Transcript
[00:00:00] Victoria Melnikova: Hi everyone. Welcome to Dev Propulsion Labs, our podcast about the business of developer tools. I’m your host, Victoria Melnikova. I’m the head of new business at Revel Martians. I’m excited to introduce today’s guest today we have a birthday boy, Abhi Aiyer, CTO, and co-founder at Mastra. Hi Abhi. Thanks for having me.
Happy birthday.
[00:00:27] Abhi Aiyer: Yeah. And thanks for this cake.
[00:00:30] Victoria Melnikova: How are you today?
[00:00:31] Abhi Aiyer: I’m doing great. 34 years young.
[00:00:34] Victoria Melnikova: 34. And going strong. Yeah.
[00:00:36] Abhi Aiyer: In depth tool’s. Age. That’s old as
[00:00:40] Victoria Melnikova: It’s an exciting year for you though.
[00:00:42] Abhi Aiyer: It’s been pretty wild. It’s been pretty wild. Pretty freaking wild this year.
[00:00:45] Victoria Melnikova: So just last week you guys hosted TypeScript Icon, which was your first user conference for Mara.
How did it go?
[00:00:55] Abhi Aiyer: It was super stressful last week. There was just so much to do. [00:01:00] We didn’t hire like conference team or whatever, so all of us were like packing bags. I didn’t pack any bags, but like other people did packing bags for the T-shirts and all the swag.
[00:01:10] Victoria Melnikova: Yeah,
[00:01:11] Abhi Aiyer: running around with our heads cut off.
But then once we got there, everything kind of like came to place. Some of our speakers were late.
[00:01:19] Victoria Melnikova: Yes.
[00:01:19] Abhi Aiyer: And then we had to like move the schedule around and, but everyone came on time and everything went well. And then people said they had a great time. So I think it was a success.
[00:01:28] Victoria Melnikova: I mean, beyond all the conference preparations, you also launched a book, you did a rebrand.
You what else? Like what, what are some important announcements that you made?
[00:01:38] Abhi Aiyer: I think four or five or something. We weren’t going to, we didn’t want it to be like a mast show. Yeah. For the whole conference. Yeah. So we just took it from the keynote. And we released a bunch of stuff, released our 1.0 beta.
Mm-hmm. So now we’re kind of shaping Mara up to be like a like more formal and less breaking and that. Then we released our, this [00:02:00] thing called the UI Dojo. It’s just our commitment. We’ve got some people have thrown shade that we don’t want to like be compatible with the UI frameworks. Mm-hmm. Mm-hmm. Or we’re not very well compatible.
Mm-hmm. So that we publish the UI dojo. It’s a thing that essentially is like a more of like a promise to our, everyone involved that Yeah. We can keep making this stuff work with any UI library.
[00:02:22] Victoria Melnikova: Mm-hmm.
[00:02:23] Abhi Aiyer: Then we released a new brand. We have like a new agent logo, which is this, or that’s our Mastro logo, and then, so that’s like an m.
Then we have agent logo. Yes. Which is, people are calling it the splat.
[00:02:35] Victoria Melnikova: Mm. I don’t know how much
[00:02:36] Abhi Aiyer: I like that yet. And then a new design system that goes along with that. Yes. And then, yeah, and then, well, we also launched, we’re gonna do another Mastra based conference in April.
[00:02:48] Victoria Melnikova: Nice. So
[00:02:48] Abhi Aiyer: that’s for more of the community.
Like nice, like true community members. Yeah.
[00:02:52] Victoria Melnikova: Yeah. So Mastras marketing since the beginning has been very much community focused. Right. So. Shane and Sam and [00:03:00] you and other folks at Mara have been doing a lot of educational work as to how to build agents. Mm-hmm. The book that Sam wrote you guys released and based a lot of your marketing on is about, it’s the principles of building AI agents.
Right. This educational layer that Mastra sits on is very important for your adoption, the open source adoption.
[00:03:24] Abhi Aiyer: I think with any framework that’s operating in like a new tech space. Yeah, or like a language or whatever. Education is like a pillar. Yeah. You know, documentation as well, like a good parallel is like when GraphQL came out or any of these new like language paradigms or tech things or whatever.
There’s a whole bunch of people who are not working in this stuff at all. Mm-hmm. And they’re not gonna, they’re gonna be here in a couple years. And then there are people who are starting, who don’t necessarily need education because they’re, you know, like the, they’re trailblazers or whatever. Yes.
They’re not the [00:04:00] ones who it really matters for. They AI’s moving so fast where the trailblazers are going, and then the people who aren’t gonna come to the product yet are already here. Education is like that missing gap. So we do workshops, we do the book, you know, YouTube videos. Yes. All this stuff.
Right? And it’s really important. It used to be like, you’d have to be, you’d be like an education course platform type of person. Yeah. To do all this. But now I think if you’re building a dev tool, you have to do it as well.
[00:04:28] Victoria Melnikova: So. It does reflect on the open source conversion because you are teaching people how to, it’s basically grassroots marketing where you teach them from scratch how to build AI agents using TypeScript.
How does that convert to paying customers too?
[00:04:45] Abhi Aiyer: The book has been so crazy because it goes viral.
[00:04:48] Victoria Melnikova: Yes.
[00:04:48] Abhi Aiyer: And people that are not even using JavaScript are liking in the book, so that’s cool. Yeah. Then there are people who work at big companies. That aren’t even in San Francisco. They’ll go to like some convention or whatever, [00:05:00] they’ll find the book.
‘cause we’re doing all that gorilla marketing stuff. And so these big names are coming from into the funnel from the book. Like Monster itself being good is like a secondary from their, their into the funnel. So that’s one sign. And then the developers are. Finding the framework and the book is auxiliary, right?
Yes. They’re finding the framework and then the education is auxiliary. So it’s kind of interesting. It’s coming from both perspectives. Mm-hmm.
[00:05:24] Victoria Melnikova: I recently went to Ted AI in San Francisco, and it’s not a typical developer conference for San Francisco. Mm-hmm. It’s something that’s, I would say, more enterprising.
And of course I saw a stack of, of your books there, which makes total sense, you know? So now you launch the second book, is that. Does it cover new things, more complex things? Is it the next step? For the first one,
[00:05:48] Abhi Aiyer: we wanted to write the second book because the principles were essentially all the primitives of AI engineering.
Yeah. Like what it, and it’ll always change, I’m sure, but it’s like the base level introduction of like what’s, [00:06:00] what is AI engineering compared to software engineering traditionally, or whatever. And what are these like new concepts that people need to know? Yeah. Because there’s all these new concepts and people make new words all the time.
[00:06:10] Victoria Melnikova: Yeah.
[00:06:10] Abhi Aiyer: And so there needs to be some type of dictionary, right? Yeah. The second thing is you start playing with these tools, and then the next beautiful question that we get asked is, how do I do this? Yeah. Thing like, okay, I know what an agent is, but like I’m trying to do this. I’m trying to like read my emails and then synthesize like follow up reports and all this stuff.
Once you have all the Lego blocks, you need to figure out how to put ‘em together. Mm-hmm. And that’s what the second book is called, patterns.
[00:06:33] Victoria Melnikova: Mm-hmm.
[00:06:33] Abhi Aiyer: And it’s just pretty much just like recipes, I would say, like is the best way of putting it. Like just different strategies on what you should be doing.
[00:06:41] Victoria Melnikova: Interesting. It still
[00:06:41] Abhi Aiyer: isn’t very like opinionated on like how you do it.
[00:06:45] Victoria Melnikova: Yeah.
[00:06:45] Abhi Aiyer: But they’re just more like methods.
[00:06:46] Victoria Melnikova: Mm-hmm. Interesting. And I was like to sit at the Mastra table at the after party, after the conference, and we talked a bit about like how people are using the Mastra and what, what is their first experience like?[00:07:00]
At Martians, we’ve had customers who try to optimize time to value within the product. And I think it’s especially relevant for new tools, AI tools because when you come into an empty product and you don’t understand the, the slang words that, you know, the new terms that are industry wide, how do you even like understand what’s going on?
You know? So understanding those principles and having those recipes, as you said, is instrumental to people being successful with Mastra. Right? Yeah. Do you find that it’s important for you to optimize onboarding and developer experience and all of those things? Like do they actually matter?
[00:07:38] Abhi Aiyer: They’re as important as Mara itself, the, like the documentation getting started.
Yeah. Even all the guides, any written content is as equal as like a bug fix, right? Like yeah, they are so important. ‘cause we get roasted on our docks all the time from the community, like. Most dev tools, founders hate writing docs.
[00:07:59] Victoria Melnikova: Yeah. I
[00:07:59] Abhi Aiyer: fucking [00:08:00] hate it too. But you gotta think about it like this. And this is how it made me feel okay about it, that I never care about documentation.
Uh, when people bitch at us for documentation. If your documentation is good, that’s one less person that want has to talk to you about anything. Yeah. One less GitHub issue, one less thing, and the whole cycle of management. And you are just being mad about docs, dude. Like, just fix the docs so they never have to talk to you about it.
Also when people like complain about features or like bugs or whatever. One person, maybe it’s a coincidence, right? Mm-hmm. But if multiple people would be saying the same thing,
[00:08:35] Victoria Melnikova: maybe you should fix that. Yeah.
[00:08:36] Abhi Aiyer: Maybe you should fix that so then you don’t have to talk to them again.
[00:08:39] Victoria Melnikova: Yeah.
[00:08:39] Abhi Aiyer: So just optimize for not talking to people.
Yeah. That’s what I do.
[00:08:44] Victoria Melnikova: That’s good. Do you ever take it personally? No. Like when people roast Mastra or dogs or whatever it is,
[00:08:51] Abhi Aiyer: I don’t take it personally anymore. Like, because this is not my first rodeo, right? I don’t take it personally, but I also try to under like, try to understand where they’re coming from.[00:09:00]
Why are they so mad about something? Because like, I didn’t, you got this for free. Yeah. You know, like it’s not like I wanted the bug there. Yeah. I think they think that I wanted the bug. Like I purposely wrote this bug for you, so you would come and yell at me. And then when I start thinking like that, that’s how ridiculous that sounds.
I don’t take it personally, you know? Yeah. There’s like so much weird stuff in dev tooling and like social media and stuff. Mm-hmm. You know, there, there’s this thing where they’re like, oh, if you have great marketing, there’s, you probably have zero bugs and you’re probably a perfect product. And you’re like, because you have this traction, you’re so perfect.
And when you’re not, you break there. Their understanding or expectation. They’re like, oh, well look at these guys. You know, they don’t have any marketing, but they don’t have the same bug. It’s like, that’s not how it works. Right. You move fast and you know, there’s just things you have to improve all the time.
[00:09:49] Victoria Melnikova: Yeah. So Mastra appeared when the AI wave just started. Let me put it like this. Like you graduated YC Sea March. [00:10:00] In March. Yeah. So in March you raised the sea round too. Mm-hmm. Right. So you were like ahead of the curve almost. Do you feel that pressure? Like do you feel market pushing you to move as fast as possible?
Like if not for this trend, like if Mastra happened three years later, would you take the same approach?
[00:10:19] Abhi Aiyer: So moving fast, we call it at Mastra, we call it making moves, right? Mm-hmm. Like making moves is our number one principle.
[00:10:24] Victoria Melnikova: Mm-hmm.
[00:10:25] Abhi Aiyer: So whether it’s AI or anything like that is the number one thing is to move quickly.
Yep. And uh, then. We move quick, you ship something that’s mediocre that you probably are embarrassed of. Iterate, iterate, iterate, make perfect, and that that itself, that is the three-year journey Yeah. Is to do that. Yeah. You know, and then you have to, you know, catch things at the right times, et cetera. For us, we got super lucky because we were not building an AI in 2024, but we were adjacent to people who were like in big SaaS companies.
And like there, there’s, there’s a dude at every [00:11:00] SaaS company that’s like the explorer, you know? Mm-hmm. Maybe they’re a principal engineer or whatever, and I, I know a couple that their job to, at the big company is to just explore stuff. Mm-hmm. And like Tinker, you know, he’s a tinkerer. Which I guess is why they call it AI tinkerers met them.
They were like, Hey, you seen GBT? That’s pretty sick. It’s like, yo, show it to me. I’m, I’m like, yeah, we’re, we’re working on some pet projects here, which will never get produced and no, never gets shared on Twitter. Yeah, they’re not allowed to. So you get to know people.
[00:11:28] Victoria Melnikova: Yeah.
[00:11:28] Abhi Aiyer: They’re like, oh yeah, this AI stuff’s pretty cool.
Like you should play with it.
[00:11:31] Victoria Melnikova: Mm-hmm.
[00:11:32] Abhi Aiyer: We wanted to build an AI product as well. We were struggling heavily just doing it. With the confidence and accuracy.
[00:11:40] Victoria Melnikova: Yeah. We’re
[00:11:41] Abhi Aiyer: using all types of tools. Yeah. And we also were trying to learn things as we go. Yeah. And when we kind of came to the, the point where we need to write a framework for this because mm-hmm.
This kind of, this like the landscape for this sucks. We fortunately got into YC at the AI agents batch, essentially. Yes.
[00:11:59] Victoria Melnikova: Yes.
[00:11:59] Abhi Aiyer: And [00:12:00] that’s when it, that’s how it feels like. Oh, it started when AI agents stuck. Yeah. Because the YC was like, this is the batch for all the agents and then boom, and then Sam Altman’s like year of agents and it’s 2025.
Yeah. Just starting. So it feels like we were there right at the right time.
[00:12:15] Victoria Melnikova: I mean, it sounds wild and I had some previously on my podcast. We talked about the batch, kind of crosspollinate each each other. You would basically sit with other YC founders and talk about, okay, try Mastra, try to build something with it and get feedback from real people trying to use it right away.
[00:12:35] Abhi Aiyer: So this is my, uh, this is some insider baseball here. This is my perspective of what happened during YC to us, because we are old as hell. Yeah. Me and my co-founders are old. And there are other people that are old too. Mm-hmm. So isn’t we, like, we weren’t alone or anything, but the first day of yc, like we just sat at this, like they have a cafeteria, like it’s like Hogwarts, you know?
Like there’s a big hall and you’re sitting there, you’re just chilling. [00:13:00] And it’s just funny because like. Everyone was like, people started talking to each other. Yeah. But the thing that they were saying in the like first day is like, oh hey, what’s up? I’m from Stanford. You know, like, where are you from? Oh yeah.
Coming in from Yale. How’s it going? And then we’re just sitting in the back not talking to anyone because like all these like younger people are just like fraternizing about like, oh yeah, I’m 24, just got up from meta. It’s like, okay, cool. And then they put us in these like groups of like companies that.
In the similar like space. Yeah. So we got, we are in the dev tools section. Yeah. But everyone in the dev tools section was like an adult and we had a couple like Gen Z kids or like 21 year olds, but everyone was like 30 plus many people had children.
[00:13:43] Victoria Melnikova: Mm-hmm.
[00:13:44] Abhi Aiyer: It’s almost like they put us there on purpose, you know, because they’re all the old people together.
But that was really good because we are all, most of those guys were like first time founders.
[00:13:53] Victoria Melnikova: Yeah.
[00:13:54] Abhi Aiyer: And they have kids.
[00:13:55] Victoria Melnikova: Yes. And
[00:13:55] Abhi Aiyer: you know how. Like having a kid is a startup in itself, right? Yes. And [00:14:00] that that’s already stressful. And then you’re in yc, which is even stressful. Yes. And then raising money and so like it was nice to be in a group like that.
We also had a lot of experience. Both life and my co-founders have children. We’ve done startups before, so that was a really good friendships we were able to build.
[00:14:17] Victoria Melnikova: I’ve heard the sentiment that. YC batches are getting younger and younger. It used to be college graduates or dropouts. Now it’s high schoolers.
I mean, the technology allows for that. Right? You can build the barrier to entry is much lower now. Yeah. For building anything. Mm-hmm. And with Mastra, you’re also enabling people to build like applications and different things, you know, with agents like super early. Yeah. It’s one thing to come to YC with zero experience.
It’s another thing to come as a seasoned deaf tool builder.
[00:14:52] Abhi Aiyer: Yeah.
[00:14:53] Victoria Melnikova: Was it still as insightful for you
[00:14:56] Abhi Aiyer: Before YC started and before we even got in, [00:15:00] Masra was not the Masra that we wrote. It was like this really N eight N style agent builder thing that honestly, OpenAI has that right now. Yeah. Which is really funny.
And it looks like that. Maybe that was a good idea.
[00:15:13] Victoria Melnikova: Mm-hmm.
[00:15:14] Abhi Aiyer: But at the time, we asked a bunch of friends and enemies, you gotta ask your enemies too. So we asked our friends and enemies what they thought, and all of them said, oh, that’s interesting. That’s cool.
[00:15:25] Victoria Melnikova: That’s
[00:15:27] Abhi Aiyer: not what you wanna hear. That’s not what you want to hear.
You wanna say it? You wanna hear actual feedback, like they hate it or,
[00:15:31] Victoria Melnikova: yeah,
[00:15:31] Abhi Aiyer: look where you’re going. But you know, yeah. If they just say like, that’s cool. That really just means go yourself. It’s like when, I hate when people thumbs up my messages too. It’s like the same, you couldn’t say yes dude. You didn’t have the ability to just type three letters.
You had to gimme a thumbs up.
[00:15:47] Victoria Melnikova: Yeah,
[00:15:47] Abhi Aiyer: same vein. So I was like distraught.
[00:15:49] Victoria Melnikova: Yeah.
[00:15:50] Abhi Aiyer: As you could tell. But then I was thinking like, why do they hate this? Didn’t give feedback. I had to like think about it.
[00:15:57] Victoria Melnikova: Mm-hmm.
[00:15:57] Abhi Aiyer: What about them? And this don’t make [00:16:00] sense. All of these people that we talked to, they wanted a control of the code.
[00:16:04] Victoria Melnikova: And
[00:16:04] Abhi Aiyer: they didn’t like N eight N, they didn’t like stacked or whatever the hell those products are. And they mentioned in conversation that they don’t know about, like, oh, I don’t know if that’s gonna work. So that gave us like the inspirations, like, Hey, maybe we should just say, screw this. Let’s go full code, full control and see what happens.
And then actually worked, you know? Mm-hmm. That was kind of the vibe. And then it was because it was JavaScript and it was actually good JavaScript, unlike, you know, some people, but like. And then things went from there.
[00:16:34] Victoria Melnikova: Why did you choose TypeScript?
[00:16:36] Abhi Aiyer: We’re TypeScript builders for, since forever, our whole career.
So Gaby was in TypeScript? Yeah. And then now this
[00:16:45] Victoria Melnikova: and TypeScript fits perfectly for the job. Or there’s some things that are not ideal. TypeScript
[00:16:50] Abhi Aiyer: fits perfectly for the job. What we realized at the time was at first that when the Python devs were using ai, yeah. [00:17:00] My JavaScript brain was like, oh, snap.
There’s probably some magic stuff in Python land that allows them to do this stuff. Because traditionally, ML and all these things have been like only libraries available on Python with certain like architectures. Yes. For the computer. Yes. You know? Yes. But then I look into all this stuff and it’s like you’re making an HTTP call to open ai.
Okay. This doesn’t need to be any, this could be anything, right? Yes. It’s just h, it’s API call. So from that perspective, like that was super validating. Like yeah, there’s nothing special here, but a HT TP call JavaScript developers already do HGTP calls. Yeah. And we have better packages for a lot of stuff.
Like, why don’t we do, why don’t we write this there?
[00:17:42] Victoria Melnikova: Why do you think Python is so popular for ai?
[00:17:45] Abhi Aiyer: I think it’s just the history, right? Mm-hmm. Like data science, big data. Yeah. All that stuff was always written by Python devs. And just, that’s how everything kind of escalated. Mm-hmm. Web applications and application development has [00:18:00] increasingly gone into other non Python stuff.
Right? Yeah. I remember when I started like Jenga was this cool and it was, I’m not trying to hate on Django, and Flask was tight and people were building full stack apps with Python and you just get some index I html and you’re good to go. But then as we know, rails is way better. Right. Shout out there, but Rails is way better.
Yeah. ‘cause it’s like a framework of like, it’s a, it’s a, it’s an application framework.
[00:18:26] Victoria Melnikova: Mm-hmm.
[00:18:26] Abhi Aiyer: In, in Ruby. Right. And that has all these primitives that you need. Okay. Cool. And then Node has all the server frameworks that exist there. So it’s not like we reinvented anything. We just brought calling HTP calls in TypeScript land now.
[00:18:40] Victoria Melnikova: Yeah. You know? Do you think that like, as AI evolves and grows, there’s gonna be a variety of things? Like is it gonna be still Python dominated or There’s room to to,
[00:18:53] Abhi Aiyer: so as of last week, typescripts the most popular language on GitHub. Wow. Which is [00:19:00] interesting.
[00:19:00] Victoria Melnikova: Mm-hmm.
[00:19:01] Abhi Aiyer: I don’t know what that means, but it’s just interesting.
That’s like a rise, you know? Yeah. And then maybe there’ll be some metrics on like how many agents are built in TypeScript versus Python. Yeah. I bet you right now it’s very much Python.
[00:19:13] Victoria Melnikova: Mm-hmm. So that
[00:19:14] Abhi Aiyer: might increase.
[00:19:15] Victoria Melnikova: Mm-hmm.
[00:19:15] Abhi Aiyer: The beauty is like any web developer listening or out there mm-hmm. Using React, you’ll soon be using all this AI stuff.
[00:19:22] Victoria Melnikova: Mm-hmm.
[00:19:23] Abhi Aiyer: And it then it’s gonna be, you’ll be building agents. No. Like, no problem. And ma, many people will forget that Python was O many or a dominant thing. Yeah, because Typescripts already here. Yeah. So you don’t really care. Yeah. You know.
[00:19:36] Victoria Melnikova: Okay. Let’s talk about YC again. Okay. So you graduated yc. Then you did a round, and you did an interesting thing.
You did a really big cap. You had 120 plus people participate in your fundraise. Why is that? Like, was it difficult to orchestrate and you had some people like Paul Grham and, you know, yeah, we had the homie and like all the, all the def to [00:20:00] stars.
[00:20:00] Abhi Aiyer: So when we were doing the raise, we all, like, me and my co-founders, we all asked ourselves like, why are we doing a company?
And we like, we were like, Hey, let’s just write down three things, like why are we doing this? And like, what’s the goal? Right? So for me, I wrote down, I wanna work with my friends. I want to make moves on anything. And I want to get Rich James, I’m not gonna say his, but his funniest one. He is like, he wants to be the richest guy in South Dakota.
[00:20:29] Victoria Melnikova: That’s a good goal. It’s not hard to do, probably. No, I’m joking. I’m, I’m joking. I’m joking.
[00:20:34] Abhi Aiyer: He’s not there yet. You know, and then making moves is like love of the game, right? Yeah. Like I love doing this stuff, making money. That’s like a byproduct
[00:20:42] Victoria Melnikova: of it. Yeah.
[00:20:43] Abhi Aiyer: Working with my friends. That’s definitely. What I wanted to do and that, and when we looked at all of our goals, it was like, wow.
We’re all like very aligned on
[00:20:51] Victoria Melnikova: Yeah. This,
[00:20:53] Abhi Aiyer: in our past, we’ve had bad experiences with different people and we are all builders and [00:21:00] we’re we’re, we were about to go on this whole campaign about builders need to be in TypeScript. Builders. Builders. Builders. Yeah. So we thought like we should raise all of our money from builders.
Mm-hmm.
[00:21:11] Victoria Melnikova: If
[00:21:11] Abhi Aiyer: you didn’t write a lick of code in your life. You don’t, shouldn’t talk to us or give us money or anything. So then we tried to do a crazy coalition round, like it’s usually called party rounds.
[00:21:22] Victoria Melnikova: Yeah,
[00:21:22] Abhi Aiyer: we did the coalition round. Got every, you know, your favorite founder’s, favorite founder type of founders in there.
Yeah. And that’s, that was a lot of it. And then we did get a lead investor.
[00:21:33] Victoria Melnikova: Yeah.
[00:21:33] Abhi Aiyer: Through Grady and Ventures. They gave us some cash, but then everyone else is pretty much founders. Or like, you know, the, like Orange Collective was a huge reason we got even got started. Mm-hmm. And they’re like a YC like series seed and series A type company or investors.
[00:21:51] Victoria Melnikova: Mm-hmm.
[00:21:51] Abhi Aiyer: So it all came like two people who are in the same mindset as us. Yes. And our cap table definitely reflects it. Yeah. It wasn’t because we, we [00:22:00] lucked into it. It was definitely like a, a choice.
[00:22:02] Victoria Melnikova: Mm-hmm. Do you think it’s gonna get easier to raise from builders as you grow?
[00:22:07] Abhi Aiyer: Probably not, but. We can try, you know?
Yeah. It’d be cool to do like a series A from all builders.
[00:22:15] Victoria Melnikova: I mean, possible. It is possible,
[00:22:18] Abhi Aiyer: yeah. Anything’s possible in the valley.
[00:22:21] Victoria Melnikova: So you’re talking about working with your friends, and I know that you brought a few people from Gatsby over to Masra that are also your friends. Mm-hmm. Can you describe like a perfect ma Australian?
Who is that like your perfect coworker? I’m
[00:22:37] Abhi Aiyer: Australian. I’m Australian. We gotta work on what that is. I don’t even know what to call this. A perfect ma Australian is someone who cares deeply about what they’re doing, who can talk and take it and give it back. ‘cause we’re all shooting the shit all the time and you know, nothing serious.
Mm-hmm. From that perspective. Who can [00:23:00] roll with uncertainty. Mm-hmm. That happens a lot in our business, you know, and then preferably someone who enjoys, like everybody else. When I started recruiting the team, it was very much like people that had relationships with me.
[00:23:15] Victoria Melnikova: Yeah.
[00:23:16] Abhi Aiyer: And like I had like a network here and then a network here, and then I brought them both into the company.
Mm-hmm. The next thing you have to do is make sure that they link up.
[00:23:23] Victoria Melnikova: Yes.
[00:23:24] Abhi Aiyer: But hopefully if you’ve been making friends properly. That stuff can take care of itself, you know? Yeah. Yeah. Building the culture that way has been super easy because all these people already know each other mm-hmm. Through this big network diagram where I’m the, the connector.
[00:23:38] Victoria Melnikova: Yeah. You know,
[00:23:38] Abhi Aiyer: and we’ve hired people’s brothers
[00:23:40] Victoria Melnikova: Yeah.
[00:23:40] Abhi Aiyer: Who are also solid and then that’s easier and like the whole family tree or whatever. Yeah.
[00:23:45] Victoria Melnikova: You know? So once again, as you grow, you know, your network is gonna be at capacity at some point. Yeah. Are you worried about how Mastra is gonna change or it’s something that you are not even thinking about at this point?
And when it [00:24:00] comes, it comes.
[00:24:01] Abhi Aiyer: I am worried about it a hundred percent. I still have like eight more people on my current list, so that’s pretty good. Until then, or then we’ll be screwed. Right. But then also I’m already recruiting.
[00:24:13] Victoria Melnikova: Mm-hmm.
[00:24:14] Abhi Aiyer: In the sense that the way I recruit is just meet really talented people.
Yep. And not try to recruit them. And then see if the stars align. I, you know, I do all this friendship posts and stuff. Yes, those are all recruiting. You know, at some level it’s all recruiting because if you wanna work with me, then you gotta know who I am. I’m not some like exec guy. I’m me. You know, I have my temperaments as well.
You gotta know who I am because we’re about to go into the freaking dungeon and like code for hours. Like we gotta be cool with this. So that’s been, that’s been really good. So I’ve met a lot of, like young people. Yeah. People who are still at big companies and I don’t push them at all. Just like, Hey, yeah, what are you working on?
Sick. If you need anything, lemme know. And then, yeah, see how it goes. I don’t know how it goes and I’m not recruited [00:25:00] yet, but just throwing feelers everywhere.
[00:25:02] Victoria Melnikova: It’s so cool. And I would say unusual for A CTO or like, you know, really. High level engineer to be so extroverted like you, I, it honestly doesn’t happen often and I think it’s a very unique combination because it allows you to create this network, you know, where that you can tap into as you evolve.
For you personally, what’s the toughest part of the job?
[00:25:29] Abhi Aiyer: I was actually talking to, to Shane about this the other day. The toughest part of the job is feeling like you are not doing anything. Like, that’s also like you’re not doing the thing that got you here, right? Mm. Like when you’re like a top engineer or whatever, the thing that got you here is making moves and shipping bunches of code and like doing that for me to go to the next level, I can’t be doing that.
I actually have to be like multiplying my insights and intuition across the whole team. So that thousands of [00:26:00] lines of if we’re measuring based on codes, thousands of lines of code are being generated. Oftentimes when people are like, when you got everyone in their cycles, right? Yeah. Like you’re doing it, you sit back and you’re like playing chess and then you think like, shouldn’t I be doing some shit?
[00:26:14] Victoria Melnikova: Yeah. But you’re really
[00:26:15] Abhi Aiyer: just puppet mastering this whole situation. And I always struggle. Then I start like new projects. Yeah. By myself and then, then I get screwed. ‘cause I stopped watching all the things I’m that I was watching and then it’s like this whole like,
[00:26:26] Victoria Melnikova: yeah. In my previous episodes, I’ve heard the same thing from founders, especially technical founders, you know, because at some point your ability to write code is just not, yeah.
What’s gonna take your company into the next stage, you know, you have to do some other more strategic things. At some point you will have engineers that are stronger than you.
Yeah.
Right? So like you don’t want to undermine the product by committing your code to the, I mean, I’m not saying that. But it’s like hypothetically, at some point it’s gonna happen, you know?
So
[00:26:59] Abhi Aiyer: what I do [00:27:00] is I prototype what I want. Mm-hmm. And then I’ll share like my prototypes and ideas with my team.
[00:27:06] Victoria Melnikova: Yeah.
[00:27:06] Abhi Aiyer: And I tell them that this code sucks. Don’t use it.
[00:27:09] Victoria Melnikova: Yeah.
[00:27:09] Abhi Aiyer: But at least I told you like where we’re headed, you know, people need to feel like they own what they’re building. Mm-hmm. Like if I wrote all the code and just threw it over the fence, like, here you go, bro.
Then they would not have the same passion. Sometimes we solve the same problems twice. Once I solve it by myself and then like so I know what to do, but I’ll give the same problem so they can come to the same conclusion. Mm-hmm. They feel like they came to the conclusion and I’m the guy who like, yeah.
[00:27:35] Victoria Melnikova: Oh my
[00:27:36] Abhi Aiyer: God.
Yeah, yeah. How did you do it? But then, you know. Yeah. But that’s where all like manipulation tactics that Yeah. CTOs use and so that’s been working really well, like giving people ownership. Right. Even, you know, if I don’t trust them, I wouldn’t give it to them. But I do trust all my people on the team, so it’s pretty good.
[00:27:53] Victoria Melnikova: So Mastra operates in, as you said, in this high level of, of uncertainty, right? The the market is [00:28:00] shaping up. Yeah. Like there’s a lot going on. How do you and your team decide on the roadmap? How do you make those goals? ‘cause you mentioned also relying on your intuition a lot. Is it a guess and game? Like, is there some signals from the market that you.
Feel like, for example, people are requesting the same feature again and again, and that’s how you know that it needs to be built? Or how do you make those decisions?
[00:28:26] Abhi Aiyer: It’s like a couple pillars. So one, there’s like net new that people are discovering and that makes our way through news, white papers, et cetera.
If it’s relevant to us, then we start thinking about it. Mm-hmm. In like how it works.
[00:28:42] Victoria Melnikova: Mm-hmm.
[00:28:42] Abhi Aiyer: For us. So if there’s new types of. System prompt strategies that we look at. Yeah. Is it something that it should be built into the framework, or is it be something that is just education? So that’s like one level.
We talk to users every day. Mm-hmm. Shane talks to users [00:29:00] every single hour of every single day, and that helps us get all of our flaws, things that they like, you know? Yeah. They roast us too. The users roast us all the time. That turns into actionable stuff to do. Mm-hmm. Mm-hmm. Our Discord community roasts us all the freaking time that turns into GitHub issues for us to do, to make everything better.
And then lastly, it’s like we have this concept of bold bets. Mm-hmm. Because we’ve been reading about ai, we’re like in this thing every day. There’s just bold bets that we want to make that we have no evidence on, but our gut is like, yeah, yo, we should do this. A lot of our bold bets have come to reality, like mm-hmm.
Like the book was a bold bet.
[00:29:41] Victoria Melnikova: Yes.
[00:29:41] Abhi Aiyer: The conference is a bold bet us making a studio like Mastro ships with a studio. Yeah. And you can play with your agents and stuff. We didn’t want to do that because who wants that? Right. That’s what we were thinking. But it was a bold bet. Like, ah, we don’t know. We feel like it’s right.
Let’s, let’s ship it.
[00:29:57] Victoria Melnikova: Yeah.
[00:29:57] Abhi Aiyer: So stuff like that.
[00:29:58] Victoria Melnikova: And if you think about the [00:30:00] book, if you stopped, like let’s say you wrote a book, you did a couple of. Activations, let’s put it like that. Mm-hmm. And then you would’ve abandoned that it would not take off. Right? No. Like it took a lot, a lot of hard work to actually make it viral.
So it took a while. I would say that it’s, it’s not about being, like, it’s about creating luck. Yeah. Right. Like you build the momentum for the book to become viral.
[00:30:25] Abhi Aiyer: Yeah. We got, we learned from Brad Flora at yc. That generosity breeds luck. This is like saying. And it’s actually one of our principles at Masra too.
Like if you ever meet anyone from Masra, I don’t want to toot our own horn, but if you ask us for help, like we would literally like just help you like, because generosity is like our main thing. Obviously maybe don’t take advantage of us or whatever, but we will help. We will do whatever. Yeah. You want to meet somebody, you want to hang out.
Are you feeling bad? Like you wanna talk to us like anytime. Right? And that’s been like another game changing [00:31:00] difference, which is like. We’re a little snide guys, you know, like we’re sarcastic but like if you wanted to actually talk, like we’re always here to talk, we’re always down to like break bread and that’s helped a lot too.
[00:31:11] Victoria Melnikova: And I also wanted to comment on the user feedback. Something you you mentioned about chain. He actually like reached out to us because we are using Mastra for our own things and it’s so refreshing to have one of the co-founders actually like. Send an email like, Hey, how’s it going? Playing with Mara, let’s chat.
Mm-hmm. It’s just something that you don’t come across often and it feels very personal. Yeah. You know, it’s like we are part of this collective as builders. We’re trying to, to make something good happen, let’s. Help each other out. Yeah. You know, it feels very communal. I didn’t
[00:31:49] Abhi Aiyer: realize how vibrant the dev tools system or your community of builders here is at sf, right?
Yeah. I met so many at the dinner you invited me to. Yeah. And then once you like, it’s kind of like once you [00:32:00] see something then you start seeing it everywhere. Yeah. Is after that dinner, I was just like, oh, you’re working in dev tools. Oh shit. All you guys are working in ductal. It’s awesome.
[00:32:07] Victoria Melnikova: Yeah, yeah, yeah.
The spectrum is, is is very interesting. And because people make trips to SF. For whatever reasons. They come from all over the place and they bring kind of like their cultures and new personalities to the table. It’s very cool because if you think about sf, it’s monogamous. It’s the same type of person everywhere.
Mm-hmm. You know? But I feel like DEF tools is, is just exciting now with AI because there’s so much going on and I see a lot of new blood also coming to the city, you know? Yeah. Younger folks moving in. Trying to build something. So it’s, it, it is very exciting. A couple of questions on the founder wisdom.
So you’re very deep in the AI space. You, you are in a building. If you were to advise somebody who wants to start something new, [00:33:00] what do you think they should focus on? So we’re talking about dev tooling, AI tooling, that space. Yeah. What are some new niche kind of like things that are still missing? I
[00:33:10] Abhi Aiyer: think vision.
Like vision models Yeah. Are an untapped potential. So for example, and I have, we have friends that are doing vision stuff, but like the whole mo, like there’s no, like, no one talks about it. There’s no like, no one’s amped about it. But I like, let me paint the picture for this dev tool, right? Yeah. I have a ring camera.
[00:33:32] Victoria Melnikova: Yes. Right.
[00:33:34] Abhi Aiyer: I want to talk with my ring footage. Why isn’t that? Like who’s working on that? Yeah. There are people working on that, but how come no one’s talking about that? Yeah. Or you go to CTV and you have like someone from the FBI. Hey, like, what? Where did that fool get stabbed? You know? Yeah. I want that stuff.
[00:33:49] Victoria Melnikova: Yeah.
[00:33:49] Abhi Aiyer: And I feel there should be more of that. Yeah. You know, in other areas too, like. Like asset generation, like images and videos and stuff.
[00:33:58] Victoria Melnikova: Yeah.
[00:33:59] Abhi Aiyer: They’re, [00:34:00] they’re like all the big models are doing that. Mm-hmm. But we’re like the open source players, right? There are a couple, but yeah, I think that’s just so exciting because those are the things that if you wanna do something viral
[00:34:10] Victoria Melnikova: Yeah.
[00:34:10] Abhi Aiyer: Make some video thing or like an image thing, right? Mm-hmm. Lastly, it’s like multimodal
[00:34:16] Victoria Melnikova: Yes. Things,
[00:34:17] Abhi Aiyer: right? Yes. We’ve been so text focused this last year. Audio is becoming a thing with all the new, like some products, but still I feel audio is not here yet either.
[00:34:27] Victoria Melnikova: Yeah. Yeah.
[00:34:28] Abhi Aiyer: So that’s interesting. If you’re gonna do rag, you have to do rag over a bunch of different,
[00:34:33] Victoria Melnikova: yeah,
[00:34:33] Abhi Aiyer: you have to do rag over a picture in the future.
Those types of things. There’s so much exciting things outside of text. So that’s the AI stuff on the regular stuff. Couple things. How come there’s no open source? Netlify Versal like thing where you can just like bring your own AWS and then you can just have like a nice UI that can do like deployments and rollbacks.
You would think after [00:35:00] 10 years of this there’d be some like open, open deployment or something.
[00:35:06] Victoria Melnikova: Mm-hmm.
[00:35:07] Abhi Aiyer: And so, yeah, that’s another thing that I think that’s missing. Like, ‘cause I don’t know, it’d be cool.
[00:35:13] Victoria Melnikova: I wonder how they, they would monetize it. Like the se the second they start monetizing it, they become yourself.
[00:35:19] Abhi Aiyer: Yeah. So you have to be sadistic. You have to be like a person who like wants to just see the world burn and then just post the thing. Maybe I should build it, dude.
[00:35:30] Victoria Melnikova: So outside of kind of like outside of the technical stuff mm-hmm. How, what would be your hardest earned advice for a young founder?
[00:35:42] Abhi Aiyer: I feel like I’ve learned everything in my career the hard way, which is unfortunate. But yeah. A couple, couple main lessons that I definitely appreciated. One was your trust is not in the company.
It’s in the people. [00:36:00] You know, your company doesn’t got your back. The people at the company do you know, so you can’t really assume that. You can’t assume anything. You know, you have to like build relationships, make sure you’re Yeah. Protected, et cetera. And that’s like through like just bad company and stuff.
If you don’t act on your ideas, yeah. Someone else is gonna do them and they’re gonna do it faster, better. And you can’t be the person that said like, oh, I had that idea. Yeah. You may have had that idea, you just didn’t do shit about it.
[00:36:30] Victoria Melnikova: Yeah.
[00:36:31] Abhi Aiyer: Anytime anyone says, oh, we should do something. I always say like, when are we gonna do it?
[00:36:36] Victoria Melnikova: Yeah.
[00:36:37] Abhi Aiyer: Well, like should is not in the vocabulary. Also, like I learned this the hard way too. Like if you don’t, like if you focus all on your work Yeah. And your family and stuff, or sorry, your work and not your family or your friends and stuff, your work is gonna suffer. Yeah. Maybe not today, but eventually they’ll suffer.
Eventually your friends will stop asking you on [00:37:00] Friday to come hang out. Yeah. And then you’re gonna be done with that big release you had, or Yeah, you shipped your thing, it’s Friday, but no one hits you up and then you’re all sad and shit. Don’t let that happen to you because then you have to go build up friendships again.
And that’s also a big thing to do, right? Yeah. So I’ve learned that as a, that’s, those are rough.
[00:37:19] Victoria Melnikova: Do you feel like you, you’ve learned from it? Like, do you feel that now you’re in a good space where it’s balanced for you?
[00:37:26] Abhi Aiyer: Yeah, just ‘cause I have, I guess those, those learnings have become like prejudices now.
Mm-hmm. Or I can see the patterns and like it’s, it helps me out. For example, I don’t hire people who are like celebrities or trying to be, you know, because they’re like, well be at your company only for what it’s good for them. Hmm. Stuff like that.
[00:37:44] Victoria Melnikova: Mm-hmm.
Can you think of like the biggest revelation you got while at yc?
Like what was the, the thing you learned and you were like, that makes so much sense.
[00:37:57] Abhi Aiyer: I have three things that were like, [00:38:00] before yc, I was like a yc. Like, who cares? I’m never gonna go. Yeah.
[00:38:05] Victoria Melnikova: I
[00:38:05] Abhi Aiyer: live, I’m from la. I don’t even, who cares about that? And then when we got in, I was like, I gotta leave my home. Yeah.
And come to San Francisco. Yeah. I was talking to my co-founders, I was like, I bet you they’re gonna force us to move here after going through the whole thing, like shipping fast. Mm-hmm. And shipping often, not being scared. I learned that that was a great lesson. Mm-hmm. Like, just not be scared of, don’t be embarrassed.
Like no one gives a shit about you. Like, why do people think? We’re like, who, who’s looking at me? Like no one cares. Right. So that was one thing. Mm-hmm. Second thing is like having a community mm-hmm. Of people doing the same thing and like, like making. Like bonds with the the people. Yeah. That was super nice.
Right? Because if we’re all gonna have to move here, eventually, at least we got started with a social network that can keep like the day to day good. Doing all the fundraising stuff. Yeah. Like we already kind of knew how to do that, but it’s still interesting to see how [00:39:00]
[00:39:00] Victoria Melnikova: mm-hmm.
[00:39:00] Abhi Aiyer: Like what they tell you to do.
And then on the last day of YC, there was a slide they were like. The amount of successful companies that are in the Bay Area, like 99.99% outside the Bay are like 0.0 something. I started laughing. I like laughed out loud and then the people around me like looked, I was like, I’m sorry, I’m sorry. I just laughed because I was right.
You know, they didn’t tell us specifically. Yeah. But that is exactly like what you say. And I remember being there, sitting there laughing and I’m just thinking like, yeah, I guess I agree now. Mm-hmm. So they brainwashed me, dude. But not really. Like I was thinking like, okay, that is probably correct.
[00:39:39] Victoria Melnikova: Yeah.
[00:39:39] Abhi Aiyer: And I wouldn’t wanna leave where like so much stuff is going on.
[00:39:43] Victoria Melnikova: Yeah.
[00:39:43] Abhi Aiyer: Can’t leave now and then, yeah, we stayed around.
[00:39:45] Victoria Melnikova: Are you enjoying this stuff?
[00:39:47] Abhi Aiyer: I wasn’t at first, but after a while you start like just getting used to everything. Yeah. You can respect the charms of the city. I wasn’t leaving the dog patch for like and doing anything.
And once [00:40:00] I started doing that, I started discovering places and it’s a cool city.
[00:40:04] Victoria Melnikova: It is. Yeah.
[00:40:05] Abhi Aiyer: No hating anymore.
[00:40:06] Victoria Melnikova: I actually heard this thing that startups that stay NSF after YC raised twice as much, twice as fast. I
[00:40:15] Abhi Aiyer: think that was on one of the slides. Yeah,
[00:40:18] Victoria Melnikova: it probably was. What do you think Mastra is gonna be like in five years?
[00:40:24] Abhi Aiyer: Hopefully still alive. That’s the first thing
[00:40:26] Victoria Melnikova: that’s, that’s very low of a bond. No,
[00:40:28] Abhi Aiyer: that’s just you. That’s the first thing. I wanna make sure that we’re still alive. I think in five years, if our plans go right, the way we’re envisioning them, we will be very ubiquitous when it comes to building agents.
Mm-hmm. Both for technical and non-technical people, as well as like a lot of these. These like experimentations that are happening in the white papers? Yes. Or like new things that were only Python, like reinforcement learning, for example, things like that are now [00:41:00] available to TypeScript developers and in Masra.
So, mm-hmm. Kind of just keeps expanding from there.
[00:41:06] Victoria Melnikova: So we come to an end of this interview, and my final question is always the same. It’s called warm fuzzies question,
[00:41:13] Abhi Aiyer: warm fuzzies,
[00:41:14] Victoria Melnikova: and it goes like this. What makes you feel great about what you’re doing today?
[00:41:19] Abhi Aiyer: Every user that tweets at us or is in Discord that says that I didn’t know what AI was, but now I built my first agent in five minutes and then now I’m gonna build this next thing.
Makes it all worth, it doesn’t matter about fundraising or anything like that, it’s just users using your is definitely the best part.
[00:41:42] Victoria Melnikova: Finally, I want to provide you the stage to invite people to try Mara. How can they find it?
[00:41:47] Abhi Aiyer: Yeah.
[00:41:48] Victoria Melnikova: What are their first steps
[00:41:49] Abhi Aiyer: to go to mara.ai? You’ll see the NPM create masra, and you’ll get started.
[00:41:55] Victoria Melnikova: And read the books.
[00:41:56] Abhi Aiyer: And read the books.
[00:41:57] Victoria Melnikova: So you can do it well,
[00:41:59] Abhi Aiyer: yeah. Nasher ai slash [00:42:00] book. Also, we have AI Agents Hour, which is our
[00:42:03] Victoria Melnikova: Oh, nice.
[00:42:03] Abhi Aiyer: Funny. Live stream on Mondays with Shane and I, we do a lot of ed like education there. Mm-hmm. And then Thursdays we do a workshop. So tons of things where you can see us.
[00:42:12] Victoria Melnikova: So these are regular, they happen every week.
[00:42:15] Abhi Aiyer: Every week.
[00:42:16] Victoria Melnikova: Awesome. Monday, actually join one of those Thursday.
[00:42:19] Abhi Aiyer: Yeah, that sounds cool.
[00:42:20] Victoria Melnikova: Thank you, Abhi . Thanks
[00:42:21] Abhi Aiyer: for having me.
[00:42:23] Victoria Melnikova: Thank you for catching yet another episode of Dev Propulsion Labs. We at Evil Martians transform growth stage startups into unicorns, build developer tools, and create open source products.
If you are a developer tool needs help with product design development or SRE, visit evil martians.com/devtools. See you in the next.
