, you were designing for whatever the capacity you needed at that time is in that case.

And that's why these recommendation engines don't tend to work very well. Because your Black Friday mixture of machine types and resources is drastically different than, you know, a random Tuesday in June. Uh, you know, those are very different scenarios. So, a tool like Cast. ai, yes, we can automate things and get it into place in the first time.

But when you turn it off, depending on how volatile your environment is, it's going to trend back towards being expensive and being inefficient over time. 

Whitney Lee: So then, if you're up in there doing all kinds of awesome things, Automation about every little part of the, of the system. Doesn't that require extremely privileged access?

Like, aren't there security implications? 

Phil Andrews: There are, um, and this is something that we, we do go through with a lot of customers. Um, you know, it's, it's important from our side that we have, uh, all of our certifications. And so we've gone down that path of, of getting the certification squared away. Uh, but also we have.

path of least privilege, right? We try to make sure that we're using the least possible, uh, resources necessary, the least permissions necessary to make sure, uh, we have just amount of access that we need. We can create and remove virtual machines from a customer's AWS instance or GCP, uh, account. However, we can't access things like secrets.

We can't access things, uh, logs or user data or application data. So we have it scoped narrow enough that even though we have operational control over some assets, We don't have data control, so we don't have access to any of the data and the data is what people are mostly concerned about. I mean, obviously they don't want anybody taking their system offline.

Uh, and that's, that's a trust, you know, that's us building trust, but the data side of things is why we're very careful in how we structure things. 

Coté New: So I, I was, I was watching a talk you did at, um, uh, what is it? KCD, the, the, the kubernetes community day. Yeah. Yeah. And, and it was great. I think, I think you can reverse engineer the, um, Some of what you were just saying, right?

Like, like I, uh, based on the, the tips that you were given, the, the, the common, I mean, the, I forget how many there were, but it was sort of like, here, here's common ways people, uh, do, uh, kubernetes wrong that causes problems as far as accounting and cost stuff, lots, lots of, it also kind of exhibited lots of, um, well, you've got to pay a lot of attention to the resources you're using, which, and, and so like, you know, It was making me think like, like I, a lot of times when, when I, when I hear about, uh, problems in kubernetes land, I often think like, I thought it did that.

Right. And so like, one of the things that I, I guess I have a more appreciation for now is having listened to that is like, it doesn't really pay attention to scheduling stuff efficiently, or it only does it one limited way. And so in fact, you've got to do a lot more work, right. Like to make sure that you're, as you were saying, like, you're not only using 10 percent of something or, or whatever, right.

Even, even the. I don't know what the phrase is in the trade, but there's that noisy neighbor or maybe greedy neighbor that's just like, doesn't allow you to, to do stuff well, while you, as you say, put your stateful stuff in a corner somewhere. Um, anyhow, like, from, this is a weird way of asking you, but like, isn't that kind of disappointing that kubernetes doesn't already do that?

Like, it seems like it should. 

Phil Andrews: Yes. Uh, we, we get a lot of asks from customers being like, Hey, we need this specific set of resource, you know, constraints and requirements. And we say, well, kubernetes doesn't offer that. Like there's a lot of customers out there that want to keep say a 10 or 15 percent unscheduled space on their nodes.

They said, we want that 10 or 15 percent empty and allowed for, for different services to be able to burst and use that space. We want to block that out. That's incredibly difficult to do in kubernetes. There's not just like a toggle that you can set that sets that up. There's some kind of back behind the scenes stuff that you can sort of do, but it doesn't really do it that way.

Yeah. That's just a common thing that people want. And kubernetes just doesn't offer it. So then we have to find kind of workarounds using within the framework of kubernetes. And that was, you know, we had talked on the previous session about placeholder pods and, you know, that's something we've kind of worked with a lot of our customers on because kubernetes offers you a ton of different tools and options and available resources out there.

Figuring out how to assemble them into a good, highly resilient and cost effective configuration is what's really challenging. 

Whitney Lee: Speaking of noisy neighbors. Am I hearing a rooster in the background? Yes. Am I not? Okay. 

Phil Andrews: My rooster occasionally crows all day long. So, and we're actually having some warm weather in New Hampshire right now, so I actually have my window open and unfortunately, uh, you guys can hear.

I'd say fortunately. 

Whitney Lee: I'm enjoying the color. Do you have other animals? Like what's your, what's your situation? What's going on outside of that window? Paint us a picture. I do. So 

Phil Andrews: I live out in the woods in New Hampshire on eight acres. Um, so we have, you know, a small kind of hobby farm. Uh, we've got Around 30 pigs right now.

Whitney Lee: 30 pigs? Yep. 

Phil Andrews: Uh, four lambs. I think we have 

Whitney Lee: lambs, 

Phil Andrews: I think. Where are we at? 16 chickens. I think. I think we've got some. Oh, okay. Okay. Well this, 

Coté New: this, this, this begs the question, when does Hobby Farm stop being Hobby Farm ? Um, when it actually turns any sort of money at all. . Ah, I see. I, I, I, I feel like maybe, okay, so.

Oh, so when you, then you become maybe amateur farmer? Cause I think amateur people can make, make a profit, but hobby, no profit. Okay. 

Phil Andrews: Yeah. I'm, it's just a big cost sink right now, but it's all right. Do you, do you do that 

Coté New: thing where at the end of your driveway, you've just got like free eggs? You just have, you just have an excess of eggs you can never get rid of.

So you're, you're trying to get rid 

Phil Andrews: of them. Not yet. Uh, the eggs are my daughter's business. So she sells eggs to her family and to her grandparents across the street. So if she gets to a point where she's running a surplus, 

Coté New: she might do that. Well, there you go. You're, you're sort of like, uh, while it's hobby farming, you're kind of like a VC.

For smaller businesses. You're doing some, some startup funding for a, for another thing, kind of running at a loss, hoping, hoping the long term return will a hundred X your, uh, you know, your 

Whitney Lee: farm cost to saving software. 

Phil Andrews: So I used to do laundromats actually. And I was really big into cost savings at my laundromats when I ran laundromats in the past.

So for 10 years, I ran three different laundromats as a side gig, as a side hustle outside of. This is my day job. Uh, and that was a good line. 

Whitney Lee: What's something you learned about humanity from running a laundromat. 

Phil Andrews: There are a lot of good people and a lot of jerks out there and you got to try not to let the jerks ruin it for you.

We met 

Phil Andrews: so many awesome people in our community and, you know, like friends that we would know just from like their appearance and stuff and you'd meet them at the store. And then every now and then you'd have something like ridiculously awful happen at the store. It was so disheartening. But you had all these nice people that you had met too.

So it's like one person for every hundred was a jerk. And you had to try to keep that ratio in mind because those bad scenarios, Holds so much more weight. 

Whitney Lee: Why are we programmed to do that? It's not a nice, nice aspect of kubernetes. into it twice 

Phil Andrews: and those were very, you know, trust breaking issues, uh, unfortunately, but, you know, but overall it was a good experience.

We, we sold the last one two years ago. Uh, so we've, we've gotten out of the laundromat business. We were too busy with our, with our life and our kids and our hobby farm. So. 

Whitney Lee: Yeah. Is there some, is there some sort of. article of clothing or something someone could drop off to you and you could be like, oh no, this is so much work.

Either a stain or a type of clothing. 

Phil Andrews: Horse blankets. 

Whitney Lee: Horse blankets? Horse blankets. 

Phil Andrews: We had, we banned horse blankets from all of our stores because they're big and heavy and have like metal buckles and stuff on them and they like destroy the machines. Oh yeah. That was our, that was our biggest issue was people washing stuff in the machines.

They're like, I don't want to wash this in my home machine. So they'd come to the laundromat and you're 

Coté New: like, exactly. 

Phil Andrews: Yeah. 

Coté New: And I guess, I guess in New Hampshire, there's plenty of, uh, plenty of horse blankets where I am. There are, 

Whitney Lee: I attended a wedding in India and I want to clean my, sorry, and I'm trying to take it to a dry cleaner.

And everyone's like, no, I'm sorry. There are too many. There's too many. And I was just like, well, how do I clean this thing? I still haven't figured it out, by the way. So, if anyone has any comments or suggestions Surely this is a problem that's been solved, but I haven't figured it out. 

Phil Andrews: Yeah, our dry cleaner was always very sensitive about things that were too bedazzled.

Uh, cause they would always get complaints about it, right? Oh, I lost four sequins. I'm upset. I want you to refund me this, you know, 500 I spent on this. People are like, 

All right, we can 

Phil Andrews: refund you the 28. 99 that you spent on dry cleaning it. Here's your 29. 

Coté New: There's an interesting, like, uh, I don't know how you would categorize it, but, but a business way of thinking kind of in that is a lot of the times the business decisions you make are based on how much it can blow up.

Like losing the sequences versus like the value that you bring to it necessarily. Right. Just like, Oh, and it's, 

Phil Andrews: and it's the same way in our current industry. Right. In my, in my day job. And we had a customer that we were saving around 4, 000 a month and. They went through with production rollout without talking to us.

And they caused a production outage for 45 minutes in their system because they clicked the wrong button or did something that they weren't paying attention to. 

Whitney Lee: And 

Phil Andrews: so they rolled it back and they're like, it's not worth it for us to save 4, 000 a month if it's going to cause instability and, you know, Well, we tried to coach you through it and we set up calls and you did it without us.

Like, and they're like, well, you know, 4, 000 a month just isn't worth it. So yeah, there's absolutely different business constraints there where at what point does savings, you know, outweigh a slight instability. And sometimes it does. But other times it doesn't. So, 

Coté New: so several times you've mentioned with customers, we'll see if we can make a laundromat connection here.

But like, like, like I, you know, maybe it's the answer to both things, but you've mentioned with, with customers at your current place, not, not the, not the laundromat, like working with them and doing things with them, like you were just saying, and I wonder, I wonder when you, when you think through kind of product managing the business or kind of building out the business, when do you decide, like, We're going to do white glove service, right?

Like, we're not just going to like, you know, put the coins in and we never see you like, you know, do the laundry and leave, but we'll actually have a conversation with you, just like you said, like we have an ongoing dialogue. Like what, what makes you want to enter that kind of business versus like. Good luck getting a hold of a human business.

Phil Andrews: Yeah. So, I mean, I would say almost everything we do is white glove and that's been a large part of our success. It is challenging to scale that because you need a high level of expertise, um, to be able to, to do that. And I mean, it was the same when I was running the laundromats, you know, if, We got a call at midnight from our 24 seven store that somebody was having an issue.

I was getting dressed and driving over there at midnight. Luckily it was only three miles from my house. 

Whitney Lee: What's the incentive to have a 24 seven store? Like you're getting called at midnight for, for what? For like 30 a profit? 

Phil Andrews: Yeah, we had a lot of, um, second shift people that would come in after they got out of work at 11 or so, so we'd be there at one o'clock in the morning.

Um, we did close during nighttime hours for a little while after the first break in that we had, um, and people were really bummed out about that. They were, they weren't upset at us. They were upset at the people who, that the person who caused us to need to do that. Um, so yeah, there is a lot of business there and we didn't get calls in the middle of the night.

Horribly often. Um, what you did, it was a very grumpy, grumbly kind of thing. Um, but we would go over it. 

Whitney Lee: And then with cast AI, is it a white glove experience to get it set up in the first place, or does it need to be like white glove all the way through using it? 

Phil Andrews: Just to get set up and do the training.

And a lot of that is because it is. Right. And we, we have had customers that did the whole setup and did their own POV, did it all by themselves, you know, got it working. They were like, yep, they're thrilled. You know, where do we sign on the paper to buy it? Um, but a lot of customers, you know, you've, you've got these poor DevOps engineers out there that are spread so incredibly thin.

I mean, they're expected to know monitoring, alerting, you know, CICD, you know, setting up the pipelines. They get yelled at by developers if anything goes wrong anywhere. And. They could go out and do it on their own, but then it's putting it on their head and on their shoulders. If something goes wrong, we'd rather just schedule.

I mean, our product is pretty easy to set up. I mean, most customers, we've got three, one hour meetings and they're fully optimized and they're ready to go. So for three hours, That's a relatively low expense to make sure that they're successful and that they feel good and that they know their way around the product.

Um, if it was, you know, I've, I've talked to some other vendors and they're like, Oh yeah, you know, we typically have a three to four month onboarding with our customers. I'm like, ouch. If you're white gloving three to four months, that, that doesn't scale real well. Like you need so many people to handle that.

Yeah, yeah, exactly. That's a lot of gloves. Uh, but with us where it's relatively easy onboarding, you know, we can, we can get them up and running. It's worth it to us to, to do that white glove service and make sure that we have, you know, sales engineers and, um, solutions architects that are able to, to do that onboarding and make it effective.

Coté New: So, so, uh, what, what, what do you think the deal is with like, so I, I used to work at like, uh, BMC software way back in the 2000s, right? I mean, I don't know. They do everything. I don't know what they do now. They're splitting themselves up now and doing all sorts of stuff. But it was, it was, this is where I learned this quandary that in the, it's probably in all of the software world, but like the infrastructure software world, where it's very hard for like a successful.

Vendor service, whatever, to like adapt to the new infrastructure that they manage. Right. And it's like, otherwise Datadog wouldn't exist. Like BMC and CA and HP and IBM or Tivoli would have just like monitored it. Right. Like, like all, all these new companies come along and I guess there's a, you know, like, uh, like, you know, the new relic people used to, whatever, there's the people who kind of like launder, uh, startups over and over again, so to speak.

But let me get to the point here. Like, why does that happen? Like, like, why can't I'll pick on them, but you know, it's just representative. Why can't the likes of BMC come in and just. Do what you do, right? Like, not that it, like what's holding them back from whatever magic that y'all have that allows you to do interesting work here.

Phil Andrews: Sure. 

Coté New: Um, I mean, 

Phil Andrews: they could, it's time, money, and expertise. You know, are they willing to invest time, money, and expertise to, to build something, right? I mean, The expertise that we bring to the table, we have a very strong depth of kubernetes knowledge on our side, uh, which can be acquired right for the right amount of money.

Any company could buy that. The biggest thing for us is the, the actual node auto scaling piece. And that's why if, you know, if you look around kind of in this space, there aren't, there's nobody else in this space that's doing it effectively. You've got Carpenter that's out there. That's open source that has hundreds of developers contributing to it.

And we still beat Carpenter by 30 to 40 percent head to head every time we go ahead. 

Coté New: And it's something like that. Like, is that. Is, is, is the moat around that like trade secrets and IP and patents, or is it just like, like your magic, magic sauce? I'm mixing metaphors, the, the secret sauce, like, you know, like why can't other people figure that out and do it?

Like, why is it only y'all and, and, you know, this is a specific question about this general, you know, generalized to every other kind of big incumbent company versus startup like this. But like, there's always something that. It seems frustrating or it's not frustrating. It's weird that, that it exists like this.

Phil Andrews: IP is part of it. Um, so we do have some good IP backing our products, uh, but time and industry, we started down this path five years ago when money was free. Right. Cast. ai was founded in 2019 when everybody was getting massive valuations, 50 X, a hundred X valuations because money was free. Nobody was worried about saving money.

We built the core of our platform back then, and we've got five years building it. It's, I mean, the, the amount of nuance and coding in there to protect against bad scenarios, to protect against, you know, negative outcomes, um, it's a lot of code, right? It's a lot of time. It's a lot of edge cases that we've covered.

It's We're still encountering today customers that we come into and they go, Oh, but I've set up like this. Can you handle that? Yes. We need to, you know, add some code to it. We need to do some handling. We might need a feature. You know, so even after dealing with, I mean, I've probably talked to a thousand customers.

We have a little bit. Yeah. We've, we've got well over 200 that are onboarded as, as live active customers. I forget the number exactly. But, um, when you see that many different environments and that there are many different use cases we've built. There's just, there's a lot of, you know, I'll call it travel knowledge, right?

But there's a lot of iteration over time that's been built into the Cast AI product. We have a good enough lead that a lot of the Me Too companies that have come out in the past two to three years and just started in the past two to three years. Can't catch up. It's, it's such a big chunk of work. And that's why we're seeing a lot of people entering the workload rights sizing space.

Right. When we talk about container right sizing, I know we had talked about it in our previous encounter, Whitney, um, about pod right sizing, workload right sizing in kubernetes. That's a much more accessible and lower risk space. And that's why you're seeing 3, 4, 5 companies come out in the past couple of years that are entering that space, because it is much easier to get into that space.

It's a lower risk space. 

Coté New: Yeah. So, so like maybe to, uh, put into like a theory or a law or whatever, kind of what you're saying there is at some point, however many years ago, several years ago, uh, a big bag of money showed up on your door and that allowed you to start working on this a lot earlier than other people and the other angle of that is it also allowed you to take the risk of starting on something that maybe was a bad idea, right?

Like, you know. That if you were at a larger company, I mean, this is almost. I don't know. Maybe it's a little bit of, of, of disruption theory in there, but there's some aspect that's missing from it. Whereas at a larger company, it's not that they wouldn't think this would be a good idea. They're just like, or I could keep money, making money over here.

I hear 10 ideas a day and I know that like 0. 5 of them will be successful. Right. And so like the way that we'll run our industry is we'll let people like y'all just like, wait for the magic bag of money. And then maybe some of you will be successful, right? And so maybe when you're at a larger company, you don't really like, you don't want that kind of like risk management.

What was it? I imagine, 

Whitney Lee: oh, at a larger company, you wait for a smaller company to get it right, do all the experimentation, and then you just buy them. 

Phil Andrews: That's very common. What is it? 3 percent of startups are successful. So you got these large companies out there watching for those 3 percent and then either throwing a huge bucket of money at it internally to just rip it off or buying, you know, whatever the most successful incumbent is that you know, that, that, that.

Started down that market. And, and for us, it was actually, this is the third startup our founders have done. And Cast. ai came out of the problem that they solved in their last startup. Their last startup, their AWS bill was just increasing like 10 percent month over month, over month, over month. And they just, every time they would try to knock it down and get it under control, it was the next month, it was still up another 10%.

They're like, we don't, we have cloud health. We have the tools reporting. We it's telling us where all the money's going, but we just can't fix the problem. Something you 

Whitney Lee: said earlier about the overworked DevOps engineer made me. Made me think of a problem or an interesting space that y'all are in, which is the people who care about cost are different people than the ones who need to implement this software and who are managing the infrastructure.

So how do you navigate, like, someone is telling me to come in here and I need to mess up all of your stuff or deal with, and, and the people whose stuff you're messing up, um, are angry about it? 

Phil Andrews: Sure. And that's the bigger the company, the, the more landmines there are. 

Whitney Lee: Yeah. 

Phil Andrews: Cause usually it's, you know, somebody in the, you know, financial side of things goes to the CTO and says, you need to cut your budget by 25 percent or you can just cut your cloud spend by 20%.

And so the CTO goes, okay, everybody go find money to save. And they go, ah, they're looking at a part of everything. Right. Because it's not useful work. It's useful work in that it's keeping the company alive, but it's not like interesting work. You know, it's tedious 

work. 

Phil Andrews: And so that's where, you know, we need to help the DevOps engineers be like, Hey, look, we're actually going to take this tedious work off of your plate.

We're going to help you guys because you don't want to go look through 700 different AWS instance types and look at the prices for them and pick which one is best and do that again next month and do it again the month after that. 

Whitney Lee: Is that the AI part of cost AI? 

Phil Andrews: It is part of it, yeah, is being, is keeping the, the analysis around the different pricing trends across the different clouds.

Spot instances is a big part of it as well. Spot instances are sort of a black box in the cloud. What is a 

Whitney Lee: spot instance? 

Phil Andrews: So spot instances are an instance that you can buy on the clearance rack. So it'd be like going into a discount store. They've got the clearance rack. It's 80 percent off. You may or may not have the shapes and sizes on that rack that you need, but it's, what's, it's what's left over.

And so you can buy those at a huge discount. The only catch with the cloud providers is they can take it back from you anytime they want. So they give you two minutes notice and say, yeah, we need to sell this at a higher price to somebody else. Um, so you need to figure out a plan B and we, we can help predict when those are going to be taken away.

And we can also replace them with more cost effective options, you know, as we move things around. 

Coté New: So, you know, Nat, to, to, to go back to laundry. Cause you, you brought, you brought up like fun pricing games. Like how, how do you, how do you think about pricing strategy and like in laundry? Like, is it pretty straightforward or like, do you put a lot of thought and experimentation in how you, how you price things?

Phil Andrews: It was relatively straightforward for us. Um, we were also, you know, all of our stores were, had effectively no competition. So it was, uh, before they would drive 20 minutes or 30 minutes to go to a different laundromat. Um, you know, and not that we wanted to, you know, gouge our customers, but when you're a small independent little store in the middle of nowhere.

You know, you have to keep yourself alive with rising costs. So you don't have to do 

Coté New: like, like Thursday, Thursday night laundry and disco specials. No, we tried 

some of that stuff when we first 

Phil Andrews: got started 

Coté New: and 

Phil Andrews: it turned out to be useless. We bought like 10 pizzas and had like a big customer appreciation celebration when we bought new dryers at one of the stores.

We had like one family that lived next to the laundromat come over and like, Eat like two pizzas and like take two pizzas home with them. And like nobody else, we were eating pizza for like two weeks after that. It's like, like 

Coté New: optors opt, uh, like, uh, opportunistically ginning. Um, yeah, yeah, it was 

Whitney Lee: in central Ohio where I grew up, there was a laundromat that was actually.

Cool to play at. There's a couple of them, actually. Sudsy Malone's, I believe, is one. But that's where, like, you would go to see shows at the laundromat. Oh, yeah, yeah. Yeah. And Krusty Punk's would get their clothes cleaned, finally. 

Coté New: Yeah. Yeah, I like, you know, There's, there's a good, good analogy with, with, with your new dryer stuff.

Like a lot, a lot of stuff in like in the infrastructure world is like, people get very excited about the new dryers and everyone else is like, I, I like dry clothes. I don't understand why this one is anything special, but why do we have pizza today? Yeah, exactly. 

Whitney Lee: So, 

Phil Andrews: uh, when I was doing my MBA, uh, one of our, my other students in the class did a presentation on his laundromat that he owned in kind of a city, you know, about an hour from where I live.

I was going through it. I'm like, that's a really interesting side hustle. I'm like, and this was when I was young and poor. My wife and I had only been married for a year. Two or three years at that point. And we didn't have a lot of extra money. We had just bought a house. We had one kid at the time and I came home and I was like, yeah, I'm like, this is a really like, that's kind of an interesting side business.

My wife's a stay at home mom. She has been our whole, our whole marriage. Um, so we were one income at the time. I mean, we weren't destitute, but we weren't rich for buying, you know, typical, typical, straight out of college, you know, young kid. 

Whitney Lee: And so we looked around to 

Phil Andrews: see, what's that? 

Whitney Lee: Character building.

Phil Andrews: Yeah, exactly. Uh, we looked around to see kind of if there were any laundromats around, we didn't find any that were for sale. There was one like an hour away. We're like, yeah, we can't manage that. Two weeks later, the laundromat that was two miles from our house had a for sale sign on the window. 

Whitney Lee: Ah. Like, 

Phil Andrews: what?

That seems like a sign. So we call them up. Yeah, a 

Whitney Lee: literal integrated sign. Literally a sign . Uh, it was just a 

Phil Andrews: small store, you know, it was pretty easy to maintain. My wife could go over there and help out with it. Um, you know, 'cause it was only two miles from our house and so we made an offer and, you know, I cashed out what little bit I had in my 401k at that point in time.

It was like 15 grand. Um, that was our down payment. We got a commercial property loan. We own the building, all the equipment inside of it and a four car parking spot out front, like four car parking lot. So it was very tiny. Um, and we bought it in 2012. Yeah. 2012. Uh, we ran that one for, that was the first one we had.

We ran it for 10 years, uh, but it was about 20, 000 a year of extra income to the family. 

Whitney Lee: Do you think that experience informed your software career? 

Phil Andrews: It definitely helped me with business sense. Um, you know, as far as a lot of engineers don't think about how their actions are going to apply and impact the overall business from a higher level.

Uh, and so that was one advantage that I had. I, I kind of moved into management and that was one of the things that I think was a good advantage there was being able to understand how your work impacted the larger business. Um, and which is, I think it was also helpful with, um, My role at Cast AI since I've been here is being able to help DevOps engineers and managers of DevOps, uh, platform engineers understand how this impacts their overall business.

We actually had a DevOps team for one of our customers get pulled up on stage at an all hands meeting and celebrated because they saved the company. Not even kidding. 

Like how many 

Phil Andrews: DevOps teams have you ever heard of getting celebrated at an all hands meeting because they saved the company from bankruptcy?

Like that was amazing that their impact, you know, by cutting their, they cut, um, cost of goods sold by about 75 percent over the course of eight months. 

Coté New: Yeah. Sometimes people really do like the new dryers. 

Phil Andrews: Yeah. Yeah, exactly. That was a case where, you know, the CEO literally brought them up on stage and celebrate them.

It was fantastic. You know, it's like, 

Coté New: like, like, like with that, you know, with, with all these people you talk with, you know, it's making me also think I've, I've, I've had conversations with people a little bit where recently where. You know, there's the notion that, uh, silos should be broken down and we should DevOps with each other more and things like that, but in, in some of the, you know, listening to the, the talk that you had and some of the conversation we've been having here, it seems like.

There actually is a lot of good utility to silos, right? Like, you know, as you're saying, there's, there's the beleaguered DevOps person who's expected to do everything. I'm sure the developers think they're expected to do everything and on and on and on. Right. But there's some kind of like, uh, like, I don't quite know where the line is, but, but I wonder what you see out there, like how, how far do you turn the knob on silos and, and it's still a good idea versus like, if we try to lower the silos at some point, It's not good, right?

Like we don't have enough division of responsibility and labor crossed by expertise. Like, how are people organizing their silos nowadays that, that works, uh, and doesn't work? 

Phil Andrews: Yeah. And so the extremes are never good, right? When you've got. Extreme cross functional meshing. You find that you never get anything done.

Um, and it looks like my camera froze. Let me try fixing it. Um, on the other hand, you've got silos at some of the bigger companies that are so incredibly locked that they're I've had DevOps engineers that were terrified, absolutely terrified about making a change. Because they didn't want the developers coming and yelling at them.

And so on that side, the silos were so strong. There was no cross silo communication. It was just all or nothing. And the ones that we found work best are the ones where, you know, the managers of, you know, say DevOps and the development engineering teams, uh, were able to actually work together and they would, um, come up with a plan to.

Collaborate, you know, they'd have a meeting once a week and they talk about, Hey, what do you need from the devs this time? You know, are you having any problems? And so they wouldn't necessarily be discussing every little minute decision. You know, for instance, the decision to bycast AI or by, you know, some other similar tool, the DevOps team would be empowered to make that by themselves, but they would want to let the development, the team know what was happening.

It'd be like, Hey, by the way, we're making these changes. If you see anything or hear anything bad, let us know, but otherwise we're going forward with it. Right, so then they're empowered to actually make that change. Sorry, go ahead. 

Whitney Lee: Have you ever seen any of those really strong silos that Get broken down.

And if so, like what strategies work for that? 

Phil Andrews: Uh, yes. Uh, not as much in my role here, but when I was at Oracle, uh, that was one of the things that we actually did at Oracle before I left. Uh, we were building a company wide threat intelligence platform, uh, at Oracle, any two divisions of Oracle working together is absolutely unheard of.

We got eight different security teams. From all the way across Oracle, able to collaborate and willing and commit to piping all their data to a central platform, doing the analysis, doing the threat hunting in a central platform, adding threats to the, to a central database, and then being able to leverage that across the entire, you know, company, 140, 000 person company.

Um, that was. Earth shattering for a company of that size to be able to get those silos broken down. I mean, those security, everybody, every security team in the world wants to like protect their little fiefdom. Uh, we need everybody to collaborate and decide on a decision forward. Uh, that was incredible.

You know, and that, that just, it took a year. 

Whitney Lee: How did, like, what was the first step? What's step one? 

Phil Andrews: Getting internal buy in from like, at least just somebody, you know, like we, we got exactly the sponsorship. Getting some champion, right? Getting some champion, you know, at a high enough level that they can at least get you a voice, you know, we, we were able to get, um, the chief of security within Oracle cloud.

And he asked for some changes and he asked for some architectural changes and he asked for some kind of interface changes. But he said, yep, you know, I think this is what we need as a company and this is what our competitors are doing. He's like, we need to get on board. We need to modernize. After that, it was meeting after meeting, discussion after discussion, who gets what.

Everybody wanted to own a little bit here and there, you know, typical politics, right? It's the same as anywhere else. You know, everybody wanted to feel like they were getting a win out of it. 

Whitney Lee: Yeah. You said you've, you said you've. Talked to over a thousand customers. Is there a time you walked into a place and like looked around and you were just like, no, not doing this and just walked back out?

Phil Andrews: Not quite that bad. Um, but I've, but I've had some new business calls where I got to the end of it. And I'm just like. I'm like, we could save them a lot of money. There's no way they're going to sign on with us as a customer. Like no matter what, and a lot of times it was because they, you know, a lot of times it was people with a very 1990s data center mentality who are now trying to run cloud native services and, you know, they've got walled gardens and they've got, you know, and it'll be.

It'll be for like a cloud native SaaS service. 

Whitney Lee: What's a wild garden? 

Phil Andrews: A walled garden, sorry. Like a walled garden. Like they um, 

Whitney Lee: they're completely isolated off 

Phil Andrews: from the internet. Like they don't allow any data ingress or outside. They're not allowed to use any SaaS services. You know, in a lot, at some of these times it'll be a cloud native company.

It'll be a five, six year old cloud native company, but they built their whole system. As if it were an isolated data center back in the late nineties, early two thousands. Yeah. And they're like, well, can we self host your system? We're like, yeah, it's going to cost you about 10 grand a month because that's the minimum size of like our whole setup.

If you want to self host it and not use the SAS platform. Well, our whole infrastructure only costs $6,000 a month. And we're like, that's our point. . 

Coté New: Right, right, right. Yeah, it, it's, I mean, that, that, it's funny is the word to use. Not that it's literally funny, but it is that funny. it, it, it is, yeah. It, it is frustrating when you have a customer and they're like, I love what you're doing there, but I wanna run it on premise.

And then as you're saying, part of you're like, but part of the magic is that. Doesn't run on premise, right? Like whatever, like cost structures we get from it, or you don't have to worry about managing it, like that's, you're going to lose that value. And yet people will be so, um, uh, you know, they'll want to work in as I, as I, I mean, it was a mishearing, but I think we should start using the term wild garden, because that's, 

that's 

Coté New: what you have when there there's a walled garden that you just let it run wild and you 

end up 

Coté New: with a wild 

garden.

Coté New: Don't worry 

about it. Yeah. So, so you, 

Coté New: you, uh, you mentioned working at Oracle and, and, uh, Did they, did they do a stack ranking while you were there? They did. Yeah. So, so this is, this is, uh, in our show, we like to have a little moment called, uh, free therapy. And I think if, uh, if, if I stay at my current employer in about a year, I think I'll be living under stack ranking.

I think I used to be in stack ranking way back when, but I was young enough, I didn't know what was happening. But like, so how do you, like, everything I've ever read is that stack ranking is horrible. Uh, and so like, how do you, 

Whitney Lee: what's stack ranking, please? Oh, 

Coté New: stack ranking is basically, well, why don't you tell us what, since you've lived through it and we're cognizant of it, what, what is stack ranking and how do you thrive and, or survive it?

Phil Andrews: So what I've seen was, you know, you've got a hundred employees, you break them down into, you know, your top, 10, 15, or 20 percent in your bottom 10, percent the bottom 10, 15 or 20%, you know, are basically on the hook for being managed out the top 10, 15, 20 percent are getting 80 to 90 percent of the reward at, you know, promotion and, you know, raise time and stock time and bonus time and all that good stuff.

And then that middle, you know, so. 60 to 70 percent, you know, might get a little trickle down. They might get a little something here and there, but generally they're in like that three category, right there. That meets expectations. You know, you're doing fine. Just keep going. But we're not going to give you a raise.

Yeah. They're the ones who get the steak knives. 

Yeah, 

Phil Andrews: that was how I experienced it. I don't know about you Koday. 

Coté New: No, well, I, I way back when. Yeah. I mean, that's the explanation. I think, I think, I think the. The, the thing about stack ranking, the, the, whatever, the thing that can be taken to be vicious is you're, you're not allowed to give everyone an A, right?

It's like, yeah, no matter what it's grading on a curve is what we, you're creating 

Whitney Lee: an us versus them. You're creating competition within a team. And so you're like, you're 

Coté New: like, even if you love each of your a hundred kids, you must kill five of them. So like, you know, like you're, even if they're doing a great job.

And so that's, I think that's where like a lot of the, especially from places like the DevOps world, a lot of, you know, just intuitively, as you just quickly intuited Whitney, you're just sort of like, Oh, right. Good job teamwork. Right. But, and yet it does seem to persist as like a management method all over the place.

So there must be something about it that like is. Cause you know, companies generally don't do stupid stuff that makes them like lose talent, that makes it harder for them to like make money. So I don't know. Like what, how does it 

Whitney Lee: work? In a way that community isn't necessarily, so even though it might be inferior, it's inferior with, with metrics.

Coté New: Yeah. Yeah. Huh? Yeah. So, so what, how do you, uh, yeah, I'm going to write that down inferior. How, how, how does, uh, you know, you, you work there like seven or so years. Like how do you, uh, How do you incorporate that into your professional life that you've got that running around? 

Phil Andrews: It was really challenging. So when I first started out as a manager, I had a very small kind of almost tiger team of four engineers and they were all rock stars.

Like it was a small team. It was easy to manage for performance. We were. One of the fastest moving teams. We got things done very quickly. We had very high quality code. Um, but we were a tiny team. We were four people. So when you're told, well, you need to give one low X, you know, one that's underperforming one, five, and then everybody else should be, you know, either threes or fours.

You know, I've only got four people and they all do good work. I'm not going to give one of them a one that doesn't deserve it just so that I can meet. And if you push back, they would, you would, you would get. Uh, an exception, a waiver, we'll call it. Now, when I had a team of 50 people across three different teams, three different engineering orgs, it was relatively easy there.

There's always one, two, three people across the orgs that are, Sandbagging, right? You know, you can tell that when they're, when they're not on a call, they're walking their dog, or they're at the park, or they went to lunch with their, you know, parent, or there's always an excuse as to why they didn't get something done.

They're never meeting their story points on their, on their tasks. Their tickets never get done. And you're like, okay. When I've got 50 people, it's a little easier, like you can find one or two that aren't doing the job. 

Coté New: Right, right. I wonder, I wonder, I mean, one theory would be that in organizations where there's stack ranking, uh, what's the phrase?

You, you have, uh, you have a lot of direct reports, right? Like you, you're, you're, is it called your spans? Like, you know, like you're saying, if you have 50 people that you need to put on a curve, that's a lot more palatable than five. Yes. Right. And, and so like. That would, if you have an organization where there's more people per each manager, then maybe stack ranking doesn't smell so bad.

Right. And, and also like you're saying, I don't know, every now and then you can find someone who needs a little, uh, motivation, maybe to be more productive or not. 

Whitney Lee: Do you stack rank your 30 pigs? 

Coté New: Kind of. 

Phil Andrews: Yes. 

Whitney Lee: Yeah. There are a couple 

Phil Andrews: who are like higher up and then there's a couple who, you know, there are definitely not my friends.

So lately my, my, my youngest piglets, they're literally about this tall. They have learned that they can squeeze out through the electric fence. If they like zip through it without getting 

Whitney Lee: caught 

Phil Andrews: and they have. Absolutely and completely destroyed about a third of my backyard is completely rooted up and completely tilled up about a third of my yard because they'll escape during the day.

And I'll be in meetings like this and there'll be out there like destroying my yard. So they're, they're pretty low. 

Whitney Lee: They're all gonna get eaten eventually, right? That's, that's the only reason a pig exists, right? 

Phil Andrews: Most of them, yeah. I mean, some of them are breeders and they, they stay around. 

Whitney Lee: Oh, so as a pig, do you want to be a big performer?

Because then you're just going to get eaten faster. 

Phil Andrews: You probably want to stay on my good side at 

Whitney Lee: least. Do you eat your over performers or your under performers? 

Phil Andrews: Both in the fullness of time. 

Coté New: That's a case where you want to be in the middle of the curve. 

Right, 

Coté New: because you're, I would imagine you look at him, you're like, Not, not quite tasty enough yet.

You're not over performing, but you're also tolerable. So I don't, I want you to not have to put up with you. Not being 

Phil Andrews: obnoxious as a pig is a big, is a big winning factor for longevity.

Whitney Lee: When you go for the next one to eat, do you go for the one whose personality you're ready to get rid of? 

Phil Andrews: Uh, there's one that's, that's pretty high on that list right now.

Coté New: That one's on a performance review. Yeah.

Well, well, uh, well, this has been, uh, delightful. What if, uh, it has, if, if people wanted to, uh, well, we've mentioned cast AI several times. I know that's just cast ai, but, but how about yourself? Where, where do you hang out, uh, on the internet? Like where, uh, where can people look up interesting stuff that, uh, that you're doing?

Phil Andrews: Unfortunately, I don't internet very much. Like I'm, I'm a, uh, was a Luddite technologists. I, uh, last week I was up in the backwoods of Maine with no internet connectivity for seven days, like no cell phone coverage, no nothing for seven days, living in a cabin with an outhouse on the woods. Uh, so that was, you know, that's how I spend my recreation time.

Um, you know, I have my LinkedIn, but that's, That's pretty much the only social platform I actually, you know, 

Coté New: there's a surprising amount of people who that's the only thing they have. The, uh, web logs, not so popular anymore or websites, the old square space. The 

Phil Andrews: longer I have spent in technology, the less I want to use technology.

Unfortunately, 

like, like it's one of those, when you, 

Phil Andrews: when you know how the sausage is made, you know, not a big joke. Um, yeah, when you, when you know how the entire internet is held together by, you know, bubblegum um, 

Coté New: you're 

Phil Andrews: like, yeah. I'm going to go out in the world and do things in the real world. 

Coté New: Yeah.

Yeah. Yeah. You know, like, especially with a lot of things we're talking about it, when I think about the, uh, one way of putting it is all the opportunity out there, uh, for, for people like yourself and other people who work in this space, another thing is like, it's the same goddamn thing over and over again, and, and it makes me think like, wow, how do like nurses and doctors put up with this?

Cause I imagine, I imagine there's some every now and then, like we have, we have a friend. Uh, a couple of the both doctors and one of them is a, um, what's, what's the fancy word for a butt doctor? I forget, but like, yeah, like he's got interesting stories.

If you work in the ER as the proctologist, a lot of what you're doing is, So, to see something from an arm, to see it from a leg is, is, is removing things. And so every time we see him, he's got some variability, but you know, day to day is a lot of the repetitive stuff. Carla Enns 

Whitney Lee: This story I imagine has the same format.

You almost just have to say the object name. And that's all the story you need. Chris 

Coté New: Tat Maybe next time we'll see him, we'll be like, yeah Just the object name. Just give us a list. Huh? Yeah. Stack rank. I want a stack rank list. Yeah. Yeah. But, but you know, I think, I think that does cause some, some people to be like, Oh yeah, it's a, unless you go chase some other domain of the it world.

Uh, it's, it's nice to, uh, be in the cabin in Maine sometime. Focus on that. Well, speaking of being in cabins in Maine. Uh, you've been listening to another software defined interview. Hopefully not in a cabin in Maine, uh, where there's no internet. Maybe you brought it on cassette tape. You downloaded 

Whitney Lee: it beforehand.

Oh, I like your cassette. Maybe we should work on that, 

Coté New: Whitney, as we might release a back catalog on cassette tape. We'll find some sponsors. Who can sponsor, uh, each season of Cassette Tapes and we'll email it out like those field note boxes of, uh, special, special things. 

Whitney Lee: Can we have it like time life books where you get like a subscription?

You get one every, every 

Coté New: four weeks. 

Whitney Lee: Put together an infomercial. Yeah. Uh huh. 

Coté New: Those books. Call a 1 

Whitney Lee: 800 number to order. 

Coté New: Yeah, 6 8 week delivery. 

Whitney Lee: Columbia Hospital. Heck yeah. 

Coté New: No CODs. Whatever that 

Whitney Lee: means. Cash on delivery. 

Coté New: Uh, so yes, if you go to softwaredefinedinterviews. com Uh, slash Whitney, what number is it?

Whitney Lee: 85. 

Coté New: 85. You can find links to what we were talking about, uh, uh, and, uh, and, uh, check it out and see how to subscribe. And with that, thanks again for being on. It was, it was fun and we'll see everyone next time. Bye bye. 

Whitney Lee: Bye.