Ep#118 ProsperOps and the Rise of FinOps AI with Erik Carlin

Host: Jon

Today's topic is ProsperOps and the rise of FinOps AI. Is this like Terminator and Skynet or something better? I don't know, but our guest today is Erik Carlin, co-founder and chief product officer of ProsperOps. It's a FinOps SaaS platform that uses AI and yes, true AI optimization algorithms and a real-time execution engine to deliver maximum hands-free savings outcomes. Before that, Erik was the vice president at Rackspace where he started to manage a w s business and grew it to one of the world's largest AW w s Premiere partners and a leader in Gardner's Magic Quadrant. Erik led the product and engineering teams for all of the Rackspace-managed public clouds, including AWS, Azure, and GCP. He has extensive experience and hundreds of complex X Scale customers optimizing their cost on the cloud. Please join me in welcoming Erik to the show. Erik, thanks for joining me.

Guest: Erik

Hey Jon, great to be on. It's a pleasure.

Host: Jon

So, Erik, you're in Texas and I have to jump to this right away because I was just out there recently and I did not get a chance to go back to some good old-fashioned brisket. Man. Let's talk barbecue real quick because that's near and dear to my heart compared to cost savings.

Guest: Erik

Yeah, you missed it. Texas is the capital of barbecue and Austin in particular, we were chatting about this earlier, if you go to the website masterclass and you look out for the master that they've selected that can create the best brisket in the world, the class is taught by Aaron Franklin who owns Franklin's Barbecue in Austin and so it's quite an experience there where you people in Austin line up and stand in line all day. It's almost sort of a team-building exercise for some organizations just to wait to get to taste, to have just a taste of Franklin's barbecue. So lots of good brisket in Austin. Next time you're down, we'll take you out to enjoy it.

Host: Jon

The last time I was in Austin, now I don't remember the name of the place, but I was at Keller. They had one called Heart eight and it was like, first of all, when you line up and you pick the meat that you want, I was like, oh, this beef rib, I want this nice one. And I'm like one, I want a couple of those. Yeah, I think I could only eat one and I'm glad I got one at the time. It just melts in your mouth. The food. Yeah.

Guest: Erik

If you've never had good Texas brisket, you don't know what you're missing. A good brisket will just kind of melt, melt in your mouth. It's incredible.

Host: Jon

Erik, this is making me hungry. I'll be right back. I'm coming the next time I'm there in Austin. I got to take some time to just go to wanting to, I would like to go to Franklin's, but I don't think we have all day to hang out and just wait to get in there. I heard you, you got to wait in line and they take your mention and they make sure they have enough for you, and then best come, best serve afterward.

Guest: Erik

Yep. Yeah, I think Franklin's is an experience if you have sort of all day, if you don't have all day, my recommendation for folks looking for brisket in Austin is to go to Terry Black's incredible. I always describe it's almost like product development. There's kind of like an asymptotic as you approach perfection, it's sort of asymptotic and Franklin's is kind of as close as you can get to perfect barbecue, but Terry Blacks are maybe one notch back without having to stand in line and it's delicious. Our CTO  is, he grew up in Austin, he's a barbecue connoisseur and I think it was in high school or somewhere earlier in his life, he built the e-commerce system for the black family to ship barbecue all around the country. So it's my own experience and also we've kind of got a connection to the blacks just through other co-founders. So highly recommend Terry Blacks if you want to, look for a good barbecue place in Austin.

Host: Jon

Okay, Erik, while this topic is not around, the barbecue in Texas is important and maybe we should have a podcast about that and the cost savings around the great barbecue.

Guest: Erik

We'll get our CTO on to do the barbecue podcast. He can do it much more justice than I can.

Host: Jon

All right, that sounds good. So Erik, today we're talking about ProsperOps and Arise of FinOps and ai. Erik, I got to tell you, it feels like props are, ops have been around a lot longer than 2018 and one would say that you're a rising star in the FinOps space. What's your story and how did it happen?

Guest: Erik

Yeah, well first of all, it's always humbling to me when people talk about ProsperOps and that context as a rising star and even knowing the name, I can remember sitting around debating names, working with our initial design team to come up with logos and colors and it's just very humbling as an entrepreneur for people to kind of know who you are and what you do. And so it's really exciting to hear that our story is an interesting one. As you mentioned in my intro, there are three founders to cross Robs, we all came from Rackspace and the three co-founders started the managed AW w s business at Rackspace. As Rackspace was changing its strategy to also embrace third-party public clouds beyond just Rackspace's infrastructure. And I kind of led the product and engineering team and we built a lot of tooling. We integrated third-party tooling.

Guest: Erik

I have a lot of experience with what I'll just refer to as kind of 1.0 like cloud cost tooling, cmp, that sort of stuff. And we were sort of taking arguably the best-of-breed tools and we were combining them with FinOps. Even discipline, this isn't like the 2016 timeframe, 2017, it wasn't even really a discipline, but we were taking very capable cloud engineers, marrying them with this tooling. And we were getting just really what I would consider being suboptimal outcomes. It was very frustrating. And so we took a step back and said, what are we missing here? How do we create very consistent, very high-quality sort of outsized outcomes for customers? We were generally operating in two areas where people needed help with security and cost optimization. Those are generally the top two areas. We really sort of focused on cost optimization and said how do we go and reset and rethink this?

Guest: Erik

And what we realized is, you know, really have to remove the human from the equation to get outsized outcomes. It's not that humans aren't smart and capable, that it's not the case at all. It's just that the cloud has become so dynamic and there is so much change and there's so much work to do that what you need is a system that is continually watching and executing very advanced sort of strategies to create outcomes. And so that was the impetus for ProsperOps how do we take and provide Auton, we call autonomous cloud savings, how do we automate as much as we can to not just deliver sort of reports and recommendations but to deliver actual savings outcomes for our customers? And that was kind of the genesis. And so we started ProsperOps in 2018 and I've just been growing the company since. So we took kind of an interesting path in terms of a tech startup where we took a seed round, we built a high growth profitable company and then we just announced a big investment round not too long ago. So that's kind of our story.

Host: Jon

Congrats on the investment round. I saw this was probably about three weeks ago, four weeks ago, I saw the investment round happen during this conversation here you've kind of indicated that you're removing the human aspect of it and I think here's my perspective on removing the human aspect of it. While humans, built the system to do it, we can't continuously monitor and make these advanced changes all the time. You would have to have an army of people sitting there around the clock and by the time you make this decision the cost is already missed or you can't place it in time or something that expired. And I think that's where the value of having AI is one of the things that set you guys apart, but the whole automated portion has me very interested and intrigued because doing this automatically, a lot of companies say, I can do this automatically. Are you able to do this autonomously without the need for intervention and without humans behind it making these decisions?

Guest: Erik

Yeah, I think it's a great question Jon and I, there are a lot of solutions that involve sort of outsourcing your FinOps and just to be a bit more specific about what we're focused on here when we started the company we said we want to automate fops for our customers. And Fops is kind of a broad domain as you know. And generally, if you kind of read the FinOps book that the FinOps foundation puts out, it's broken in into two categories. You have rate optimization which says to pay less for the stuff that you use. And then you have usage optimization which says just use less stuff. And these are really at a high level the two categories of ways in which you can optimize your costs. Where we started and debated different avenues, but where we started on the rate optimization side of things.

Guest: Erik

So that's kind of the world of our eyes and savings plans and there's a lot of options for customers to either manage that in-house or to outsource it, but generally, when you're outsourcing it, you're outsourcing it also to another human and that could be a function of the sort of skillsets or bandwidth. And I think the difference that you're pointing out with sort of AI and automation is that this is not about ops having a more capable ops army or staff or an engine room of people. Our mission and what we've built is a platform that can ingest state changes as they're happening in your cloud environment in real-time, and I pulled some stats for this just to be able to communicate some actual numbers, but our platform is ingesting and this number grows every single month, about 20 million state change notifications every single month. So every time you know, start an instance, stop an instance, make some sort of the change in your environment that is effectively creating an event. Our platform is where we ingest that and we're constantly checkpointing all of these state changes and that creates an opportunity to then evaluate whether or not the change in your computing environment would necessitate some sort of change to optimize your I and savings plan portfolio to improve your savings.

Guest: Erik

And so we're running about 650,000 Athena quarries a month at this point to look not only at this real-time telemetry data that we're collecting but also sort of curve-based lagging data in our platforms, blending that, it's creating a picture of what your environment looks like every single hour we checkpoint your environment and then we have several AI tool sets and algorithms that are looking to understand is there an opportunity to optimize this environment on any given month? We're making almost a quarter of a billion calls to AWS API calls and it turns out that if you look at all of the optimization activities that happen like an aggregator across all our customers, we're talking about 850,000, I'm sorry, 85,000 of those per month, which works out to between one and one to two actions per minute. So every single minute since we started this podcast we're 15 minutes in our platform's taken 20 to 25 actions across our customer base to optimize their environment. And so we're also employing very advanced sort of strategies that no human would execute. Sometimes we're orchestrating complex transactions, and this isn't just a matter of purchasing this commitment, it's about to expire, please renew it for me. We're talking about very advanced strategies. We're talking about sort of real-time monitoring complex sort of orchestrations of these commit discount instruments in ways that can increase savings as well as kind of decrease risk for our customers.

Host: Jon

Erik, help me understand ProsperOps a little deeper. I know we jumped right into the AI portion of it and one of the questions around ProsperOps is whether you guys handle our RI’s conventional standard, all of the above.

Guest: Erik

So we started with just AWS and we still survey a WS today, although we're working towards expanding to other platforms. We started initially when we launched the platform in early 2019, it was just convertible to our eyes. Then when savings plans were launched we blended savings plans into the portfolio. And then just recently we launched a new feature which we can talk about if you want on this podcast Flex Boosts. And Flex Boosts blends standard ROIs into the portfolio as well. So in the same way that you might deploy in your personal investment portfolio, you might deploy a blend of equities or various securities in ways that each has kind of a different risk-benefit profile and you're sort of creating a mix based on being able to achieve a certain savings outcome. Our platform is now taking advantage of all of those different instrument types and blending them for our customers to create a maximum savings outcome. So it's all of the above rights now.

Host: Jon

Nice. Now just for the audience, a little backstory is that our I savings plans are conventional everything is a billing construct and ProsperOps is handling that. They don't touch your instances, it's just the billing portion of it, but the cool part about it is that it's doing this all automatically behind the scenes for you for the best cost savings. Now if you try to inject a human into the aspect of it, the only thing that I see and that it's a concern is that you're missing out on a lot of potentials and you can't see, you're only focused on what's right in front of you versus the broad spectrum of everything. Diving into it a little more, we're going to talk about Flex Boost in just a little bit, but I want to talk about some of the ROIs and the capabilities that you're able to do. And when you say automated and looking at it, are you looking at this customer account? Are you looking at a broad spectrum of all the customer accounts and what's happening in an environment taking into consideration RI’s availability within this region, within this availability zone, and the cost differential between everything?

Guest: Erik

So ROIs and savings plans generally operate in the context of an AWS organization. This is within an AWS now. And so we have customers that in some cases have thousands of AWS accounts that all roll up to a single organization. And so we're ingesting all of this telemetry again, both lagging in real-time, our platforms blending it, and then we are effectively normalizing and sort of rationalizing what that looks like in aggregate across the entire organization. And then our optimization algorithms in our AI are looking at data kind of in the context of an org. Our platform is also multi-organ in the sense that we also have customers that support, that have multiple organizations, but each of those is sort of managed at least right now on an independent. And I say right now because this is the way that Amazon rules are, that could change in the future, but right now things are looked at on an org basis.

Guest: Erik

We do have console views in our console that will aggregate data across organizations in terms of how we present it to customers so they can understand the total number of actions that our platform took on their behalf across all their organs or their total savings or their total effective savings rate, these sorts of things. But generally speaking, it's happening on a per-org basis. The exception to that, which we'll talk about in a little bit is Flex Boosts where that's a unique product where we can leverage the scale of ProsperOps. We have, and we're approaching almost a billion dollars of annual computing usage under management. And so at that scale, we can do some really interesting things where we not only look at an individual organization, but we look at our total customer base in totality and we can do some more interesting things by looking at aggregate.

Host: Jon

When you say more interesting things, are you able to say, I know all customer data is very secure and confidential and you're just looking at the ROIs in general? So you say Customer A is utilizing this type but they don't need that and customer B could utilize that and I can sell to either of them and kind of trade back and forth to get the most cost savings effective for my customers.

Guest: Erik

Yeah, that's right. And just to reiterate your point, you said earlier, one of the reasons we started with rate optimization is because we are operating in a world of abstract billing constructs. Our platform doesn't need an agent, it doesn't need access to customer data. We don't have access to even the control plane, any sort of right operations to the control plane I e we can't start an instance, or stop an instance. We have access to cost and usage metadata if you will, and we have a lot of customers that operate in sort of regulated secure environments. We have FinTech, we support gov cloud, and we have other regulated sort of sectors. We've passed a lot of very high sort of security scrutiny in terms of the access our platform needs. So the data we have is actually sort of what I will refer to as almost usage cost and usage billing metadata.

Guest: Erik

But our platform is sort of again blending telemetry and creating a picture of the environment where we can do the current AWS is lagging generally let's call it a day. And so our platform can create a real-time view of what your environment is doing right now. And this is something else that when we talk about just computers and algorithms being able to do things, this is just one example where it's ingesting all this telemetry, these 20 million state change notifications and it's creating a picture of what the environment looks like right now. And so when we're taking action for a customer to optimize it, it's not based on what the environment looked like from the curve that's delayed by a day or more. We're talking about what happened in the past hour. And then coming back to your question with Flex Boosto in particular, Amazon has an ROI marketplace where when you make a standard RI commitment that's a standard RS are very rigid commitment, they're inflexible.

Guest: Erik

Once you commit, you can't exchange it, you can't do anything to sort of change it. But the one thing that you can do is you can make it available in the marketplace for somebody else to take over that commitment. So if you're running, I don't know, an R five and you buy an R five standard RI and you no longer need that because you switch to an M five or an M six G or whatever, you can list that RI in the marketplace for somebody else to effectively take that commitment. And so what we've done with boost is we've said the general challenge with the marketplace has to do with RI liquidity and kind of counterparty risk. So if by standard I and you list it, you have no kind of assurance that someone else is out there that's going to be able to take over that commitment.

Guest: Erik

So you make that commitment and it's a rigid commitment and you're stuck with it if you can't dispose of it on the marketplace if you don't need it. But how do you know if that's something that someone else might take from you if you don't need it? And so again, because we now have almost a billion dollars of annual computer management, what Flex Boosts is doing is it is strategically and intelligently layering in standard ROIs into our customers' environments, specifically in regions and instance types where there is sort of a mass of those in use across our customer base and we have various algorithms that de-risk and shard across our customers so that we don't. So that commitment is kind of spread, but what that means is that if a customer's usage changes at any time, and by the way, let me point out to you, we don't require any communication with our customers about what their usage patterns are going to look like in the future.

Guest: Erik

Often if someone is doing this by hand, they're talking to the teams doing capacity planning, trying to figure out whether are we going to increase our usage, decrease our usage, et cetera. That all goes away with proper props. Our algorithms can just watch what happens in real time and then just change the environment to match. And so when we built Flex Boost, we wanted to continue with that low context sort of autonomous experience for our customers. And so basically the way it works is our platform is automatically deploying standard RI commitments and if a customer's usage goes away, our platform sees that and we can basically what's kind of referred to as a market maker where in the same way you might have, if you sell your house, you might have an agent that brings the buyer as well and so they'll be able to sort of be a market maker because they can sort of connect the seller and the buyer as one agent.

Guest: Erik

And that's effectively what ProsperOps does, where before we even list that particular commitment on the marketplace, we already know who the buyer is and so we can list it and then in the account of our other customer we can buy that particular commitment. And so through the marketplace and then looking at our customer's usage in totality, we can sort of shift commitment between our customers safely. And this is another advantage again of just sort of an algorithmic approach and AI approach where this, it's doing things that just a human at an individual company couldn't do or just most of the time wouldn't have the time to do. I want to reiterate the point, there are so many super capable FinOps practitioners and teams out there in our view, this is a function of just certain things being so complex that it's only achievable through algorithms and sort of real-time execution engines. And the cloud is just getting so dynamic, so complex and the world of fops is just getting broader and broader and broader. And so there are so many things to do on a fops practitioner's list. Our mission is just to take more and more work kind of off their list, automate it, give them outcomes that exceed what they could do on their own, and free them up to go focus on more high-value kind of fops activities.

Host: Jon

Erik, I think you guys have advanced insights for reselling from one customer to another. You already have the buyer lined up so then you're going to sell it and quickly pick it back up and exchange it between customers you indicated for you're not trying and FinOps teams are very capable of doing this. And when you have a small environment, 10, maybe 20, 25 instances managing for it, it's not bad. It's easy to do, but environments are growing, accounts are growing, and organizations are growing and the key here is how you manage all of them and see and have an insight into all of those and the sale and the EC2 instances, then that's gone around the RIs that you need to utilize what's being used, what's not being used, what's being wasted. And using ProsperOps allows you to take that off your plate, allows automation, and does not worry about that. But still using the ops culture of having the reports and how things are going and being able to evaluate the changes is key. The other thing I want to touch on is that you don't have an agent, don't have to worry about that very limited permissions. You only need to see what the curve file is to make and have these rights and permissions to purchase and sell our I on our behalf, correct?

Guest: Erik

It's the curve file, it's permissions to execute on these RI’s and savings plans. And then there are a few lightweight read-only permissions that give us access to that real-time telemetry that I was talking about. And so our platform is blending real-time telemetry with lagging data and then the AI figures out what to do and then we submit it to our execution engine, which then goes and orchestrates in some cases very complex transactions to make that plan of reality.

Host: Jon

Nice. Erik, I want to jump over to savings plans a little bit and one of the questions I have for you is what is an effective savings rate and why is it important?

Guest: Erik

Yeah, so you know, we're mentioning this idea of in a smaller environment it's generally manageable. There are a lot of advanced strategies that our platform executes on behalf of our customers. Now with Flex Boost, we're bringing not just advanced strategies, but we're bringing the benefits of our scale to our customers. Where the larger ops grow, the larger the flex boost layer is and therefore the more savings our customers get. So you're getting advanced strategies, you're getting the benefit of our scale, but at the end of the day, the only thing that matters for customers, or I should say the chief thing that matters is are we doing as good? Are the results that our platform is generating for our customers, are they as good or better than what a customer could do on their own? And is very, I used kind of personal investing as an example earlier and we kind of like that as an analogy.

Guest: Erik

We kind of base the company off of in the personal investing space, you know have to say, which is like a DIY platform where you log in and they give you all sorts of reports and recommendations and you can research equities and decide when you want to buy and sell. And so we kind of view that the v1 CMP tooling, you can log in, there are tons of reports, there are tons of recommendations, but ultimately it's on you to decide what you want to do. Then you have kind of human advisors that can manage your money and you generally have to pay them a substantial fee because there are humans involved in doing that activity. They have expertise maybe more than you have, and they can generate higher returns. That would almost be outsourcing your I and savings plan management to say a managed service provider or something that where there's a human's going to sort of take over those activities.

Guest: Erik

What emerged in the personal investing space is this concept of a robo-advisor, which is algorithms ai, they're watching the markets in real-time and they, you know, define a set of constraints in their platform that how their algorithms execute on your behalf and manage your investments on a day-to-day basis. And that was kind of the model for ProsperOps in the early days. Our original tagline was the robo-advisor for cloud until we realized that no one knew what a robot advisor was, so we had to go redo the tagline. But conceptually it's the same thing where what our customers do is they define a set of constraints. There are a few settings in our platform, customers define those settings and then our platform will autonomously execute within those constraints very similar to a robo advisor. And the question with all of those options, do it yourself, human advisor robo-advisor is like who can generate the best outcomes in terms of ROI, right?

Guest: Erik

And so what we realized when we started ProsperOps is that there is no cloud FinOps, cloud savings metric that kind of quantifies how much you're saving. If people are familiar with our RIs and savings plans, they're probably familiar with concepts like coverage and utilization. How much of your usage is covered with the commitment of the commitment you've deployed? How much of it is matching and used versus being unused? And these are not unimportant metrics but they're kind of input metrics. They don't tell you at the end of the day, what was my savings rate. How much did I get off lists, right? You go to Amazon's website and they say you can get up to 73% off the list price. Well, that's like some super obscure random MRI that's not what people get. What do people get? And we realized this doesn't exist, that this metric does not exist in the world of ops and if we are going to create a platform that isn't just going to deliver recommendations, but's going to deliver an outcome, we need to have a definitive quantifiable way of customers measuring what was your savings rate before, what was your savings rate after ProsperOps, what's your savings rate if you want to consider option B or option C?

Guest: Erik

And so that's basically where the effective savings rate was born. If your listeners want to learn more, they can go to effective savings rate.com and we have a very simple microsite set up with some blogs that we've written and a Google sheet where you can calculate your own effective savings rate. But the key thing here is that this is not a ProsperOps, I mean we sort of created this metric but it's, but it doesn't favor ProsperOps. There's nothing special about it for us. It's a completely cloud-agnostic metric. You can calculate ESR on AWS, Azure, and GCP. And the formula is quite simple. All it says is the numerator is how much did you save off the list price, right? So if your on-demand rate was a dollar and you saved 30 cents, then your numerator was 30 cents. And the denominator is what would you have spent if you paid the on-demand rate?

Guest: Erik

And in this case, it's a dollar. So if you save 30 cents and you would've spent a dollar, then your effective savings rate is 30%. And so the concept of ESR is very simple In practice kind of getting all of that data can be a little bit tricky. And I mentioned there's a public Google sheet with commands you can execute, you just plug in the numbers and it sort of calculates what your ESR is. These are just like AW W C L I commands that you can execute but at the end of the day, you know have to kind of unwind. You've got utilization coverage, all of this complexity, you kind of need to peel that back and just get at this core data of how much did you save and what would you have spent if everything was running on demand.

Guest: Erik

But that is your effective savings rate. And so we have, what we do is whenever we customers interested in exploring our service, our process is to run a savings analysis. And what that savings analysis does is we take a very, very lightweight provision permission, even more, lightweight than our active discount management permissions. We ingest some cost explorer data, some current data, and what we'll do is we'll calculate your historical effective savings rate over the trailing 12 months, and then we will take our algorithms, we'll hold your usage data constant and we will play our algorithms out on your data and project forward in time and show you what can happen to your effective savings rate in the future. And we take a percentage of savings. Our model is not a percentage of spend. Another problem with sort of the V1 like cloud cost optimization tooling is that they take a percentage of spend and in that regard, their incentives are misaligned with you saving money because when you save money your cost goes down, which means their fee goes down.

Guest: Erik

And one of the things we wanted to create in this model is we want incentives aligned with our customers where they want us to save as much money as possible. And so by taking a percentage of savings, we don't make money unless our customers save money, they get the vast majority of it. But when we measure the effective savings rate after our service is in place, we do it, we include our charge in that. So in the savings in the numerator, we subtract our savings share from that. So it's kind of like your net savings over your on-demand equivalent amount. And so we know we've collected thousands of data points at this point and we know that generally on AW WS 75th percentile effective savings rate is somewhere in, let's call it like 25% range ish, somewhere in the 20. So Amazon says gets 70% off, which is not an untrue statement.

Guest: Erik

Some risks offer 70% off, but if you look at in practice with coverage, with utilization, with the discount, the term commits that people are willing to make all this sort of stuff. Generally speaking, the 75th percentile effective savings editors around 2020, that's called the mid-twenties. The platform net of our charge can take that above. If you can get above 40% effective savings rate, you're in the 98th percentile. We have multiple customers that are north of 45%, 46, 47, and 48%. And with Flex Boosto that we've just launched, we broke the gauge in our console, which is a high-quality problem. It maxed out at 50% and we've now upped that to 55%. So we expect as Flex Boost is, we have customers now actively turning it up. But as that rolls out that we're going to see customers again net of our charge that have effective savings rates kind of north of 50%.

Guest: Erik

And this is just the power of, again, humans are super capable, smart, brilliant, all these sorts of things. There are just some things that are so complex that if you can apply real-time telemetry automation, very intelligent algorithms, and very advanced strategies, you know can just create superior outcomes. And the wonderful thing about FinOps is that it's just quantifiable right there. There are lots of things in life that are subjective and so is this tool worth it? I'm paying X amount, I'm getting a productivity gain or I'm getting some kind of subjective benefit. The wonderful thing about Fops is it's quantifiable. And so you can know with our platform what your effective savings were before our service, and what your effective savings rate is after our service. And again, we measure that net of our charge. So we have a graph in our console that shows you this ESR trend and it shows your baseline and then your post-ProsperOps ESR I tell our customers, as long as that post-ProsperOps ESR line is above the baseline, you kind of in the money so to speak.

Guest: Erik

And the money flows in different ways. We have customers that buy through the AWS marketplace, not to be confused with their marketplace and all sorts of things, but at the end of the day when our customer's effective savings rate is higher with our service than prior, that is functionally equivalent to us writing you a check at the end of the day. So it's almost like, and we quantify this amount in our console, so we are a platform that does not consume the budget. If you want to go buy some tool, some DevOps tool or CI tool or whatever, these are all great tools, we pay for them, et cetera, but it's a consuming budget. What's unique about this idea of FinOps and AI and producing outcomes and our pricing model is that we actually will put money back into your budget because we are creating more savings than our saving share is.

Guest: Erik

Our saving share is a portion of it. And so at the end of the day, and this honestly confuses some, it's a different animal, a lot of procurement teams haven't seen this before, but what we are at the end of the day kind of writing our customers a check and if we're not writing our customers a check, they should fire us. They should turn us off. We make it very easy for our customers to see whether or not our platform is creating better outcomes than before. And we're very transparent. We show that kind of front and center and that's just a key part of our platform. So ESR in my opinion, is critical. As you see the rise of finos ai, you see the rise of more automation platforms, et cetera. How do you know whether A is better than B is better than C? Well, that's the same as again, going back to personal investing, let's look at the history of this platform or this particular investor or et cetera, see what kind of returns they've created, and that gives you a way to quantify the outcome that you can expect from a certain platform. So we view ESR, it's just useful even if you never use ProsperOps, everybody should know their ESR.

Guest: Erik

Just like anything you want to understand, you want to figure out what metrics kind of make sense and measure them, sort of compare that to kind of industry standards. And so I mentioned some of the data around ESR, how are you doing today in this area of cloud after people clouds generally the largest expense in a company and rate optimization I mentioned can save upwards of 40 or 50% off that bill. So it's very worthwhile for everybody to understand ESR, know where they're at, and then be able to use that to just assess what makes sense for them as a company. So we view it not only as helpful kind of in the rise of ops AI and customers needing to quantify between different options but just in general as a useful way to measure how well people are doing in the discipline of fops.

Host: Jon

Everybody, real quick, we're talking with Erik Carlin, co-founder, and chief product officer at ProsperOps now the topic today is ProsperOps and the rise of FinOps AI. Erik, I have two more questions for you before we got to wrap things up. The first question I'm sure is top of everybody's mind is once I implement ProsperOps, how soon before I realize my savings?

Guest: Erik

Savings start as soon as the system is implemented, and as soon as we turn the system on savings, incremental savings begin flowing immediately. There is generally kind of a step up in savings once the platforms are turned on and then as our strategies play out, as our platform learns, your environment and adapts, generally what happens is there's a step up in effective savings rate and then there's an increase in SA savings rate over time until you know end up, again it depends on your environment, there are lots of factors involved in terms of regions and platforms and cyclicality, et cetera. But generally speaking, our platform's going to get our customer's effective savings rate over time to basically as high as it can be for your particular environment. So step up in savings initially so you get that immediate savings benefit and then that just generally grows with time. And so you kind of asymptotically approach the theoretical maximum of your environment.

Host: Jon

Is it, let's put some timing around it. Is it like I'm going to see within the console my effective savings Ray where I can make a change or inaction and I can enable this in a couple of hours in a day? What's your recommendation? I wait a little bit of time so the proper ops can learn my environment and then enable these changes.

Guest: Erik

No, I think this is something else that I bring up. Great point Jon. There's sort of this conventional thinking that says I want to optimize my environment and I'm going to do some right sizing, I'm going to do, I've got these old instances that I need to upgrade to latest gen, I've got this legacy application, I'm going to turn it off at some point. So I'm not going to kind of enable ProsperOps yet, right? I want to first go do this sort of optimization activity and then I'm going to turn ProsperOps on and that's generally conventional thinking and I think in that 1.0 world that is necessary. I think another benefit of AI automation, ProsperOps, real-time watching your environment, and being able to do advanced strategies to both increases as well as decrease your commitment is you no longer need to do that.

Guest: Erik

You just turn to ProsperOps to point us at your environment, it's going to generate incremental savings and you don't need to perfect your environment before you do that. If you want to change out your instance types, go for it. If you want to decommission that legacy application, go for it. Whatever you need to do in terms of continually right sizing and cleaning up your environment, rearchitecting, et cetera, you can go do that in parallel while you're getting incremental savings with ProsperOps. And to be honest, it's a bit of a fallacy because the idea that you perfect your environment and rearchitect it, and then it's static is a misnomer. The cloud is always dynamic, it's always changing. There's something new that's coming out and so for multiple reasons we just advise our customers wherever you are now you enable ProsperOps, just turn it on and our platform will generate incremental savings now with increasing sort of savings over time.

Host: Jon

I know I said two more questions and which would leave me one more but I just thought of another one I want to ask you is, so first of all you're saying get rid of your old thought of I want to optimize my environment first before I give it to you. Given that ProsperOps does all this AI stuff, just turn it on, let it do its thing and you handle your environment. We will handle the rights, the cost optimization behind them, and their purchasing. What about a customer? So say I'm customer A, right and I've just purchased all these ROIs and now I'm locked in for one or three years and another are some instances where companies weren't able to do much for a customer because they had just purchased these, they're brand new into their contract and you're like, yeah, there's not much I can do for you. That, is ProsperOps able to help me out in those situations saying Hey listen, doesn't matter when you purchased them, we can handle it. Or is there an optimal time that you can come in and help?

Guest: Erik

Yeah. In your example, were those purchases done well or were they kind of done maybe not so well,

Host: Jon

Let's just say I'm new at it and I think I did it well. Yeah, but

Guest: Erik

Gotcha. So another important factor is we can inherit, there's almost nobody today that has no commitments of some sort. We occasionally run across it, but for the most part, people have made some sort of commitment, whether it's savings plans, our eyes, et cetera. Our platform can inherit whatever historical choices you've made. Doesn't matter. Our platform can inherit it. Now in some cases, we can, depending on what it is. If it bought a three-year, I don't know, some crazy obscure instance type and some crazy region with some crazy platform, we might not be able to, you might be stuck with that standard RI we have to AWS has rules, and even though we're automating, we can't subvert the rules.

Guest: Erik

It's very important to us that we are basically following the AWS terms of service and their guidelines. That's not always the case with providers by the way. I don't want to say more about that, but just so your listeners know, you've got to be careful. Certain techniques are being deployed out there that aren't things that I would recommend customers do relative to the ABS terms of service. But suffice it to say, if it is adjustable, our platform can sort of fix historical problems that may exist. There are certain things that we just can't do because if you make a commitment to AWS and they provide no recourse, a good example, is a lot of people saving plans are great to compute savings plans. We love to compute savings plans. We use them as part of our portfolio. One of the downsides of computing savings plans is that they are an immutable commitment. And by that what I mean is if you tell AWS I want to spend, I'm going to commit to $50 an hour for three years,

Guest: Erik

Once that commitment's made, you have no recourse to change it. You are paying AWS $50 a year. There's a marketplace that you can use to dispose of commitments with convertibles. There are advanced strategies where you can reduce hourly commitment. You can't do that with savings plans. So if we come across a customer they've purchased lots of compute savings plans and they are very highly dangerously high in their coverage or over-committed an example of a scenario where just by nature of the platform, there's nothing via automation or AI that we can do to rectify that until those expire. So there are certain cases, I mean this is sort of why we advise customers to engage our platform because we're going to create a portfolio that sort of avoids this scenario. But if we do encounter a situation like that, we may be able to fix it or we may not. It just sort of depends on what aw, you know, rules allow us to do.

Host: Jon

Nice. All right Erik, so my last question is how can I find out more information about ProsperOps?

Guest: Erik

ProsperOps.com would be the great place, the first place to go on that site. Lots of options in terms of requesting a demo or chatting with somebody on our team. But ProsperOps.com

Host: Jon

Nice. I like it. Everybody, Erik Carlin, co-founder, and chief product officer at ProsperOps. Erik, thank you so much for joining me.

Guest: Erik

My pleasure, Jon.

Host: Jon

All right everybody, my name's Jon Myer. Thank you for watching the Jon Myer podcast. Don't forget to hit that, like subscribe and notify because guess what, we're out of here.