CEO as Experiment Officer, Design Partner Personas & Rebranding w/ Sha Ma @ Topogy

What happens when your team can't keep up with internal tool requests? People build their own solutions, Shadow IT spreads across the org, and six months later, you're untangling the mess. Retool breaks that cycle. Their AI AppGen platform gives teams a govern place to build. Dashboards, admin panels, workflows, so everything stays secure. Your teams get unblocked, you don't inherit technical debt. Head to retool.com/elc and stop being the cleanup crew for Shadow IT. That's retool.com/elc. We kind of lived with the product as it evolved and shaped up as we learn more about what features we're on as build, what value we're adding from talking with our design partners and doing validation in the marketplace. It's almost kind of like renovating your house, right? And so whenever you move into a place, you know, I used to kind of tell people like, don't do the renovations right away because you want to live in the house for a little bit, to understand like, oh, this is kind of my natural habit. I tend to look out that window and therefore I want something there. And the kind of path that I take from one room to the other. And once you kind of build that, like, you know what you want, you know, kind of, you know, how you live in the house, that's kind of when you start renovating. So that everything feels like, okay, this is very custom. This, you know, fits that right. And so I actually feel like the branding exercise we went through is very similar to that. Welcome to Engineering Founders, the show for engineering leaders making the daring leap to start their own company. In this episode, Shama CEO and founder at Topaji shares her playbook as the chief experiment officer. We deconstruct how she tests on boards and scales out different experiments with new AI tools to accelerate her entire team. We also explore parts of the founder journey from identifying the right customer personas and finding design partners to the full story behind Topaji's rebrand. Let me introduce you to Shaw and Topaji. Topaji is an AI native cost optimization platform designed to turn infrastructure complexity into clear actionable insights for finance and engineering teams. Topaji turns the unseen connections within your infrastructure into clear data driven actions, helping engineering teams optimize cloud cost performance and focus. Shaw previously was CTO at catalyst.io and industry leading customer success platform. She was VP of engineering at GitHub where she was responsible for core platform and ecosystem. It was part of the leadership team that took send grid public. Enjoy our conversation with Shaw Ma. Shaw first off, just want to say welcome to the show. Thank you so much for joining us on this Thursday. How are you doing? Good, good. Thank you for having me. I'm really excited about it. Yeah, so to get right into it, one of the things I was really excited to talk to you about is experimentation and the role that that plays within how you lead the company. I think to start off with a quote that you shared with me is that you see the CEO as the chief everything or the chief experiment officer. So I wanted to get into that a little bit because I love the way that this shows up into how the company operates. So I think maybe bring us in a little bit to what do you mean by that and what does this look like? As a startup CEO, especially in the age of AI, I tell my team, there's no better time to be trying something new, doing something new at a time like this. The pace of innovation is so fast and there's so many tools out there available to us. But at the same time, we're a very small team. There's only seven of us and we want to keep things lean, but also accelerate, right? And so because of all the AI innovation, there's a lot of expectations, you know, from everyone that thinks just move a lot faster. You can do a lot more with smaller teams, you know, and there's like a ton of new tools that are coming out every day. Like the rapid pace of, you know, just evolution in this space, right? Like there's new tool you hear about like every other day and you're like, oh my gosh, like how do I not have this be such an overwhelming experience for my team? Because I can't also, you know, have my team trying out all the new tools and we're not getting anywhere from product development perspective. So that's kind of where I became the chief experimental officer for our startup. So back in February, when MCP first came out, I was like, oh, this is a really cool tech. Let's see what we can do with this. And we set up an MCP server and all of a sudden, I'm like, oh, this is really great because previously back even just six months ago, I was developing, you know, we were doing rags and we were setting up vector databases, chunking up all the embeddings and feeding it into the thing and all of a sudden, I can have just, you know, an LLM talk to my database and, you know, skip all of that, right? And I have this thing that I can just pass in as context. And this is really cool. So basically introduced that to my team and we started kind of actually building really around the concept of MCPs and how do we actually talk natively behind the scenes to our data. And so, you know, that was one of the introductions. And then another one is that, you know, again, when we started very small, you know, I was looking at all these new AIS tools coming out. And this is early days of bolt and lovable. And I experimented, I think it was with bolt.new at the time. And I told my lead developer, Tom, I'm like, hey, I think we can actually just create our entire onboarding experience with this. Let's take a look. Let's feed it, you know, the UI that we just got from our designer and see, you know, where it takes us. Tom was able to get our entire onboarding experience up and running in one afternoon. I think he got us 80% there and he was like, oh, I ran out of free credits on this platform. And I'll just do the rest by hand, but it's not bad, you know, it got us 80% there. So we were able to build our onboarding experience in one afternoon because of that. So really, you know, I've been like, you know, I've been enjoying a lot of the experimenting, having fun with a lot of the tools out there. If I hear something new, I try it out. If it's good or if I see a, you know, very specific use for it, I bring it back to the team. Debbie, one of my other developers, she came from a Ruby on Rails background and we've been primarily developing and going in the back end. And so I was like, Debbie, you got to use cursor because this will help, you know, bridge all that gap and, you know, you'll be a going developer in no time, right? And that's really helped her, you know, adapt to a new, new language and new platform. And then it's another example, we have, you know, again, very lean team. So we have one designer, Lucy, and basically, you know, she's like, oh, I would love to have like a design partner to be able to brainstorm and just kind of bounce ideas, you know, off of, and then. So I was like, oh, you know, I got the perfect tool for you. And this was before I think my make came out, right? And so this was with Anthropic Cloud Desktop. And so basically, I'm like, you know, describe what you're trying to build to cloud and look, cloud can generate an actual web page with all these different blocks and elements and things like that, right? And, you know, you can tell cloud to like, you know, give me interesting visualizations and we were able to kind of experiment with like, Sankey charts. So that became a really good design partner. Yeah, so we've been having a lot of fun because, you know, I've been telling my team, I'm like, things are moving so fast. If we're building like we were like even a year ago, we're doing something wrong, right? And so when we're looking at, you know, what does it mean to be an AI native startup? What does it mean to be a startup today? We're constantly thinking about like what tools can we adapt into our flow to one, accelerate us, but then also what can we actually build into our product and into our platform so that we can offer an experience that nobody else can because they started two years ago or three years ago, right? And they have that, you know, for them, it's a whole change management process, but we're building, you know, it's an AI native startup from the ground up. So and that's been, you know, super fun and exciting. These examples are so powerful because I think it provides like this like comprehensive understanding of how all of these different relationships are changing. Let's use the bolts and lovable example with Tom because I think like that type of partnership, like I think also is so emblematic of like the early stage relationship of idea happens. Let's see what we can do with it and then kind of push it and extend it and see how far it goes. Can you talk to us in a little bit of like what did it look like to initiate that early experiment? And then what was kind of like the the pass off like and like what made it go so fast? I think that maybe like I'm trying to like really zoom in into like the granular details of this relationship because I think like that speed is like so critical right now for early stage companies. Bring us into like that those early moments and like what set up the speed to deliver on that type of experiment? Yeah, so you know kind of just taking a step back to kind of my previous role as like engineering leader, right as VP of engineering of a large organization. I'm probably not the most like innovative or the forward you know adoptive technology by any means right usually it's you know someone from my team they find a really cool tool they show it to everyone and then I'm kind of on the receiving end of it. It's like hey you know once you find a really good tool. How do I scale this so that our entire team can benefit from it so we can actually introduce it into our process and flow. You know now being on the startup founder side of things you know we've kind of completely flip that on this head because you know I don't want my team to be spending a lot of time experimenting trying new tools and taking up a lot of time when we need to be accelerating our product roadmap and just getting the product out there. So usually you know I'm now the one who's leading a lot of that research and experiments saying like hey I heard this thing just came out let me play with it and if it's really good or if it's useful for specific application you know let's apply it here right and so from my perspective. Really there's two things that I tend to optimize for when I go do research and experiment with these tools the first one is how do we accelerate our team. You know how do we bring a tool into our existing stack that just makes everyone go faster whether it's you know through that kind of brainstorming phase or whether it's like hey we already have a product and you know we just need to get the code done so let's roll it out right. And so what we know is that we have a very rare opportunity right now as an early stage startup because there's so many lines of code and so much you need to do just to go from zero to one. There's a lot of boilerplate things it's like we need these integrations we need a basic way of authentication. We need to hook up these initial modules so we can start you know doing our CICD so we can start kind of getting poor requests and you know up and running right. And what is AI great for it's amazing for zero to one because you're not dealing with 10 years of tech debt or complex code base that's you know all over this place distributed teams you distribute code base you're basically saying like these are the boil of play zero to one things that everyone needs and so like let's just leverage AI to generate that we only get this opportunity once because we're going from zero to one. So what can we do right and that's kind of where the bull and the loveables right you have a clean slate you know exactly what you want to build you also have an experience developer who can actually take you that last 20% which is what a lot of non-developers struggle with to get things in production right or to get things into production quality. And so that just you know became a perfect blend of like hey we need exactly zero to one we need a lot of generated code in order to get started and we have a developer who's experienced enough to actually get it to production quality. And this just perfect storm just kind of helped everything accelerate for us at this stage so like I said like our timing is impeccable right any earlier we don't have tools like this there is a lot of like early days of rapper technology or chat bots or you know things like that right that wouldn't have you know made it into our kind of development stack and then just with the pace of evolution with the pace of all the new tools that are coming out these days it's a perfect blend of kind of what we need and what's available out there. I imagine to it removes a lot of kind of like the toil of like the early stage building where he said like there's a lot of the boilerplate things that have to happen so you know you got to do that. And so anyway to accelerate that and to get to some of the more nuanced or strategic levels of the business like yeah absolutely like nobody wants to be spending like a month doing onboarding like let's just get out that out of the way so that we can start onboarding customers right which is our ultimate goal so I think that's like a really important of like where should your priority be like that it's like such an important thing to reflect on there. I want to go into the relationship with Lucy in the designer and like how that created leverage because I think like for a lot of engineering leaders listening to this that are making that transition like the design side is like maybe like a gap or an area of like less experience. When you were kind of working on like setting up that design partner with Lucy like how did that start to accelerate things like in what was kind of like your experimentation and then with Lucy like onboarding her into that flow and then how is that kind of impacted things downstream. Yeah so from a zero to one design perspective things actually are a lot harder than you would think because a lot of times you're like well you know this is the data we want and you know once I see it it makes sense but a lot of it is like we have a lot of complex data so what we do is we help other teams using AI to optimize their infrastructure from an efficiency and cost management perspective. So we are bringing in a lot of different data different pieces of information but like how do we make that makes sense without just showing your generic like mega bar charts and you know lying graphs and like overwhelming people with like excel spreadsheet like tables. When it comes to you know looking at all their instances and how much each you know what's the unit economics for each thing and then all that stuff right you know so you kind of get this like you know okay I have a lot of data and I want to present them in such a way that it's easy to grasp and it's easy to action on. Without overwhelming people but it also needs to have all that information and so a lot of times you know when you think about that you tend to kind of gravitate towards like you're the standard way of you know visualizing this data like there's the pie chart there's the bar graph there's a line graph right and you know here's the time series you know type of thing and so I think what we introduced is so especially with you know a small team with one designer. You can really rabbit hole down kind of this path of you know this isn't very creative this looks like just like any other tool out there right and so we start a kind of bring caught into the mix and we're like don't worry about the data like let's just explore different interesting ways of visualizing like let's build a webpage that tries to visualize these like three things and give me like five different diverse ways of visualizing this right and so those are the prompts that you can give to these you know large language models and caught has always been very good at generating actually. Working code actual web pages so not only does it kind of give you these different visualization actually almost build the web page for you with those components and so that's been a great tool you know for Lucy to basically say like okay I want to visualize this data. We have this API from the backend but instead of just your standard ways of visualizing with bar charts and filters what else can I do to make this like make more sense and easier to understand right and we've gotten a lot of like dynamic and diverse visualizations ideas out of that. Including kind of a stinky flow chart. This is where once you have some basic templates a designer can very much add that human touch. We actually came up with a heat map for visualizing kind of cycles right so like if you have weekend downtime and you have low usage that's a great opportunity for auto scaling and scaling down your traffic over the weekend. And so we use the heat map but then you know Lucy was able to say like oh instead of you know the kind of the standard red and green and yellow which isn't really friendly from an accessibility perspective. How about we use dots right like the bigger dots represent you know have your traffic and the smaller dots represent so so you're not only getting kind of the visual heat map but you're also getting kind of the size where it matters. Right and so she's bringing a lot of like human creativity layered on top of the AI recommendations saying like oh he might be a really interesting way to visualize this than your standard bar chart. If you're an engineering leader then you know this cycle your teams focused on building product but someone in ops needs dashboard marketing needs an admin panel finance needs a custom workflow the request pile up you can't get to them all. So people start building their own solutions shadow it spreads and eventually you're the one stuck cleaning up tools that were built with duct tape and good intentions. The tool breaks that cycle their AI app gen platform gives teams a govern place to build the tools they need so everything stays secure and under your control. Someone could type build me a customer admin panel that manages accounts from Postgres and they'd get a real production ready app with proper permissions built in your team gets unblocked and you don't inherit a pile of technical debt down the road. It's weird of being the cleanup crew for shadow IT head to retool dot com forward slash ELC and see how other engineering teams are democratizing app building without creating chaos. Because honestly we could all use a better way to handle internal tools sometimes you just need to retool learn more at retool dot com forward slash ELC. What's so fun is how this illustrates just like how can kind of amplify your abilities and talents as like the human founder leader co founder and like team member. And so like the way you describe that I think it's like great it's like I can see the acceleration of like this like partnership and pairing in a really powerful way. I want to dive into how you're applying AI into the product but I figured it's probably a good time to talk about the origin story behind topogy and then your transition to a founder so bring us into the story and the origin story here. Yeah absolutely absolutely you know I'm an I'm an IT graduate engineering school I've been an engineer and engineering leader my entire career for the last two decades. I've been in roles like VP of engineering at sun grid going through an IPO process VP of engineering at GitHub where I managed a fairly sizable team of 150 to 200 people. And then I was the CTO of a series stage startup in the customer success stage and then we eventually merged with our second largest competitor in the space and then for the last nine months of my career there is head of engineering actually had a little bit of private equity experience. And so kind of coming out of all of that one of the key realizations that I had is that for engineering leader we usually manage some of the largest teams inside of a company. But rarely are engineering leaders tapped to become the next CEO or they rarely have a lot of presence or talk a lot during board meetings you know we're get tapped for board positions you know later in their career. So why is that and how do we actually bridge that gap what are some of the things that engineering leaders can influence at the company level at the board level. And then this kind of ties back to my journey at catalyst which is my my last role as head of engineering so when I started at the company we're very much you know fast growing startup coming out of it was December of 2020 the peak of SERP we had just raised a huge round of financing for a series fee and we're going very fast. So at the time the mentality was like hey growth at all cost if we can spend our way out of this let's just buy this versus you know kind of that can cut up down to level time let's just hire more people you know spend our way out of this right and you know we know that quickly shifted so you know starting in late 2022 2023 the economy shifted and basically we have to adapt super quickly so everything turned from the growth at all cost to how do we operate the business efficiently and what does that mean. And at the time we actually didn't even we cared so little about our finances we didn't even have a full time CFO we had a fractional company that we're working with and so a lot of that responsibility actually ended up landing in my kind of domain from a technology perspective because a lot of the spend was in you know engineering headcount. It was an infrastructure cost because we were going so fast we initially started in AWS and then moved to DCP you know we were kind of splitting our cost between the two clouds and we added a lot of complexity through this process getting to multi cloud not because we choose to but just because of the course of decisions made along the way right and so there are a lot of things that needed to be re examine and looked at both from an engineering efficiency perspective as well as just you know making our product more performance and faster. And so when I started that evaluation process of like how do I go about thinking about this how do I actually you know so my charter was to get our gross margins into the seventies preferably upper seventies which is kind of the industry standard right because at the time as a you know serious be stage start up going through that high valuation. And then having to immediately think about efficient growth you're basically saying like what are the metrics like we need to one we need to extend the runway for the business to get gear up for our next round of fun raising and then what metrics do we need to get there right and so for business like ours you know here's what our expectations are in terms of growth trajectory and top line business revenue growth and here's what we need to do to kind of show efficiency in terms of our burn rate as well as you know cost of good sold gross margins and numbers like that right and so. So my number was kind of getting our gross margins in the upper seventies and a lot of the contributions from engineering so gross margins there's the revenue factor right and then there's the cost factor and then the cost factor a lot of that is cost of good sold or you know kind of what is it cost you to power our product in the production environment. So we took a finer comb we did all the low hang fruit starting with getting off of you know founders credit card and renegotiating all of our contract but there's only so much you can do without actually looking under the hood and kind of replacing some pieces or tuning some engine parts and so we looked at all the tools out there and we actually just didn't see a lot of tools that was you know meant to solve this problem so this cloud got more complex there's a lot of tools that you know focus us on one aspect of your cloud spend so they look at things like you know maybe just helping you by reserved instances or helping you optimize Kubernetes or helping you like figure out your spot instance usage and things like that right and so from an engineering leader perspective you know I didn't want to have to piece together like nine different tools I needed a holistic view and we were multi cloud and a lot of our spin was an even just limited to the cloud we had snowflake we had data bricks we had data dog from an observability and we were able to do that. And so I need a tool that kind of give me visibility across all of my engineering spend but also kind of help me optimize and understand like you know what I need from each one of them without me having to like build very specific knowledge every single time and I just didn't find a lot of tool out there. And we actually did turn in house build our own tiger team and then just went category by category and what we've noticed that there's a lot of common patterns that people look for in cloud savings right there's a handful of things there's always like you know here are your top three or four services that contribute to 80% of your cost on Amazon or GCP you look at things like network transfer cost your ingress and egress cost you look at like under utilization like right sizing over provisioning right you look at all these things I haven't even touched in their definitely. Not being used in production but I'm still paying for them so idle instances that I can maybe just shut off so there's a set of things that we've found like category by category that every engineering team that goes through this process is pretty much looking for those patterns. And I think one of the realization is especially with how good that AI is getting these days like what is AI good at right AI is really really good at pattern recognition. And so what if we fed all of these knowledge like these patterns that we keep looking for in these production environments into AI and what can AI actually tell us. So you know imagine a world where teams don't have to build up their knowledge in every single one of these different categories about how do I actually decrypt this very complex invoice from data dog and try to match all those different units to how my team is actually using this dashboard. How about like AI just tells me that and tell me what are the optimization opportunities right. So that was kind of a realization for us is kind of that founding story tying it back to the earlier realization of how to tie engineering output to company level impact. I feel like going through this journey at catalyst I really felt like I actually had a company level metric that I was owning and driving and optimizing for as a company you know see level executive at this company. And it was really on me you know it's given how much focus there was on growth efficiency to report to the board in terms of how well we're doing. And so basically we were able to keep our infrastructure spend flat while doubling in traffic the last two years I was there based on all the optimization efforts that we went through. And then when we actually merged with our second largest competitor in the space we went through that exercise all over again but because of the knowledge that we built up actually made that process you know very smooth and we're able to identify some savings immediately. The story behind that I mean like to pull back the hood or and all of that is incredible. I think for people listening to this like they they so understand like the pain that it was to go from this like zero interest rate phenomenon to a huge swerve into operational efficiency. Like this dominated our twenty twenty three topics like ever it was like all right how do I move to become more operationally efficient as you're talking about this journey like there's just so many other people who like their stories that I'm going through as we're as we're talking about this. So one of the things you mentioned earlier on our conversation was this idea of as you were reflecting on these different areas like talk to us a little bit about like the making your product AI first journey or making AI first and native out of the gate like talk to us a little bit about what that journey has been like and what does it look like so far. In the day twenty three everyone was going through that same process right and so I talked to a lot of engineering leaders in the similar positions and everyone pretty much kind of had that same idea of like we looked at a lot of tools out there and everything seemed very manual like I have to tag everything. I have to do a lot of dashboard maintenance I just wanted to get to the heart of it and kind of actually figure out very quickly what can we do so we can then switch our attention back to feature building feature because ultimately that's the goal right like we don't want to spend all of our time doing you know cost optimization or tech that like that we're not. You know focused on actually you know growing the business and so a lot of engineering leaders ended up having to pause what they're doing in twenty twenty three build this Tiger team build knowledge in each of these categories and then just go product by product or platform by platform to try to optimize their cost so that they can get to what the business needs from a gross margins perspective or kind of lower their engineering cost because a lot of times in order to extend the runway engineering leaders weren't giving many choices they're like hey. But your infrastructure cost or we have to downsize we have to you know have a rip right and so you're kind of forced to like okay how much savings can I find on infrastructure so I can keep building on the team and the features right. So I think that was kind of a lot of what we were faced with back in twenty twenty three and so we're all doing the same things we're all looking in the same places for these things and so how do we collect all of our you know kind of common knowledge and experiences in this area and basically use AI to actually solve this problem. The other thing with AI is that AI doesn't get tired and AI doesn't need to pause what they're doing in order to do this right and so a lot of times what we found is that having to pause what you're currently doing to go look at infrastructure optimizations or to look at you know efficiencies or cost you're actually basically really impacting kind of the flow of your business right. What if something can just keep running in the background you know twenty four seven and keep an eye on this. The other thing that I think you know this is one of the things that I'm like just drives me really mad is just that you get surprised right and so when you're managing a large organization like actually any organization that's bigger than a handful of people you know as a engineering leader you don't have full control over what makes into production right. Anyone of your developers could basically say like hey you know I rolled out a new feature I changed configuration and then you know code reviewers they review the code everything looks good and then three months later you're like hey I got a call from finance and how did our bill go from this to that. Or you know in some cases it's kind of like I got hit up by an account rub for early renewal and I'm pretty sure I just renewed this contract you know last month but like apparently we exhausted 60% of our credit you know somewhere along that right and then usually it's kind of like okay now it's going to be a very investigative journey around like let's figure out what actually happened in the last three months that you know caused this you know because we didn't catch it early enough but the worst is that you know well now like we have to pay this because we already owe this much money and so this is a very reactive process which both makes you look really bad in front of finance they're like what are you guys doing over there like how can you not know about this this has been happening for the last three months and you're like well I don't have full like control over every single line of code that goes into production and everything has been code reviewed but you know as a developer you're like I don't know I just put this dashboard out there where I turn on this thing like you know this happened at least three times to me you know at catalyst where it's kind of like oh we introduced this new feature on Databricks and you know we just basically used all the settings out of the box and we didn't even look at the garbage collection setting and the cost the charge didn't actually increase our Databricks bill but we found it three months later on our S3 charges where you know it increased by a lot and we're like okay and then so kind of then going back and figuring out the correlation between all of that right basically what we end up with actually on engineering teams is this I call a sawtooth pattern things look clean for a little bit but then you know over time things get progressively worse and you know non-optimized you get that call from finance or your CEO is saying like hey what is this charge what's happening here and then you take action you pause what you're doing you clean things up a little bit and things you know get back into a pretty good state but then over time they get bad again right and so why you know have all that waste when you have this kind of sawtooth pattern when you know AI just can basically keep an eye on things in the background alert you when you need to pay attention when things start to tip up or you know always let you know that hey like based on this trend you're gonna be hit with an early renewal right and so you might want to do something about it or like hey we actually detected this common pattern that usually trips people up and so before you even put that into production like we'll give you a comment on your poor request that's as before you make this configuration change this is expected to increase your infrastructure cost by $5,000 do you still want to go make this change right so we want to take a much more proactive approach to how you manage its process and so you don't have to spend your entire time like you know pausing what you're doing going to investigate and you know spending time to fix the issue right we're just catching them as they happen so we can alert you and you can take action on them you have such a deep understanding of like especially from your personal experience like going through these massive changes and the patterns that exist so like number one there's like this this huge level of of noticing and really deeply understanding the space the next side of it though that I really have appreciated is the phrasing of your questions I wonder like what would it look like if something could run in the background and then not derail progress and do this there's like almost this idea of like envisioning a world that's different and then like the third is like having a really clear understanding for like the unique and special moment of how a technology solution like AI can uniquely in this moment really solve this issue and really accelerate these things am I am I kind of sensing those things like is that is that like fair or there are other things you'd expand on yeah because I mean I think a lot of startups you know start out this way right it's a personal pain point and you're like it's got to be a better way of solving this I can't believe you know somebody hasn't solved this problem already why are we still living in a world like this where you know this is acceptable right and then you're like okay we're at a place where the technology is actually good enough to solve this and so let's imagine a world where we actually don't have to live this way or you know kind of deal with these pain points right and you know at the end of the day it's actually just you know solving for waste it's not you know right like so why pay this three months of things that you didn't realize you were paying for but you end up paying for because you used it then actually you know just you know doing the right thing from the from the get go and keeping it that way I kind of at a metal level want to reinforce how powerful it is to use the language imagine a world if and then to really describe like the better state in such and I think like what's so powerful is like the clarity of the terms that you use like imagine a world where like you don't have to build up knowledge in these areas you don't have to stop what you're doing and that you can automatically learn these optimization areas like in like automatically like recoup these cost savings like the power of that language is critical so for people listening that that's the kind of clarity and like the phrasing that I think is really powerful as you're communicating your idea to different people so that was just more so like the way you describe it I think is really powerful can you talk a little bit more about like how maybe that persona journey progressed so you started with people like me and then the journey of it from my understanding is that you sort of expanded or really differentiated that persona into a couple different company categories so one that you're talking about like the zero and just rate phenomenon company who is like we're going to spend the money to like make the thing happen the other side of it is like the extend the runway company or maybe even the like now AI native company with spiraling costs and how do I address it? So I guess bring us into how you've thought about that persona differentiation and I kind of want to then get into like how does that sort of impacted or work with product discovery and validation and how you've started to build out the product because of that. Yeah, so this actually brings back the title Chief Experimental Officer but from the other perspective right so like as a startup founder you're always experimenting whether it's like hey adoption of new tools and introducing that to you know the team as I kind of spoke about earlier or like external validation that discovery validation kind of process for what product do I build? Do I have you know how do I get to product market fit? What am I hearing and when I do my outreach is what messages are landing? What's not landing? How do I do A/B testing you know on some of this right? And so I think initially it's just going out and validating some of your assumptions like hey this personally impacted me but did you feel the same way like when you went through this journey like are you getting a lot of like you know emotional reactions like almost visceral like yes I know exactly what you mean like yes like you know right or are people are like no not not my top priority right now and and so think when we set out to you know kind of do this discover we're looking for a lot of tech leaders, engineering leaders they're going through that similar growth stage journey right and so they started out like you know Series B, Series C, they're looking at potentially you know the next round of funding or an IPO or they recently IPO'd and now all of a sudden they're getting an overwhelming amount of interest from Wall Street, from people who are scrutinizing all these numbers and they're like oh we need to look better like what is this growth margin thing and how do we improve it and right and so that was kind of our initial hypothesis is that those are the kind of companies that we're at a scale where we need to care about this or we're at a scale where either investors care about growth efficiency or the public right because now our public company and the Wall Street is expecting that but as kind of we did research we're learning that a lot earlier stage companies are looking at this number and some of them because right you know they had the high valuation during the Zerp era and now they almost need perfect metrics to get to their next round of financing and the best way to do that is actually to extend the runway by themselves a little bit more time and so this is where efficiency matters right how do you do that without having to go through riffs or layouts how do you find efficiencies in your systems so that you can have better understanding of all the different pieces right because a lot of things probably gets added over time so if you look at you know a 13-year-old architecture it's a lot more complex than you know two-year-old architecture just because pieces get build out over time do you still have visibility into you know all pieces of your complex infrastructure and do you need them all and at what point does something grow so significantly that you're like oh is there a better technology to solve this problem now maybe this wasn't the right tech you know when we put it into place but it was you know we needed at the time it was simple it was straight forward but it's not scaling well with us and with our use case right and so those are the things that you start to kind of you know take a you know microscope and start kind of looking at and examining you know so as we start talking to people we're learning that you know earlier startups are now saying hey like where is startup that's built that post the Zerp era so we kind of want this discipline to begin with you know so we want to kind of start you know understanding our growth what we found really interesting is that you know with the latest wave of AI startups people are like whoa AI is so costly like you know we cannot not care about you know AI cost right just because tokens and it's totally different way of thinking right now it's not just even an optimization problem it's also like people are using so many tokens they're getting rate limited on one platform so they actually have to spin up multiple platforms and managing like hey when I run out of tokens here I'm going to use the tokens from this and so like basically they're adding to their overall architecture and infrastructure complexity because they actually have to build a gateway to kind of say like where am I going to get my next token from and how do I power all the demand that I have where AI tokens and inference and then on top of that what can we do knowing that that's very costly you know all the input tokens output tokens and it's when we have done a lot of optimization what can we do to actually understand where we're spending the money and how do we actually start doing optimization around it right and this is at a time where things are evolving so fast you're introducing new technology latest models are coming out the frontier models are coming out all the time and the APIs for getting that information is not very mature yet because nobody's focused on that right and so there's a lot of just kind of the perfect storm scenario and yeah it's it's just been really fascinating in terms of like you start with a hypothesis of like hey we're going after the growth stage companies and we're using AI to solve an infrastructure optimization problem and then you're feeling kind of the tug towards the oh there's a ton of startup companies that are now like just you know consuming these AI tokens and there's zero visibility and zero optimization around it you know one company we spoke to I think they went from a couple thousand dollars a month to over 110k a month of AI token spend in a course of less than six months right and you know for them it's it's you know it's great as a business they're getting all that traffic and on that demand but also they're like oh we can't not think about our infrastructure cost right because you know this is just a crazy ridiculous amount of infrastructure cost growth that we've never seen before. So which customer do you go after or when do you revisit assumptions for maybe the persona or is the product flexible where where it can drive and power all of those and I guess like I'm trying to understand like maybe decision making and strategy around how all of this like clear understanding of how these things are shifting that impacts like maybe early customers that you start to that you start to work with more because you're like okay there there seems to be a growing demand there we're feeling the pull stronger there like can you talk a little bit about maybe that thought process or the questions or the journey there. Yeah so kind of through our design partnership process we look for different representations of different companies and different needs and so basically you know we don't want to fall in the trap where as a startup you're like okay the loudest voice wins or the you know kind of the we only have two design partners so basically we're building a custom product for them right and so we're very intentional you know given this problem space given kind of all the people that we did validations around kind of picking three design partners that are very different from one another in terms of the stage of the business the amount of spend they have and kind of the problem they're trying to solve for right and the level of maturity around their phenops and kind of the notion of infrastructure visibility and cost management and so we have a YC company that's earlier stage they're like hey you know our system is not super complex you know we're going through a contract renewal process and we're not very confident in our forecast can your tool help us with some of the forecasting aspect we're working with another company that just recently went through a merger and so the parent company actually had a very mature cloud operations they have cloud engineers who kind of you know make it their job to kind of do a lot of these fine tuning but with the merger they're like oh now we are multi-cloud and we would love to see things in one place and we would love to kind of quickly catch up kind of the company that we acquired into this process and so how do we do that right and so and then one company we're looking at they're building like AI training models they're looking at GPU consumption right and they're definitely on the forefront of a lot of that and so we're working with these different partners to kind of give us you know just different input points in terms of what do people need as we build out this product and what's most important a couple of things that we kind of distilled from that is that you know one focus on the big complex hairy things that AI is really good at right and so we actually you know if you look at kind of cost of goods sold a lot of it is consumption based pricing with very complex systems like AWS where there's thousands millions of different types of compute and storage instances you could use there's different combinations of our eyes and savings plans and spot instances how do how do we actually optimize for that that in itself is a really interesting mathematical like you know optimization problem right and you know things like data warehouses like snowflake data bricks like you know how do I think about query optimization let's look at like long running queries and understand what's actually happening and how do I optimize for that and so like those are the things that we tend to look at using AI because machines again you know you can just look for these patterns try to understand what's happening behind the scenes versus those are the ones that usually take a long time for a human to be like okay why is this query running so long what are the errors is causing let's let me look through all my logs and try to figure that out right you know machines can't answer those questions a lot faster and so we are not as focused on kind of the license based or seat based pricing because those are you know pretty straightforward you know you have this many number of developers therefore you're buying this many number of license you know somebody you know on boards are off boards you increase or decrease that number and you know they're tools for doing that but those are you know much simpler so we try to go after the really complex problems where we're like okay this is where machine learning this is where AI like it's really good for right and so let's use the technology where it's you know best suited and look for this pattern so those are some of the things that you know I think we think a lot about with the startup is like you know do we have the right design partners are they diverse enough that we feel like we're building you know for the population which is kind of being skewed one way or the other and are we solving the hard problems and building ourselves that moat leveraging the right technology that we have today that we didn't have a year ago or two years ago or three years ago I think it's really interesting to see sort of like that expansion of like really complex capabilities or like use cases and scenarios as a moat because as soon as you started to describe like the complexity of the different scenarios that folks in face and how they can change over time like I'm imagining like the company that just got acquired like their need like probably reflected some of those like series B or series C startups that were trying to like get to the point of like reducing costs to enter into that type of acquisition conversation so I'm like oh I can see like as you expand those capabilities how they can kind of create reinforcement of the tool and like the long term use of it Absolutely and because you know we're also built in an age of AI right like our tool gets better the more use cases that we handle and the more cases that we see right because those are the patterns that we can add into our own knowledge base to feed into AI as kind of additional patterns that I can detect and look for So one of the things I was excited about is like kind of right now is like a really special time in the journey here and that you all just completed a rebrand and I think this is like a really special moment for you and for the company I just wanted to get some like lessons and insights around branding for folks listening in who maybe are at different phases of that journey so tell us about the rebrand what triggered it and then maybe some insights into how that process went and what you might encourage folks to apply in their own rebrand Yeah sounds good so when we started this process we actually incorporated pretty fast because you know we actually went through a fundraising process and you know at the time I was pretty naive I was like oh you know this is my idea this is what I want to do but I don't even know how to like take an investor's check so what do we need let's start from there okay we need a bank account what do we need to do to get a bank account okay we need to be incorporated what do we need to do to get incorporated and find a company name and a domain so we kind of went through that process pretty fast and we didn't you know think too much about it and I think that was the right thing to do because you know in talking with some of my founder friends right everyone's like hey right now people are buying making their buying not buying decision not based on a name right they you know they know you they know your team they're based in basing it on kind of their trust and that you can build this product that you're you know talking about and also kind of how good your product is in solving this problem for them right there's a couple of other things that you know I think just are my own like pet peeves like as a you know leafy engineering I was like okay I'm always up for experimenting with new tools and like seeing what cool startups are out there and that they're building but you know I always ask like how long is it going to take my team to get up and running on something and so you know to me that kind of whole like seamless onboarding how do I get onboard it super fast it's really critical and so we kind of designed our startup so that you know our onboarding it's like oh while you're on the call with us we can get your onboarded like it takes you know minutes like you know if you had the right credentials you know write the person on the call with us right and so that's something we focused on in the beginning but kind of back to the the re-bending story so we came up with the name super fast we started kind of building this product we fundraised but at the end of the day you know we were kind of sitting on this name and we're like does this actually represent what we're looking to do and who are we and so we started asking those questions and it's kind of like you don't need to worry about your brand or your name until you do right and so while we were building with design partners nobody cared what we were called but you know we were looking forward we're like okay but we want to start like marketing about it this we want to start talking about on LinkedIn we want kind of people to start like you know organic traffic you know discovering us and kind of understanding you know what we do so what can we do with the brand right and so our old name was kind of more like it was called Fintey it was really around like you know how do we help bridge you know finance and engineering teams when it comes to Finops right but then you know we decided that like we're an you know AI native company and having the name Fin in our name doesn't really mean it's modern and you know innovative as you know we want it to be and also nobody can pronounce it and my designer hates having a capital T in the middle of our name so what can we do to actually you know make it more us this is you know I think one of the really fun things about startup is everyone is trying to help be helpful and there's so many kind of people in your network you can tap into and just pick their brand and so I called up a new friend from GitHub who's done a lot of the branding aspect of the Octocat right and I was like how do you how do you go about like figuring out a name like do I just like start like building lists of words and seeing what's available from a domain perspective like you know what is that process and it's actually really interesting process in that like he's like well what do you want this name to represent like the image that people think of right that's your brand attribute and so he's like you know think about what we did at GitHub like what does it mean to be get hubby you know we're empathetic we collaborate with developers we're a little bit nerdy you know right and so he's like at the end of the day like what do you want that name to kind of generate from a feelings perspective right in motions perspective when people look at it and so um so he's like start with brand attributes like come up with the list of words that you want to represent your brand and then so we kind of start it with that and we're like you know infrastructure is so complex these days we just want to make infrastructure complexity into like giving people clarity helping them grow helping them grow efficiently and our not here to replace humans like we really want to be helpful right we want to work in the background um so you know we want to be collaborative we want our AI to actually bring out that human creativity right kind of like how i described the use case between Claude and Lucy right like you know Claude came up with a heat map Lucy made it like that give that like human creativity that human touch um and so that's kind of really what we aspire to do from a naming perspective how do we take a really complex concept and just make it simpler and then how do we you know bring the kind of human creativity and the efficient growth out of things while we were like the kind of the workhorse that runs in the background um and so we started with a lot of that kind of brand attributes um how we want to think about the brand uh we made mood boards right so kind of what do we like about other brands you know and why why do we like about them so we kind of put together um separate mood boards originally you know we did some of the exercises as a team but that kind of then led to group thinking so we actually then when we did the board exercise um we did them individually and we kind of shared like a lot of the what we like what we don't like about it um and actually a very interesting thing came out of it because I think again you know kind of coming from more of a developer background we always thought like oh you know if we're going to build like a developer tools product our webpage needs to be dark like dark mode um and then uh when we looked at all the different doctoral websites out there we're like whoa like you know we like we put everything on to make them a mood board and we're like you know how do these like if you didn't zoom in they all looked exactly the same yeah like dark kind of hacker code ask yeah exactly right you can't tell like you know five different sites side by side you're like you can't tell who's who and so that's kind of that gave us like okay so we actually want to differentiate from the pack we actually want to stand out let's actually just go for you know a light side experience with you know very easy to read kind of developer friendly clean you know that kind of talks about what we do right and ultimately that's um what we want and that represents that level of clarity that we want to build kind of back into the brand attributes and so so yeah so you know like things that you you never thought about much when you're kind of in an engineering role and you're like wow there's a lot of like art and science to something you know like a branding and naming and so um we did come up with the list of words that we kind of started thinking about like that gives us these type of meanings um and I think we kind of use the theme around kind of like oh what about like you know mantle or like you know tectonic or you know magma or things like that and then our designer Lucy came up with um topology topology gets confused with topography sometimes and I think Lucy was actually thinking about topography when she said the word but she's like oh I think there's a lot of really cool like contours and you know heatmap type of things I can do you know with this concept building into our product but you know being the math nerds that we are we're like oh topology is actually a you know we're building like network topology and we're building system graphs um you know behind the scenes to help explain like what's happening um around your infrastructure and how to optimize it and so this is a networking problem and so the topology actually really represents um what we're doing and so we're like you know we want you know a shorter punchier version of that so how about we shorten topology topology right it still has that very like nerdy feel to it and you know and it's kind of like quirky you know as a startup name and so easy to remember you can say like hey we're topology like topology and so that was kind of like where all this came from and then because of all these like math concepts we actually started nerding out around like topology it's like a math and math co-representation of systems and like you know subway maps are you know topology and those impossible shapes like the tour is the climb bottle the movie is strip right and so we're like oh there's so much we could do um around this brand and uh it's actually really interesting one of um our teammates he's from so he's Spanish speaking and he's like oh topo topo actually means like a little mold that you know like um in in Spanish like the animal the mold and then we got all excited um you know we're like oh it's like the little mold digging for like infrastructure savings and finding you know optimization opportunities for you yes yeah so yeah so overall we just got really excited about the name and kind of landed on the name topo g so uh so we're going through that reverning process and you know i don't have episodes going to air but hopefully we can you know have the website and all that done um you know before so uh you can see the end product from that sometimes the mascots choose you you know like maybe you don't know if a mascot's right for you but sometimes they choose you absolutely the little mold digging for digging for infrastructure savings like it couldn't be a better match and i think it's so interesting as you kind of describe this because like this process is so thoughtful and it really is like architecting like the soul the identity the brand of the company and it sounds level of alignment where the mascot chooses you like that level of serendipity like that probably takes a little bit of time so i think like you know when reflecting on like that process like it probably does make sense in the journey or you're talking about like the order of operations and like what you optimize for which was like at the beginning of like how do we get this company up and running so then we can get to onboarding customers as early as possible knowing that such an important value for you so it's like it's almost like yeah it's like a really good like almost like parental principle branding choice it captures like the feel that we're kind of at this point where we want to do big marketing push and we have the time now the the mind chair to really start to articulate what we want to be great moment so i think it's just like artfully from like the sequence prioritization and then the like framework for what this brand identity process was like then it's really really elegant it's almost kind of like renovating your house right and so whenever you move into a place you know i used to kind of tell people like don't do the renovations right away because you want to live in the house for a little bit to understand like oh this is that window and therefore i want something there in the you know kind of landscape in the yard or you know this is kind of my natural like path that i take from one room to the other and once you kind of build that like you know what you want you know kind of you know how you live in the house that's kind of when you start renovating so that everything feels like okay this is very custom this is kind of you know fits that right and so i actually feel like the branding exercise we went through is very similar to that we kind of lived with the product you know as it evolved and shaped up as we learned more about what features we're on as build what value we're adding from you know talking with our design partners and doing validation in the marketplace that evolved with us until we're like okay this is what we're building and this is what we want to be versus you know kind of having like the name kind of the first week that you're you're doing this and try to kind of build into that name fantastic shout we've got some rapid-fire questions if you're ready to ready to jump in the right first rapid-fire question what are you reading or listening to right now oh so i read a lot of different things right now because mostly it's research oriented right and so to become experts in a lot of these spaces i also want to I always want to stay kind of a step ahead of the team and so i'm you know a lot of times it's reading AWS documentation and gcp.com like you know it's it's reading about all these different you know tooling it's reading about like you know the best practices around prompt engineering and things like that so that we stay current and what we do and we can apply that into our product so i do feel like a lot of you know stuff have been like work focused next rapid-fire question a founder resource that you found most helpful founder resource the network so um definitely i feel like i've connected with so many people that you know throughout my career i'm like oh this is a really interesting person i want to keep in touch with that person but then you kind of lose touch over time but you know i think as a startup it gives you an actual like really legit like reason to reach out to people and you know just check in on them and learn from them right and so i've reached out to so many former colleagues that i've worked with previously people want to be helpful like everyone is like one it's i'm so glad you reached out like it's great hearing from you like after all these years like tell me what you're doing and like how can i help right and so tap into that definitely what is it trend you're seeing or following that's been interesting or hasn't hit the mainstream yet yeah so this is an idea that i've been kind of thinking about a lot is that you know there's definitely been kind of this ai hype cycle that you know i've been following right earlier this year it was all about like you know ai replacing humans right it was kind of like hey you know don't hire anymore until we can justify that ai can't do this or like you know as a developer like what do we do like we're all going to be out of jobs you know kids shouldn't go into computer science anymore see that hype cycle has kind of calmed down a little bit because people are now like no no like after we've been burned a couple of times or ai has elucinated and added like a database or deleted a database or we found that it wasn't as good getting into production people are like you know back down like hey ai is not going to replace any humans anytime soon but you know maybe this is more of a back to kind of that agentic copilot style where you know it's going to be a good partner to humans where you can really be there's a good you know collaboration opportunity where humans can focus on the things that humans are good at and ai can focus on the things that ai is good at. I love the question of how do we actually get ai running in the background and imagining a world and like what's possible when that happens that's great. So I have one more question for you. Is there a quote or a mantra that you live by or a quote that's been resonating with you right now? Yeah well a quote that basically my dad told me when i was a kid that i've always thought was a really good one is that where there's a well there's a way you know opportunities come to those who are ready for them right or luck i guess luck is just opportunity for people who are ready for them when they come right and so those are the things that i tend to think about a lot you know how do we you know pay that path for ourselves and how do we kind of like you know believe in something so deep that basically you're forging the path. Shaw thank you so much for an incredible conversation diving in to your story the top of the story the problem space and so many of the different nuances and ways that you've approached building the business we really appreciate it. Thank you for having me here that's been actually a lot of fun just just talking to the story and yeah if you're listening to this and you're wondering how can i connect with other engineering leaders in my city pull up your phone right now and go to elc.community click our chapters page you can see that on the menu on the left find your local chapter click join we're hosting virtual and in-person events all the time and this is the best way to help you get involved expand your network in your city and support your leadership and career growth so pull up your phone head to elc.community join your local chapter and get involved a huge thank you to all of our local leaders who make community happen and thank you for listening to the engineering leadership podcast [Music]