Latest / The MonkCast / Good Wigs & Golden Paths: Dev Tools in the AI Era with Victoria Melnikova

Transcript

Good Wigs & Golden Paths: Dev Tools in the AI Era with Victoria Melnikova · The MonkCast

• speaker-0: Thanks for joining me for this monk cast. My name is Kate Holterhoff. I am a senior analyst at Red Monk. And today we're going to be talking about DevTools and product market fit. ⁓ Very apropos for a Red Monk audience here. ⁓ And my guest today is Victoria Melnikova. She is the head of new business at Evil Martians. Victoria, welcome. speaker-1: Hi, hi everyone. ⁓ thank you so much for bringing me to the show. I'm very excited for this. speaker-0: Okay, so I wanna tell the story before we get started because it's ⁓ pretty funny. So Victoria and I got to meet ⁓ about a month ago when I was ⁓ last in San Francisco, and we recorded a podcast for her own show, which is called the Dev Propulsion Labs. Y'all should check it out. and I I tell you what, it was the most sci-fi setup I've ever seen. There were all these amazing like DSLR cameras and like lots of mics, and we had the over the ear you know, headphones and it was it was so legit and there were like folks there, ⁓ you know, checking the how the audio sounded and it all was like super professional. I was I was deeply impressed. And ⁓ so the show just came out this week. This one will ⁓ be be coming out for a bit, but and and Victoria, what was the exciting thing that happened to to ⁓ to our show? speaker-1: So let me just say this. When as we were recording, Kate and I were talking about different tools that we're using for podcast recording and Kate was like, you know, I I just go like very simple tools, you know, it's it's kind of like a setup that, you know, never fails me. And I'm like, Yeah, you know, we tried to do like a little cool in person thing here and record at different studios and guess what it did fail me. That that very exciting setup that we ⁓ that Kate just described. Basically what happened, the the video got corrupted, sadly. So we were only able to pull some stills from it, which look awesome. But ⁓ we have audio, which is great, but sometimes when you take a risk, it doesn't really pay off. So speaker-0: They do look good. It was it cracked me up. ⁓ you know what else I liked about it? ⁓ so I used to study literature and there were these ⁓ there was such a beautiful book ⁓ layout behind us, like a bookshelf with very well curated literature chosen, ⁓ and some like nonfiction as well and like cute little ⁓ objects or whatever. And but what interested me the most was that to the side, there were a bunch of books that were not chosen. And so I just really wanted to talk, like have a conversation with whoever set it up. But anyways, it was you know what was great about it was just the fact that we were in the same room and we're able to have that conversation live. So yeah, the video is truly secondary. But ⁓ but I just I thought it was amazing because I am so self conscious about not being an audiophile. Like I I'm just I feel like I'm hacking it. But, you know, I post COVID we all have to be A V experts and so I'm leaning in. I'm doing the thing. I'm I'm you know, the awkwardness is okay with me. But ⁓ but it It was great. You know, there but for the grace of God. That's all I'm saying. You know, ours failed too sometimes. They haven't lately, you know, knock on wood, but ⁓ it does happen. speaker-1: It does happen and to be honest, the more I deal vi with video, the more I understand how intricate and it's just a lot of work to do like redecided video production and my hat is off to people who nail it. ⁓ it's it's a lot of work. So a lot Sometimes it doesn't it doesn't really work out but it's a cool story to tell, so it's speaker-0: Absolutely is. You know what I the only thing I was missing, Victoria, was I was waiting for you to drop on LinkedIn ⁓ you know, three things or top five things I learned about B2B sales from my video failing with Kate Holderhoff. So maybe you can do this after afterwards. Jump on it. This is like, you know, this is a good AI slop opportunity for everybody. We all we all need to know. speaker-1: That that's good. Just today I saw this report by SXNZ that like a huge percent of LinkedIn posts is actually ⁓ generated. I'm not gonna contribute to that. I try to write my own posts. Sometimes it's a little tiring, but I feel like we're all over the slope at this point. Like speaker-0: ⁓ my god. The world thanks you. ⁓ but you know what? Sometimes for a summary or something, it's it's okay. But I I hear ya. You know, you definitely want to have the the person in there with a little taste, making it sound good. So okay. Great. so ⁓ okay, so so I w I obviously wanted to start with that story 'cause it's it's like too good and it w it would like made such an impression on me. And ⁓ anyways, you're great. So but let's introduce you more fully to to ⁓ you know, the red muck crew here. So like Be d who are you? What do you do? Talk to us about your background. speaker-1: ⁓ I'm Victoria. I work at Evil Martians and I've been with Evil Martians for it's coming up to six years. So really I'm 32, so it's like the the vast majority of my professional career has been with Evil Martians. And Evil Martians is a consultancy for startups. We do design and development for AI, DevTools, and Cybersec startups. ⁓ and DevTools is really our home. And my job has evolved through the years. I joined as a manager and I just helped run accounts. ⁓ and now I lead new business ⁓ which includes marketing, business development and sales. ⁓ and it's a lot of hats to wear for one person, but we're just about like fifty-five people, so a very lean team. And we work with about fifty startups every year. So As a Lean team, we just have to wear a lot of different hats. And we work mostly with early stage startups. So we ourselves are kind of an early stage startup, ⁓ but profitable. Very profitable. So ⁓ we just make it ⁓ we kind of try to adopt the same ⁓ methodologies our clients do. So we speak the same language and ⁓ just stay really efficient when we integrate into their teams. ⁓ so my personal background. ⁓ I actually went to Smith College and I studied Portuguese and economics there, which was t like a crazy ⁓ weird couple of things to choose, but ⁓ I kinda didn't love investment banking. I wanted to do something a little bit more creative. And while I was at Smith I found this job which was called Fellowships ⁓ Webmaster. It was very mysterious. ⁓ I had to update the fellowship's website at Smith College and do like some social media work and I really enjoyed it. So after college I tried to find my way to ⁓ computer science or web development and I didn't know where to start. And I always thought that it's not a woman's job to do, which is weird, I know. And that's how I stumbled into the agency world and I started with web design ⁓ agencies. I really loved it. ⁓ just I was thinking about it the other day and I thought that agency work to me is is a lot like being a pirate. 'Cause you kind of like travel the seas and you see so many different places, you meet different people, but you never own that new place. So like I think it's kind of like pirate ship. So ahoy. ⁓ That's how I feel about it. Great. speaker-0: visual. Yeah. Martians are kind of like pirates, right? You know, you got the at least the the evil part of it, right? This is like this is a good joining of your metaphors here. I like this. Yeah. speaker-1: So yeah, and as I said I joined the Velmartians and stuck around and here we are. speaker-0: Okay. So you've been there for six years. And how long has Evil Martians been around? speaker-1: Evil Martians is twenty years old. We just celebrated the twentieth birthday, which is kinda crazy. And it was founded twenty years ago by two Russian founders. And I think initially we stumbled across ⁓ well, actually I think the story originated around Ruby on Rails. I don't know too much detail, it's kind of mysterious like what happened then. But yeah all we know is that startups loved to choose Ruby on Rails because it allowed to build quickly. and we we had an amazing engineering team from day one. Some of them are still with us. Like Andre Sidnik has been with the company for like 19 years. And ⁓ we just found that Ruby on Rails is a very efficient ⁓ language framework to work with and startups really enjoy it and can benefit from it tremendously. ⁓ and then as our team grew we found that Rails indeed scale and we found our ways around performance and databases and infrastructure and just organically picked up a lot of traction amongst Rails companies. And we also did a lot of open source work. So ⁓ some of the well-known evil Martian projects are Post CSS, browser list, Tanya Cable, y Yabida d there there's a lot. I think it's over a hundred projects at this point. Nano ID, ⁓ left hook, there's a bunch. And ⁓ we found that we really love working with technical teams and developer experience is huge for us. So naturally we kind of found our path to developer tools companies. So we had a few successful kind of like experiences with DevTools and cybersecurity companies like Teleport, Tance, Bolt and Lago, some there there's a a big list. And we kind of fine-tuned our position into focus on developer tools startups because we really understood the struggle of bridging the gap between technology and actually making it successful, right? So like how do you make money with open source? How do you make money with a developer tool. That's our kind of sweet spot, that intersection. So that's that's where we stayed and we continue thriving in that space. speaker-0: Okay. Okay. Thank you for that history. That that's helpful. And it's funny because Red Monk's been around for about that long too. So it must have been something in the ether at that time. You know, that was that was the fa founder ⁓ moment. okay. So all right, so let's get into the topic at hand here. I'm excited to talk to you about product market fit. ⁓ as it relates to DevTools, because it's something that you folks have been ⁓ thinking about a lot, and it's also newly important right now. I mean, in our AI era, what it means to have PMF is like shifting ⁓ dramatically. And I guess the tools around it are are shifting as well. And like how do we gauge success? So I want to talk about ⁓ some of the ways that you're thinking about it. and ⁓ so so there's a very good blog post. I I I was actually reading all kinds of your blog posts. You've got a lot of interesting stuff on there. So shout out to to the Evil Martians ⁓ blog here. ⁓ But ⁓ in terms of PMF, you say it's not bullion, it's a spectrum. ⁓ product signal and revenue signal are separate and the gap between them is informative, and that the weakest product metrics will tell you where to look first. And I think all of those are like super interesting. So I guess to begin with, like talk to me about you know, wha how you came to these determinations. Like what are your methods for, you know, making these sort of assertions around ⁓ PMF as it relates to DevTools. speaker-1: Yeah, it's actually so you are referring to a big piece of research that our CEO, Irina Lazaro, did. And Irina plays a key part in in that intersection that I mentioned earlier, technical startups and business, right? Like how do you make money with it with your technology? And ⁓ so ⁓ Irina has finance background and she understands really well how to do research. with LLMs. She is she's mastered that. And at Evil Martians, she's worn a lot of hats. Right now she is the CEO, but she has a long history with the company. She's been, I want to say like coming to 10 years with Evil Martians. And she was a manager before, then she became a CFO, then she became a CEO. So she really like climbed that ladder. And the way Irina thought about product market fit is like this. So We have developer tools. This industry is actually growing ⁓ as AI adoption is growing because it's easier to become a builder than ever before. ⁓ but when we think about developer tools, we we can look back at different success stories that took place and documented data that supports their success. So when we talk about product market fit, people typically say something like ⁓ when you reach product market fit, you can feel it. You know, like you wake up and it's your your product has grown. customers are more forgiven about your mistakes because the value that you give them is more than the potential mistake that you can make, you know? There's certain like really vague cues as to what it is in reality, like how do you measure product market fit? So Irina studied companies that are successful, profitable, ⁓ growing a their ARR, ⁓ perhaps. And she analyzed ⁓ the metrics that are available ⁓ on the internet to derive her methodology. And she called it product market fee compass. And as you Kate alluded to, there are two ⁓ important parts to it. First one is the product metric. So Do people conver let's say do people stay with the product? Do people return so churn and conversions inside the product? Those are some metrics that we're looking at. And the second one is revenue. So is you you a is your ARR growing? Like what's going on on the on the money side of things, right? And what we found and I think a lot of technical founders will relate to this. ⁓ technical people often fall in the trap of building. It's really a comfortable space, right? Like as you build, you feel like you're closing tasks, but you don't have to go outside of your comfort zone because ⁓ you can just focus on code, you know, yes, there are pressures, but at the end of the day it's deterministic whether or not it works, right? Like there's very little like hesitation or negative feedback from the code. Sales, however, is much less comfortable. And a lot of technical founders actually struggle with with that. Successful ones somehow solve for for sales, right? Like they are not afraid to talk to customers. They often have strong personal brands. They do a lot of outreach. They do a lot of public speaking, they do a lot of things that are not comfortable for the vast majority of engineers. So basically, when we look at product market fit and various metrics that exist there, and you can find it actually at our homepage, Ilomarshans.com, ⁓ the little kind of like how'd I say, like a little form. There has all the metrics that you can test to actually see what it is about your company that needs to be done. But it comes down to very clear things. Either you need to improve your product or you need to work on sales. And if your product is good and your churn is is not that high, we will probably tell you to go and do more sales. So it's very simple, right? It's just two things. You either work on the product or on sales or go. ⁓ so yeah, in in a nutshell, ⁓ what we see typically in very early stage dev tools is very simple. You need to focus on sales and marketing and go to market. ⁓ instead of focusing on perfecting the product further, adding features and all the other traps that, you know, builder founders fall into. speaker-0: Okay. And one of the things that I thought was interesting is how this blog post differentiates between dev tools and typical like generic SaaS products. ⁓ so that there's like, you know, ⁓ with a generic SaaS, you have like higher conversion and like higher net revenue retention and faster time to value. So like I and actually yeah, faster growth. Right. Okay. ⁓ so I I thought that was interesting. I want to hear more about it. So talk to me about like for one, I I feel like there's some overlap. ⁓ but for two, I I get that like you weren't the one that did the the research initially. But like ⁓ maybe even what have you seen with with some of the companies that you've worked with? speaker-1: To be honest, right now I feel like the landscape is shifting because AI is really challenging how things were. ⁓ so I feel like we find ourselves in a very peculiar situation. And I was just talking to someone the other day about startups that we see succeeded in twenty twenty-six. Like I'm talking Superbase, Work OS, Planet Scale, you know, DevTools or infra companies that have existed for a while. Like, I wanna say like five plus years, right? ⁓ they did something, first of all, with AI, their growth only accelerated. Before AI was just steady growth, and they did a lot of legwork on the ground here in SF to do that. ⁓ but then as AI became ⁓ more powerful, they started growing like almost like ⁓ hockey sticks, right? Like really exponentially. So I I think that right now the the the landscape is kind of shifting, so I'm not sure what it's gonna be like in a year. But if we talk about a traditional dev tool, ⁓ it's a professional tool that an engineer uses on a daily basis. ⁓ it does not really have a golden path. So if you think about ⁓ th this is gonna gonna be a weird example, but if you think about a Photoshop, what's a golden path inside the Photoshop? inside Photoshop. It's a professional tool where designers spends their whole day what's the golden path? There. It's r it's really hard to say because it's highly customizable and engineers, just like any other professional, ⁓ tend to customize their tools to their liking, right? If you think about to the early day early days of AI, where, let's say like last year at a CephRuby conference, we I asked a bunch of engineers like Joseph Alien, creator of Elixir, or Brent Gessler or some others, James Arthur, about their ⁓ setups. Like what is your AI setup? Guess what? Every single one of them had their custom like ⁓ individual setup that might not work for other people. So this just goes to say that ⁓ developers, the way they interact with the tools is not as linear as an average size product could be, which leads to what you alluded earlier, like how do you actually measure effectively that golden path inside the the product, etc. So ⁓ and and now if we talk about the economics side of it, it's also tricky because now with usage based price and kind of trend things are changing as well. So To be honest, I'm not gonna give you a clear answer because I feel like the the traditional developer tools and AI tools are kind of like morphing and changing the way the industry is gonna work. So ⁓ all I can say for developer tools that ⁓ for years we've seen trend with product-led growth where amazing product, amazing developer experience, open source offering. allowed for steady ⁓ growth and, you know, solid traction over time. I'm not too sure that that same principle is gonna stay with us just because it's super hard to break through the noise and go to market is becoming more tricky by day. So yes, ⁓ I gave a super vague, like all over the place answer, but honestly like I feel like that's where the industry stands right now. speaker-0: Okay. Yeah. ⁓ I I agree that the the floor's moving under our feet here. That's ⁓ I don't think that's controversial. ⁓ I guess what I am interested in is in terms of AI, are we seeing even greater overlap now between what DevTools and SaaS have offered? Like I is that something that's ⁓ meeting as well? Like I you know, what what is ⁓ I don't know, i is is is that speaker-1: Is speaker-0: definition changing. Because it it's like a big part of the blog post that like founders in particular, they're trying to use metrics that SaaS companies have used in order to gauge their own success. ⁓ so, you know, can you speak to that at all or or is that something that would be a little bit more ⁓ in the weeds? speaker-1: ⁓ I can say this. because it's easier than usual to become a builder. ⁓ I think a lot of DevTool companies are gonna change the way they offer their product. Like for example, if you think about companies like Netlify or Superbiz or Versell or whoever else who could be ⁓ adjacent to software creation, right? I'm talking about vibe coding or agentic coding. Those tools need to become more accessible for non-technical folk. And I do envision those remember how I said that in DevTools there is no golden path inside the product. Yeah. Now maybe there is, you know, because if we think about those tools that are adjacent To ⁓ to AI kind of software production, there could be a more of a B2B SaaS. You know, it's at this point, it's like a j like a a write and joke, right? Like B2B SaaS founder, like AI, whatever, Claude. ⁓ but I do kind of welcome that idea because ⁓ anybody can b build an app, you know? Mm-hmm. And and that's a clear golden path to me. Like launch and soft then on the internet. So ⁓ yeah. I do think that they they will merge and and merf and become something new. I also think that there is a trend. Like for example, Canva ⁓ did that I think like a couple of years ago already when they offered kind of like ad hoc payment for a certain thing. Like for example you want to create a poster so you can just buy Canva for one day, you know, build your poster and ⁓ not even create an account. So I do think that DevTools could become something of sorts, like a commodity similar to Canva that anyone could use ad hoc when they need it and then kind of move on and forget about it. So speaker-0: Okay. That is interesting and helpful. ⁓ yeah. Okay. So let's I I I wanna dig into the AI part of this now because you've also been thinking a fair bit about agent experience. ⁓ yeah. So I you know, I had Adam Devander on the podcast recently and he's amazing. And we and we talked a lot about what AX is gonna look like. ⁓ so I'm interested in in your take and like what you're seeing. Like How are you advising ⁓ DevTool companies who say that their competitor is showing up on Claude when they, you know, type in a certain problem? Like, how do you suggest that they are able to become more visible and lead for this this new ⁓ diverse range of tinkerers and folks who are building now hobby projects that are are also ostensibly going to introduce these these tools into the workplace and that sort of P L G motion that has always been so important ⁓ in the DevTools realm. speaker-1: Yeah. Agent experience is a is definitely a hot topic and this term was coined by Matt Bielman, ⁓ th one of the co founders of Netflify. And the idea is that agents are gonna pick tools for you, right? And whoever sets the whoever plays by the rules is gonna win because LLFs are gonna suggest ⁓ to use their tool over others. And I think ⁓ right now it's a little bit more ⁓ complicated than that because it's not only about who LLMs is suggesting, because ⁓ agent experience is not only about kind of like almost like rankings, right, of tools, but it's also if you do specify your code in agent to use a specific tool, are they gonna succeed or not? Right? So it's almost like you're gonna have churn or really disappointed customers if you don't nail agent experience. And we see it not only in DevTools, but also in other like B2B SAS applications where companies are thinking about going headless and they try to develop tooling for their customers to actually be successful when they use agentic coding with their tool, right? So to give you an example This could be ⁓ I don't know, let's say like a shop. or I d it's hard to say 'cause I don't want to to give out any like under NDA details, but basically using MCPs and CLIs to create accessible pathways for your customers to build something custom on top of your solution, right? ⁓ once again, I'm like I'm trying to be careful with not like speaker-0: Yeah, sure. speaker-1: not name any specific names, but so yeah, so when we it's almost like a pen test. Is an agent able to complete the fun the whatever function all the way? How fast can they do it? Are they like is it effective? Is it nine out of ten success rate or I don't know, two out of ten? All of those things are super important. And ⁓ even if you nail agent experience and you're amazing at it, there is a flip side to it. Because if your product is really accessible for agents, it's accessible not only to good agents but bad agents, right? And now when we live in the age of like crazy incidents, that vulnerability ⁓ is becoming an issue for companies. So we have to be like really careful about how we master agent experience. And then to kind of come back to your point, the way LLMs are gonna serve information is also key and it's one of the, you know, ⁓ big discovery channels for your customers. ⁓ so thinking about how LLMs are viewing your product, are they reading documentation? Do they have access to open source projects? How do LLMs in general source their information, right? All of that comes into play and it is gonna be the new kind of like Google Link in I think in a couple of I don't wanna say months, but also not years. Like in a year in a year and a half. I don't know. It's that's good. ⁓ speaker-0: Good. Yeah. I I ⁓ what's good about it for me is that you're you're giving me concrete dates. I mean, I love that. Most ⁓ most folks I talk to are are in the it depends realm. And here you are giving us like in a year. Okay, well we'll check in. That's good. Yeah. okay. So yeah, ⁓ love that description of the importance of agent experience. Is there one thing in particular that you recommend? So you mentioned having an MCP, you mentioned good docs. Like say a product has been around for a while and the way like between versions, there's some bad information going around. And so Claude is saying, hey, ⁓ you know, you need to do it this way or you need this particular library, and that's not true. ⁓ because, you know, the the product has shifted, it's got new features, et cetera, et cetera. So like what you know, for for use ⁓ instances like that, or just in general for like, you know, trying to rank priority of like what it is, what is the most important thing to improve AX? ⁓ you know, what what are you recommending? Like what what is the the top thing to to take stock of ⁓ in terms of like assets ⁓ online that are that are gonna help developers and and these agents that they're using to become aware of your product? speaker-1: Yeah, I think this is gonna vary by case ⁓ a lot. And as you said, I think that like every company has certain baggage that they bring. and typically we try to not give kind of like generalist ideas. ⁓ first of all, check out our agent experience article if you want kind of like a general idea on ⁓ on what to do. And also we have a couple of articles on L L ⁓ optimization for your website and product. I think Those are good to check out. But beyond that, I would say that it really varies and it depends on your strategy. So what are you trying to do? Who's your ICP? Always trying to think about the customer and how what's the smart application of AI for your particular product? How are the people gonna be using it? Are they gonna use it in the terminal? ⁓ like Let me even say this, like at what stage in their life will people resort to your product and where would they go to find that information? So I would almost go back to basics and think about your user journey and how they would be using your dev tool, even if there was no AI. And then with AI thinking about creating shortcuts for them to do it easier, faster, better. And there are obviously metrics that you can put in place to measure that. ⁓ progress and yeah so i i it kinda like it has to be tailored to your product and the reason why I'm being so cautious is because remember how we all lived through that age of like slapping on an AI chatbot to to just show that you have AI? We we don't want that. We don't we don't want that. We want good dogs, we want accessible information, we want smooth developer experience when using your product. So ⁓ yeah. speaker-0: Okay. ⁓ so I'm interested in some of the insights that you have gained from talking with all these dev tool companies. ⁓ so I'm and and you know, d our audience is always interested in how that relates to the developers, right? so what is your sense around hiring right now? Are you seeing a lot of these companies downsizing significantly? ⁓ I mean the most recent one I saw was Redis had ⁓ cut jobs. ⁓ but I mean it's it seems like it's every week. ⁓ So I don't know. How are you advising both founders in terms of the size of their teams? ⁓ and how are you advising developers ⁓ in terms of like their careers? making sure that they're they're going to have a job ⁓ in the next you mentioned a year, things are gonna be changed in terms of cert ⁓ of search in five years, right? Like what sort of advice are you giving about just the the entire software development ⁓ role. Enterprise, startups, the gamut. speaker-1: It's actually interesting. ⁓ so we definitely see layoffs in bigger companies and it's all happening under the same umbrella of, you know, using AI to to 10x production, etc. And I remember for me it was like a big ⁓ kind of like crisis moment when that wave kind of like died down a little bit and people started talking about token maxim and actually token span being like outrageous, right? And we kind of rolled back on that idea that AI is gonna replace humans because it's just not there yet. Then we had that whole like fable situation where ⁓ I'm not even gonna go into that, but basically basically speaker-0: Yeah. speaker-1: where I stand today is an engineer that can use AI to be more efficient is gonna be first a happier engineer. Like those people are gonna stress much less about their future than the ones that choose to not ⁓ adopt AI. ⁓ they're gonna be super valuable and they're almost more valuable than other engineers at this stage. ⁓ and companies are gonna seek them. And companies are gonna be willing to pay a lot of money for engineers who really know ⁓ how to use AI. ⁓ and when I say ⁓ know how to use AI, I mean sustainably, not just creating a one-off kind of like quick ⁓ solution, but building something that could become a foundation for many years to come. And I think it's really hard to not buy into certain narratives that are happening online where people are too optimistic. and you know there is that whole like software factory story and you know there's a lot going on ⁓ on X. ⁓ but I would say that for each individual engineer, I would just recommend finding your own comfortable way of working with AI. trying to open your mind to it, give it a try. ⁓ talk to your peers about how they use it because for me the the big unlock with AI happened when I actually learned a real life kind of like setup that ⁓ Irina used. at the time it was like Zed in con in conjunction with Claude Code. ⁓ and for me personally it was like it made a huge difference because it ⁓ removed mundane routine tasks that I would spend hours on. So basically the consensus that I see amongst engineers is that wow, I'm finally free to build faster and better and I don't have to spend time on things that I don't enjoy doing. I can focus on my creative tasks that are strategic and just, you know, ⁓ it's kind of like you have an army of of junior engineers that can help you build faster. But you're still the one leading the way. So similar to design, I would say that you have to have taste and strategy to be successful with AI. You have to learn from others on how to do it effectively. and you have to be open to trying new things because new models come out like every so often and ⁓ the experience does drastically change over time. Right. So if we look back at the experience we had a year ago versus now, it's like a huge difference. So if you tried it a year ago, give it a try again. That's ⁓ that's my kind of invitation. ⁓ Okay. Yeah. speaker-0: Yeah, didn't didn't know if you, you know, had any sort of backroom insight into into the hiring 'cause yeah, the you're right. X is I don't know, not the best place for developers to be right now in terms of folks talking about, yeah, not only the the toking maxing hilarity, which I I feel like that wave has sort of crested and died, but ⁓ but more just in terms of like, hey, I replaced my entire team with this, or you know, I'm I'm gonna try to be like a one person unicorn or whatever. So it's just ⁓ I don't know, it it's pretty bleak. speaker-1: I mean, ⁓ I can say this. ⁓ all the top DevTool companies that are in my network are hiring senior talent. speaker-0: To speaker-1: They all are looking for people who know how to use AI. Those are the two facts. ⁓ I do know that people build a lot of software as a one person show, but I'm yet to see a unicorn. I'd love to see that. I'm yet to see that though. speaker-0: Yeah. All right. Okay. So you mentioned design. I do want to talk about that a little bit, ⁓ because that's also a big thing that Evil Martians thinks about. And I myself was in an interactive team back when I was a front-end engineer. And so I interfaced with the UX teams and the designers. And it was a digital marketing agency, so obviously the marketers were the ones telling us what what we should be creating. and ⁓ so the the things you're doing are of deep interest to me professionally. But ⁓ yeah, so one of the things that I I thought was cool is that you there was a whole blog post about how you create illustrations. And that there there was a moment of like using ⁓ of trying to find that sweet spot of like using too much AI and then sort of going back and including the human and like deciding what part of the process was the best place to to incorporate humans. ⁓ so it it ⁓ it just it was very thoughtful. I and I I enjoyed that because I I do feel like design is going to become a huge differentiator when it comes to ⁓ marketing, just in general. ⁓ so I interviewed Wes Boss about Lando Norris's website. He did a whole series of videos about that. and I I but I feel like it goes deeper than that. And so I I felt like the way that you folks are thinking about ⁓ just even illustrating your blog, like and and almost as like a proof point of like how is it that AI is going to change the way that we create visuals and think through UI experiences. Because of course, the other part of the AI X conversation ⁓ is that agents don't need a good UI. Humans do, right? And so there there's also this grumbling that maybe we'll just get rid of These marketing sites in general, because it won't be humans that need to find the docs or whatever. It's all just going to be markdown or something. Like they they can just find the text part. They don't need this visual gooey experience in order to make changes to an application. And I'm just interested in what you're seeing behind the scenes in terms of like how Evil Martians has approached the question of design and AI. speaker-1: Yeah. So we get ⁓ leads and customers that have used agentic coding without a designer and now they are in a situation where it's no longer sustainable. Like it it's just not possible to untangle by themselves and th they struggle with it. So every feature release becomes a pain. ⁓ because it's it's just they have to address so much and ⁓ the product starts to look like AI slob mess just because you know there is no system to it. And there are definitely ways to clean that up and to build a solid foundation like create a design system and maybe design linters that allow you to still use a gentic coding but keep it maintainable and and looking consistent and cohesive ⁓ across the board. so basically my my stance on on this is simple. The operator who runs AI has to have taste and certain design skills to make ⁓ the final product look look and feel good to humans. I'm not gonna speak on the on the agents because I just don't know. Like it has to be ⁓ accessible and scannable for agents, but I'm I'm not sure about visuals. Maybe LLMs are gonna evolve to the point that they will also have taste. I'd love to see that. and they will be judging tools by the looks of them. ⁓ I wanna talk about the illustration for a second because ⁓ so evil Martians, we everyone on the Evil Martians team is writing for the blog. And our blog is quite popular among the engineers and designers. We have about Half a million readers on our blog every year. And we really care about the details. So an illustration for us is like the opening to the article. Sometimes, like recently we published an article about inertia JS and it features Bob Ross and it has like different Bob Ross references inside the article. It just like it's a nice little touch, you know? And because we we are of a Martian, so we have a very telling mascot. It's like a Martian is running after like a human humanoid. ⁓ and it's just very dear to us and it's a funny little character. Th there's there was just a lot of like potential in it, you know, to to use it for our marketing materials, including illustrations ⁓ for the blog. And we thought that like out of all teams, we should be able to nail like one shot AI process for this. And men, we were wrong. Like it was just so hard. ⁓ we tried I mean, we also tried when image generation models were just emerging. So it's a little bit like now it's like water under the bridge, you know, a lot has has changed. But we basically arrived at what I call the last The first and last human mile. So the the operator has to prompt the AI to do it well, potentially do even some preparation work. And then the last mile, the cleanups, the shades, the little tweaks, those are still human. And eventually they will probably become AI, but the models are just not there yet. ⁓ so we see that this hybrid where the operator has to run the command. properly and then finish up the task properly. And if something goes wrong has to correct it along the way. And I think that applies to any application of AI. It has like with coding it's the same, right? Like the way you prompt the coding agents is super important. And then the way you kind of correct it and give it feedback is extremely important as well. So it's all about that human in the loop, at least for now. And yeah, and we were able to s ⁓ actually our editor, Travis, you said something that kinda stuck with me. I want to share this metaphor 'cause I think it's really funny. Like AI is like a wig. If it's a bad wig, like it just makes you ⁓ like it it destroys the trust. Like you all you can think about is that wig, you know? And if the wig is good, you're just not be able to you won't be able to tell. So we We want to do good wigs and operators that do good wigs they have to have taste and a certain level of expertise to to make that happen. So that's where we are. speaker-0: That is the perfect metaphor. I can't tell you how many movies I've watched and been like, that wig is distracting and I can't pay attention anymore. I'm done. This is it. Gotta invest in the wigs. okay. Or otherwise just don't use a wig. Just speaker-1: We want to be the Game of Thrones of of it all, you know, like we want to do good weeks. speaker-0: Yeah. Beautiful. such a good, such a good metaphor. On that note, let's go ahead and wrap up. So Victoria, when folks are interested in hearing more about the exciting stuff you're doing at Evil Martians, or your own personal musings on living in San Francisco, doing all the cool stuff there, your own podcast, whatever, where do you direct them? speaker-1: Yeah, so Evil Martians dot com has everything about Evil Martians. We have a blog which is called Evil Martian Chronicles. We post about once or twice a week. I think it's a it's a great resource for engineers and designers to check out. We also have a YouTube channel which has SF Ruby meetup ⁓ mm recordings, my podcast which is called Deaf Propulsion Labs, ⁓ different ⁓ talks at different conferences that we do ⁓ you know in different corners of the world. So ⁓ Evil Martians on YouTube and Evil Martians dot com are the resources to check. speaker-0: Okay. And you're on X. What is your what's your hand on it? I'm on speaker-1: X. My handle is V Melnikova underscore E N. Okay. ⁓ speaker-0: ⁓ Fantastic. All right. Well, so my name again is Kate Holterhoff. I'm a senior analyst at Red Monk. My guest today is Victoria Melnikova. She is here from Evil Martians. And ⁓ if you have enjoyed this conversation, please like, subscribe, and review the monk cast on your podcast platform of choice. If you're watching us on Red Monk's YouTube channel, please like, subscribe, and engage with us in the comments.