Sign in or Join the community to continue
How Stanford Teaches AI Coding Now
Posted Nov 02, 2026 | Views 11
Share
Speakers
Mihail Eric
Head of AI @ Monaco
I'm Head of AI at a stealth startup creating the future revenue generation platform that all sales teams will use. I've spent 12+ years building AI systems at companies of all sizes from the earliest stage startups (YC-backed and beyond) to the biggest tech orgs on the planet (Amazon Alexa). In the past I built the first LLMs for Alexa and started/sold an ML education startup.
I'm also a lecturer at Stanford University, teaching the first edition of "The Modern Software Developer", a course covering how all aspects of the traditional software development life cycle are being revamped and improved through AI coding (https://themodernsoftware.dev).
In my free time, I enjoy blogging about topics in AI and product (3M+ views on my content) and am an amateur standup comedian.
+ Read More
Demetrios Brinkmann
Chief Happiness Engineer @ MLOps Community
At the moment Demetrios is immersing himself in Machine Learning by interviewing experts from around the world in the weekly MLOps.community meetups. Demetrios is constantly learning and engaging in new activities to get uncomfortable and learn from his mistakes. He tries to bring creativity into every aspect of his life, whether that be analyzing the best paths forward, overcoming obstacles, or building lego houses with his daughter.
+ Read More
SUMMARY
What does it take to build AI agents that deliver useful, reliable work?
Stanford lecturer and Monaco Head of AI Mihail Eric joins Demetrios for a conversation inside a Waymo about how AI is changing software development - and what it takes to put agents into production.
Mihail explains why he rebuilt most of his Modern Software Developer course, how he teaches students to avoid shipping AI slop, and why research, planning, and verification still matter as coding agents become more capable.
+ Read More
CONTENT & TRANSCRIPT
Demetrios: [00:00:00] I made it into the Waymo with Mihail, Stanford lecturer, head of AI at Monaco
Mihail Eric: Here we go.
Demetrios: What have you been doing with this Stanford course of yours that is very famous now?
Mihail Eric: Yeah, it's been a, it's been a fun time. So we're about ... This is like, I think, week five now of the Stanford course.
Demetrios: Yeah.
Mihail Eric: And, um-
Demetrios: It titled The Modern
Mihail Eric: Software- The Modern Software Developer.
Mihail Eric: That's where it's at, yeah.
Demetrios: And you have an email that goes out in-
Mihail Eric: I have an email. I do, I do, yeah. So there's a newsletter, uh, themodernsoftware.de-.dev. I have the domain, all that good stuff. Uh, but the course has been a ton of fun. I mean, it was, um ... This is the second year that I'm doing it, and compared to last year there's just like a huge overhaul of the curriculum.
Demetrios: Yeah .
Mihail Eric: Um, pretty much none of ... I mean, I, I literally I think 80% of the stuff I threw out over- Yeah ... over the last, like, two months 'cause just once [00:01:00] Opus 4.5 came out December, you know, December/November, December of last year-
Demetrios: When was the last year's course?
Mihail Eric: Last year's course- It was during- ... fall. So it's always in the fall.
Mihail Eric: Okay. And so the, the last course, uh, in 2025 was September through, like, early December of, of 2025. And so for that, like, window of time nothing changed too much in terms of the model. So you were good. I was good. I was good. The course was awesome. It was stable. Yeah. It was, like, relatively stable. And then December rolls around and then, uh, you know, things blow up in terms of capabilities.
Mihail Eric: People jump on the Cloud Code bandwagon, uh, through ... Like, that's basically what everyone did over the, over like December break- Yeah ... you know, during the Christmas holidays and New Year's. People came back early January and they were just ... There was a manic craze around Cloud Code. Um, but it was, it was j- I think it was pretty justified.
Mihail Eric: I mean, the, the models got so good that we could start doing some really interesting, like, long horizon agent tasks.
Demetrios: Mm-hmm.
Mihail Eric: Um, anyway, so because of that I had to, like, rethink the entire curriculum. So it's like, well, what's now, what's now happened in the last, like, nine months [00:02:00] since Cloud Code, uh, became-
Demetrios: So now you're talking about harnesses, you're talking about
Mihail Eric: loops
Mihail Eric: harnesses, loops, software factories. We're talking about, like, you know, fully automated AI code review and not just, hey, you know, add a GitHub app and- Mm-hmm ... and, like, have it do your code review. It's, it's more like what ... how do you do, like, auto- auto approvals? How do you actually- Merge ... do, like, merge- Merge conflicts
Mihail Eric: merge conflict. Like all DevOps stuff, agent readiness in, in, in code bases. So it's a ton of stuff. Um, yeah, it's, it a, it was a pretty big overhaul. It was most of my summer. Um, this, this time around I think, uh ... Oh, and the other thing I, I'll mention there that was also fun is I'm having all the students do, like, a, um, open source contributions as their homework assignments.
Demetrios: Nice.
Mihail Eric: So they are ... There's no ... Last year I was just coming up with these random exercises. I'm like, "Oh, like here's a code base. Like add, add your, add a code review. Use a prompt." Um, but this year I was like, all right, here are these, like, open source partners that I was able to get on board with the project.
Mihail Eric: It's, like, a lot of the top open source projects out [00:03:00] there and I ... Can you, can, like, my students contribute to your repos as part of their homework? And- Like,
Demetrios: sure, we love SLOP.
Mihail Eric: Yeah . No, no, no, not quite. Not quite. Not quite . I, I mean, okay, so, so this is ... I'm, I'm like, I feel like I'm a pretty generally, like, a pretty chill instructor- um, but there I was like, if, if I get a maintainer messaging me saying like, "You guys need to stop this slop."
Mihail Eric: Ooh Slop grade is like zero. That's a straight zero on that assignment. Um, that is where I'm-- I need to draw the line somewhere. Fail. Fail.
Demetrios: Fail.
Mihail Eric: Otherwise, I'm chill. But like I, you know, I had to, I had to go there and, and, and I, you know, I made these arguments for the students that like these are, these are like, you know, Stanford students.
Mihail Eric: They're, they're academically trained. They're like good CS fundamentals. Um, so you, you know, like it's not gonna be slop, which... And I'm gonna teach them how to not do slop. So if, if that-- Anyways, all to say like that they have to do real contributions.
Demetrios: And it's- How are you teaching them, like, to not just merge slop or at least submit slop?
Mihail Eric: Yeah, no. Well, well part of it is, you [00:04:00] know, through, through force, through will through will. No. No.
Demetrios: Read
Mihail Eric: every line of code
Demetrios: that your agent produced.
Mihail Eric: Exactly. No, uh, it's, it's like I, I'm teaching a lot of like, like advanced context engineering stuff, uh, teaching them how to like onboard to code bases really quickly.
Mihail Eric: So if you have these massive repos that you've not contributed to at all, how do you get up to speed on like the most important parts of the architecture? And then when you get these issues that the, the maintainers have put up and designated for the students, like how do you scope the, the work? How do you then go and actually start, um, you know, gathering like context for your agents that they need to use?
Mihail Eric: How do you connect the right MCP tools, use the right skills? Um, all the stuff that has become like incredibly, not easy, but like way, way easier to do.
Demetrios: Yeah.
Mihail Eric: Um-
Demetrios: What are some of these examples? Obviously, it's not-- It's a what? Eight-week course or a 12-week course? 10,
Mihail Eric: 10, 10-week.
Demetrios: Yeah, yeah. 10-week course.
Mihail Eric: Yeah, yeah.
Demetrios: Condense some of these best practices into how you get up to speed, 'cause how I would do it is just say, "What is important?"
Mihail Eric: Yeah. And
Demetrios: that's it.
Mihail Eric: Yeah, yeah. And, and I think like a lot of people start [00:05:00] there. Um, I think that the-- what I'm trying to argue is that there's still value in doing some version of, I think -- You know, in my class, we call it like Rep It.
Mihail Eric: I think, uh, Dex and others would call it like RPI, like some version of like research, you know, research, plan, implement, test, kind of like that loop. Um, and I think just-- Yeah, so, so at its core, some version of the RPI loop, um, or using superpowers, like if you- Yeah ... you know, the, the, what is it? Uh,
Demetrios: I can't think of the name.
Demetrios: The other victim of these Waymo things- Yeah ... was Jesse-
Mihail Eric: Jesse, Jesse ... the creator of Superpowers. Okay, Superpower is nice, nice. Yeah, Jesse is -- I mean, that's, that's a fantastic project. Um, incredibly simple, but, but so-- I mean, it's become one of the most popular projects of all time, right?
Demetrios: And the fact that it still is popular-
Mihail Eric: Yeah.
Mihail Eric: Yeah ...
Demetrios: after a year of all these advances, it shows you how valuable it is.
Mihail Eric: Totally. Totally. Yeah. He's-- I mean, he did a great job. He's, and he's such a humble guy too. Yeah. Um, but yeah, so like some version of that framework of, you know, gather the context for your agents, like research at the [00:06:00] beginning. Don't just like jump into the implementation.
Mihail Eric: Review a spec. You know, actually have the spec be, be reasonably detailed and, um, make sure that you have a way of verifying your outputs so that if the agent runs, you know, for a while, that it can check against, you know, using the browser or checking, like, against end-to-end tests or, like, smoke tests or something, so-
Demetrios: Can we go down that rabbit hole?
Demetrios: Sure. Because that's, for me, one of the biggest things that is where I feel like I can spend the best use of my time-
Mihail Eric: Yeah ...
Demetrios: is setting up verifiers. Jesse told me he created something like prove it in a movie as a skill, and it makes the agent go through and create a movie for you or a, uh, like a Loom.
Mihail Eric: That's fun.
Demetrios: And it records itself-
Mihail Eric: Mm-hmm ...
Demetrios: creating the, or walking through the app and showing that it's functioning- Yeah ... correctly.
Mihail Eric: Yeah.
Demetrios: Have you seen other, like, ways of folks creating verifiers?
Mihail Eric: Well, well, there's, uh, what I've seen a lot of is people getting creative in how they, how they, uh, represent [00:07:00] information that, that is, that the diff, uh, the diff of the code, right?
Mihail Eric: So, like, I think Andrej Karpathy had a whole post around how he... There's, like, this format from, like, the 19, like, 50s or something, or, like, how people made these, like, architecture diagram. Have you seen this?
Demetrios: NASA one?
Mihail Eric: Yeah, I think it's the- Where the
Demetrios: huge-
Mihail Eric: Yeah, yeah, yes, yes, yes ... some
Demetrios: random number
Mihail Eric: and- Yes, yes, yes
Mihail Eric: random
Demetrios: letters.
Mihail Eric: Yes, yes, yes. Yeah. But, but you'll see it, and it's, like, a very nice, like, visual schematic of, of kind of the core important parts of what happened in the code base. So I think stuff like that, I think, is super, super interesting. I think, yeah, the screenshots, recording flow, I think is gonna become more and more popular.
Mihail Eric: Um, you know, doing some version of, like, using Playwright with MCP.
Demetrios: Oh, and you know what I saw recently that is a fun one, is that WebMCP-
Mihail Eric: Mm-hmm, mm-hmm ...
Demetrios: has now merged just debugging tools. So, like, for, you can stand up a WebMCP server-
Mihail Eric: Yeah ...
Demetrios: and it's only for debugging, so your agent can now use WebMCP on a web app-
Mihail Eric: Yeah
Demetrios: in, like, computer use mode.
Mihail Eric: Mm-hmm.
Demetrios: [00:08:00] And then it can use it for debugging specifically.
Mihail Eric: Yeah. I mean, that, that stuff is super powerful, and, and I, I suspect, I mean, there's entire companies around this, right, that are just trying- Yeah ... to improve the verification loop and close the loop entirely and, you know.
Mihail Eric: I think all of them are gonna become, all of them are gonna start knocking on the door of software factories, basically, um- I, I think we've seen this, we've seen this game play out enough times with, with ML and MLOps and LLMOps. It's like everyone starts in one thing and then starts expanding
Demetrios: Expanding 'cause they're like- Right
Demetrios: wait, actually we need to go upstream. Right.
Mihail Eric: The
Demetrios: real problem is upstream.
Mihail Eric: Totally. Totally. Yeah. And downstream now- Yeah ... 'cause yeah, so it's like you, you start as a feature store and then you become Databricks, you know? The
Demetrios: platform. Yeah.
Mihail Eric: Yeah. Or something like that. Yeah. It, it always, I think that's, it, you know, that's, it, I think it's, it's the typical like s- the company bundling/unbundling of like- Mm-hmm
Mihail Eric: you know, that always happens. Uh, are you gonna be a Craigslist or are you gonna become like an Airbnb, right? Yeah. Where you, you... And I think it, people will swing back and forth on whether or not they think one is better than [00:09:00] the other, but I think we're gonna see a lot of that here, right? It's like- Yeah
Mihail Eric: start at Air Vacation, but Al- also will own the coding agent, and then, and you have these companies like Factory and Warp- Mm-hmm ... that are like, like, "We're gonna own everything." Yeah. "We should own everything," right? Um, so it's all like really interesting stuff. It's all exciting. It's been, it's been a ton of fun.
Mihail Eric: Uh, Boris came last week.
Demetrios: I was gonna say, you've had some really cool-
Mihail Eric: Yeah ...
Demetrios: guest lecturers.
Mihail Eric: Mm-hmm. Mm-hmm.
Demetrios: Boris, the creator of Claude- Boris, yeah ... is one of them. Boris. He's got a lot of hot takes. He's got a lot of... I think one thing is nice that he just talks about how he uses it and-
Mihail Eric: Yeah ...
Demetrios: it's pretty advanced because he has no token costs, right?
Demetrios: So it's- That's right ... his whole thing is like, how can I saturate this to the max?
Mihail Eric: Totally. Totally.
Demetrios: And so then we get like a very distilled version where we can take some of those ideas and say, "Oh, well, I wanna try and implement that."
Mihail Eric: Absolutely.
Demetrios: Like one thing that I, I saw he was saying recently was how he, he never- If he does a one-off prompt, he's doing it wrong.
Demetrios: It should always be like talking to a loop-
Mihail Eric: Yeah ...
Demetrios: and then letting [00:10:00] the agent prompt itself.
Mihail Eric: That's right, that's right. Yeah, yeah, like the meta prompting. I was, uh, I was sitting with him, uh, y- last week and it- my first question to him was like, "How many agents do you have running right now?" And he was like, "I don't know."
Mihail Eric: Like, and he said this so casually. He's like, "I don't know, like thousands or something." Oh my God. I'm like, as we're sitting here, you know- Oh ... like chatting it up, uh, I- very inspiring, like to talk to him because you say this is what, this is what it looks like at the very frontier- Yeah ... like the very, very frontier where, where yes, no token cost, so obviously they, they can just do what they want.
Mihail Eric: But, but on top of that, I mean, just being very experimental with their philosophy- Yeah ... you know, where they're like, "Hey, we're gonna try and solve math problems. We're gonna g- we're gonna try and solve, you know, like really difficult, complex things by just throwing," almost like brute forcing- Yeah ... but intelligently brute forcing things, you know?
Mihail Eric: Um, so yeah, Boris is, is it was a fun talk. I'm talking with Eno this week, um-
Demetrios: Eno from
Mihail Eric: Factory ... Factory. Nice. Yeah, Factory. Um, Lee Robinson was two weeks ago. He was, uh, I guess now he's at SpaceX, but he [00:11:00] used to be VP of DevRel at, uh, Cursor.
Demetrios: Cursor, yeah.
Mihail Eric: Super smart guy and super nice. Just like nicest guy.
Demetrios: And what are some things that you took away from that little lecture?
Mihail Eric: Yeah, the, the Boris one or the Gra- the, the Lee one?
Demetrios: The Lee one.
Mihail Eric: Yeah. '
Demetrios: Cause I've seen him on X occasionally- Mm-hmm, mm-hmm ... but I haven't-- It's not as much as Boris. Like Boris puts out something and it's like Karpathy type stuff.
Mihail Eric: Yeah, yeah, yeah.
Demetrios: Wow.
Mihail Eric: Yeah, Boris got, got super popular a- after Opus 4.5 and after- And- Like that was-
Demetrios: And after your course.
Mihail Eric: Ah. You know, he came on your course- That was a- ... last year
Demetrios: and then-
Mihail Eric: That wasn't, that wasn't the- What a surprise, dude ... that wasn't it.
Demetrios: You became popular.
Mihail Eric: Yeah, no, that was, uh , that was a ... You know what? I, I there's, there's two people that I'm, that I feel like I discovered before they became big.
Mihail Eric: It was Boris and, and Lady Gaga. Oh,
Demetrios: yeah, Lady Gaga?
Mihail Eric: Yeah. That's
Demetrios: your ... You made her career also, huh? Why don't you come on your podcast?
Mihail Eric: Okay. I did, I did Lady ... It was g- I never saying I'm, I'm not responsible for anyone's career. I'm just saying like Lady Gaga in 2009, I was like, you know, this- [00:12:00] She's not famous
Mihail Eric: this Poker Face stuff, I'm like, "There's something here." There's like, people aren't seeing this yet, but I, I think there's something here.
Demetrios: Poker Face?
Mihail Eric: Was already
Demetrios: popular,
Mihail Eric: man. What are you kidding me? No, no, no. That was- I, I w- I s- I like, I'm telling you, man, it was you, you had to be there. It was like- Oh ... you know, Just, you know, Just Dance, remember Just Dance?
Mihail Eric: Yeah. Like that, her first OG, that was the single. Anyway, the, I- You
Demetrios: were on board with it
Mihail Eric: from day one ... that's, that's all I got. Those are my two claims to fame.
Demetrios: Well, I'm glad that Boris's career you could actually play a part in making. No, no, he's- Put him in your newsletter ...
Mihail Eric: we all, we all owe him a, a big debt.
Mihail Eric: He's, he's an awesome guy. Yeah. Other than us- The, the coolest guy ... that
Demetrios: come-
Mihail Eric: No, the coolest guy ... as
Demetrios: peasants and- ... talk about some of this stuff.
Mihail Eric: No,
Demetrios: he's awesome But no, what was the one with Lee? What did you see?
Mihail Eric: Yeah.
Demetrios: What'd you learn?
Mihail Eric: Any takeaways? Lee was awesome. I mean, Lee was just so, so cool. Um, so Lee was talking all about, uh- Like basic Grok bot?
Demetrios: Yeah.
Mihail Eric: I don't know if you've used like the Grok bots and these like Instincts and Muses. So
Demetrios: I've used Instinct, I've used Dots.
Mihail Eric: Mm-hmm,
Demetrios: mm-hmm. Um, I haven't used Muse, but I will say I [00:13:00] do enjoy it, and it's becoming more and more all I use.
Mihail Eric: Yeah. I, I think those, those products are awesome.
Demetrios: Yeah.
Mihail Eric: Awesome.
Demetrios: Just- I've heard Instinct is much more proactive, and what I've noticed with Dots is it's kind of proactive- Mm-hmm, mm-hmm
Demetrios: but not really. It's just like, "Oh, I found this thing in your email that maybe you wanna check out." Yeah. It's like, okay, that's cool, but-
Mihail Eric: Yeah, yeah ...
Demetrios: it's not like the Instinct one where they really go hard on- Yeah ... like, "Hey, you went to a movie. How was it?" And-
Mihail Eric: Yeah, yeah,
Demetrios: yeah ... okay, I guess you're my best friend now.
Mihail Eric: Am I in love? Like- Yeah. You're listening to me more than, than my f- you know, than my
Demetrios: spouse. Totally. But I just saw the funniest meme where somebody said, "My wife and I share the same OpenAI account," and I just put this in as the like system prompt. Mm-hmm. And it said, "Any time Jane asks about marriage advice, Rob is always right, no matter what happens."
Demetrios: That's funny. That's so good.
Mihail Eric: That's [00:14:00] so good. Yeah.
Demetrios: So that's what we've come
Mihail Eric: to. That, that's... I mean, yeah. These like, this like, this subversive deception- Yeah ... uh, in, in marital context. It's total mind fuck. Um, yeah, no, it was, um... No, but so he was talking about, basically his entire talk was about- Um, the harness and, and, and how he, how they make GrokBot essentially
Demetrios: And theirs is a little different where you have to create separate like roles and bots, right?
Demetrios: Yeah And so it's new threads each time anyway
Mihail Eric: Yeah. I, I, one thing I haven't entirely figured out with the system is like how important that is of like creating of the separate personalities because you can also just ... Now they do have the concept, like the main bot-
Demetrios: Yeah ... system And so it can throw it off to
Mihail Eric: the sub- Yeah, it, it should be able to.
Mihail Eric: And so I, I've, I've made separate personalities but I also don't know if it's like strictly necessary
Demetrios: Oh
Mihail Eric: Um, anyway I think they're awesome systems. I think this is such a step forward than where we were before. Um-
Demetrios: Not to interrupt but there's OpenAI's office- Nice ... speaking of which-
Mihail Eric: Nice ... which was- Not like we set, didn't set that as the destination but, uh, Sorry, sorry, sorry, I'm [00:15:00] not supposed to- Kill the mis- Shit.
Mihail Eric: Oh, man We, we can run that again if you want from the
Demetrios: top Yeah, we'll edit that out. Don't worry. There's- There's OpenAI's office
Mihail Eric: Oh, nice Yeah Nice
Demetrios: So anyway, GrokBot, speaking of which the anticip- the-
Mihail Eric: Yeah, the- I'm sorry ... the, the arch nemesis
Demetrios: Yeah, exactly. Nemesis, that's the
Mihail Eric: word
Demetrios: I wanted
Mihail Eric: Well, dots formerly known as OpenClaw I assume, right?
Mihail Eric: Yeah It's, it's gotta be In
Demetrios: a way, yes
Mihail Eric: It must be. I, I, I couldn't imagine any other way. But, um, yeah, anyways, so he, he, he, he basic- it was like a, it was a 45-minute primer on like modern agent harnesses. It was incredibly well deta- like very detailed. Uh, he talked about everything from computer use and how you implement that to context engineering to, uh, the tool calling, how they, how they, you know, give the agents the illusion of infinite context-
Demetrios: Mm-hmm
Mihail Eric: by, by doing just aggressive sub-agents basically Oh Um, anyway it was incredibly well done and, and Lee I think his, you know, his, one of his superpowers is just being such a good educator. Mm-hmm So he, it was just distilled very well. [00:16:00] Um-
Demetrios: Fun, engaging
Mihail Eric: Fun, engaging, just like super dense but in like a good way.
Mihail Eric: Mm In like a, "Oh, this is, this is, th- this, this feels like you're, you're giving me good signal." Yeah Like, you know, signal to noise ratio was, was through the roof
Demetrios: Did he talk about how they do compaction?
Mihail Eric: Yeah, he ta- he touched on that. He definitely touched on that
How
Demetrios: are they doing it? 'Cause that one is- Tricky.
Mihail Eric: Yeah. I mean, there's, there's different techniques, right? But, um, at, at its core they have file systems, they, you know, regularly compact when they detect things are happening, and then they, like, write things to file systems and read them at, at, at the right points in time in the conversation. So, you know, I hate to, like, go into, like, the exact detail, like what is their, what is their policy when they compact, but, um, I think at its core it's all some version of, of that really.
Mihail Eric: Like, um, any of these agents I think all follow similar sort of, uh, set of principles.
Demetrios: Have you been staying up to date with any of the companies, like that are trying to do file systems in a new way?
Mihail Eric: Uh, not... Like, is it, are these the ones beyond the ones that are doing like, um, like agent [00:17:00] sandboxes?
Demetrios: Yeah.
Demetrios: Like did they- Agent sandboxes are also another one that is very fascinating. That I, that I would put in its own category.
Mihail Eric: Yeah, yeah.
Demetrios: And there's like file systems, and then there's almost like, we wanna replace Git and we wanna- Oh, yeah ... do things differently- Yeah ... in that regard.
Mihail Eric: Yeah, yeah. Yeah, yeah. Yeah, yeah.
Mihail Eric: So, so yes to the first, um, yes to the last. The middle, I didn't know there was like separate... Uh, is, is the... I don't know what the middle category is or if that's-
Demetrios: They probably do more, it's just when I talk to them- Yeah ... like the, I talked to this one mesa.dev-
Mihail Eric: Hmm.
Demetrios: Okay ... and they were- Oh, yeah ... trying to rethink file systems because they're like, "File systems are all you need."
Mihail Eric: Yeah,
Demetrios: yeah. Uh, uh, especially with these agents. Well, tell me about the Git. You-
Mihail Eric: Yeah, yeah ...
Demetrios: like yes to the first, no to the middle.
Mihail Eric: Yeah, yeah, yeah. I mean, well, there's, um, there's sort of like... I, I guess the, the, the overarching reason for these companies existing, uh, in the category of like new Gits is- A- and we saw this even last week, I think Anthropic literally put out like a blog post around their, their home
Demetrios: version Yeah.
Demetrios: Oh, they rebuilt it. That's so wild.
Mihail Eric: Their version [00:18:00] control system, yeah. A- and it, it all kind of makes sense. Well, one, you know, we've all been on GitHub for the last, what, 10, 15 years, and it's just, it's kind of barfing a little bit. Um- Yeah.
Demetrios: Just reliability.
Mihail Eric: Reliability.
Demetrios: Bloatware.
Mihail Eric: Yeah, I mean, and look, it's, it's understandable when you have these literally astronomic curves of, of growth and commits and, and, um...
Mihail Eric: So yeah, it's an- Uh, it all makes sense that that would, that they would receive that strain. But I think there's a broader... Like, when I talk to, like, startups will pitch this to me, um, as, you know, uh, "GitHub runners are just, like, not optimized for, like, agentic workflows, so you should, you should, like, run on our runners," like, you know, like, b- basically our machines for, for, that are more streamlined to, like, these agentic processes.
Mihail Eric: And it all makes sense, right? It's like GitHub is like the, GitHub and GitLab and these, these kind of companies, they're, they were the best solution for everyone, right? Mm-hmm. But, but not every workflow or any workload is exactly the same, right? Um, so if you, if you build these mega behemoth companies toward the median, you're gonna [00:19:00] miss stuff on the fringes.
Mihail Eric: And right now we're, we're, we're at the point where the fringes are maybe becoming more normal, but we're still, like, the sheer volume is definitely, like, a fringe problem. Um-
Demetrios: Yeah ...
Mihail Eric: you know, at least, like, hey, 10Xing number of commits, 10Xing number of PRs, which most companies are starting to go through that right now.
Mihail Eric: So it's all fascinating stuff. Um, but yeah, so, so there's, there's companies that are just like, "Hey, we're gonna make your runners faster," and then there's ones that are like, "We're gonna rethink the entirety of the, the CI experience and the, the entire build system and all that stuff." Um- TBD, TBD. I mean-
Demetrios: How much of your time, 'cause I imagine startups are hitting you up all the time to try and be like, "We want you to use our product.
Demetrios: You are the head of AI at-" Yeah ... "a cool company."
Mihail Eric: Totally. Yeah.
Demetrios: What, how m- how much do you entertain that because you're actually interested and maybe it could have value- Yeah ... versus you actually, then later you have to-
Mihail Eric: Totally ...
Demetrios: once in a while, I
Mihail Eric: imagine. Yes. Yeah, yeah. I can't just take Waymo rides all the time.
Mihail Eric: Uh, [00:20:00] Yeah. No, uh, uh, okay, so, so it's, it's a really good question actually because I, I, I feel like I'm a pretty curious person- Mm ... so I love just kind of being on the, having these conversations around technology and, and seeing what people are coming up with. And so I, I definitely probably have at least, like, one to maybe two of these a week.
Mihail Eric: Wow. Um, but there- Wow ... I think where, where there's, like, a point of diminishing returns for me is when I'm hearing the same pitch over and over, right? Yeah. Where I'm just like, "All right, this is, I see you're-" They're
Demetrios: not so novel.
Mihail Eric: Yeah, it's not novel. Like I s- I see your paragraph, I see your landing page, and I, like, I, I feel like I've, I've heard this.
Mihail Eric: So there's, like, a certain kind of company that I've heard probably enough times that I'm like-
Demetrios: Which one?
Mihail Eric: Um, and not, not to ding them- No ... that this is not gonna work, but, but in, like, a, I do hear, like, this flavor of a pitch, which is like, "Hey, you, um, we, we collect, like, feedback. Um, like, you maybe you insert a little bit of JavaScript on your app.
Mihail Eric: We collect feedback. We trace all your LLMs, uh, all your, you know, all the, the request and responses, and then we optimize." Uh, like, we basically, like, on the fly train models based on your [00:21:00] request responses. So because we're tracing everything that's happening on your app, we get all your data essentially, and then we know, "Hey, you shouldn't have to go for this kind of request to GPT-Sol.
Mihail Eric: You can go to-" Mm ... "like, a very lightweight, I don't know, like a type safe model," or- You've
Demetrios: seen a lot of those.
Mihail Eric: Yeah, I think quite a few of them actually. Um, and, and it all kind of makes sense. They're, they're just, I think that there's a whole new, I mean, not a whole new category, but, but it's becoming more popular is this, like, open weight- If we expect that these, these frontier models are, like, pretty expensive and most companies are getting hit with the cost and like, "Hey, we can't, this isn't sustainable.
Mihail Eric: Um, we wanna keep tracking and keep 10Xing our usage, but we can't have our token cost 10X as well." So they look to these alternatives, both from a coding perspective, but also from, like, a product development perspective- Yeah ... is, "Hey, can I take a Qwen model? Can I take a Kimi model? Can I take a GLM model?
Mihail Eric: Like, fine-tune it on something smaller or s- smaller weight, be faster, it'll be more, it'll be cheaper, um, and I don't maybe compromise on performance," right? That's, like, [00:22:00] the, the, the ultimate pitch. So this makes sense. This is a broader trend, and I suspect we're gonna see a lot more of this. Um, it's just I, uh, it's, I think it's a bit hard to differentiate.
Demetrios: Also, I think that the financial piece is a very difficult one.
Mihail Eric: Yeah.
Demetrios: It's maybe not for an engineer the top of the pile of priorities.
Mihail Eric: Right.
Demetrios: Because there's also th- things you can do when it comes to just optimizing your skills and how they take in the context that can save you money too, or just a simple router-
Mihail Eric: Totally
Demetrios: so you don't have to be optimized to the gills-
Mihail Eric: Mm-hmm ...
Demetrios: on your, uh, cost saving-
Mihail Eric: Yeah ...
Demetrios: analysis or ways of doing it So yeah, that one's, uh, I can see that being a hard sell.
Mihail Eric: Mm-hmm, mm-hmm. Today, traces, I mean, you maybe heard this, or if you haven't heard this, this is gonna be a hot take, traces are like the new oil, right?
Mihail Eric: Like-
Demetrios: Oh.
Mihail Eric: Like, company [00:23:00] traces or data traces are, are where, like, a lot of the juice is. It's, it's like a, it's like the modern incarnation of, like, training data. Okay. It's like, you know, company's training data is the new oil, right? Um, and so there's so much value in traces because that effectively captures all of your customer interactions, all of your, like, app-level interactions.
Mihail Eric: And so when a company comes to me and says like, "Hey, put this little bit of JavaScript and through this, this additional layer, your, your, your, your customer's supposed to go through, like you're just capturing all of our traces as well." And I just, I've kind of got to a point where, uh, we have a tracing solution that, like for evaluation and observability, that of course captures our traces 'cause that's what we host and you- Mm-hmm
Mihail Eric: Braintrust in this case. Um, and we also use Jedra Labs for this. But I'm not, I, I don't feel like I'm at a point where I wanna keep just like willy-nilly having everybody-
Demetrios: Trying-
Mihail Eric: Yeah, have this d- A new one Yeah, try a new one. Like- Like, oh ... oh, now, now it's- Yeah ... routing to these guys, you know? And these guys, and these guys- And these guys, and these guys
Mihail Eric: and it's
Demetrios: incrementally better.
Mihail Eric: Right, increment. Is it, so I'm just, it's, it's a tough sell, [00:24:00] but, but this is, I feel like that's the equivalent of when people started doing like feature stores- Oh ... in like 2020. There were- There were
Demetrios: so many of them and you're like-
Mihail Eric: Yeah ...
Demetrios: is this really needed?
Mihail Eric: Totally. Totally.
Mihail Eric: And, and- What about-
Demetrios: Yeah ... MCP kind of analytics and understanding how people are using your MCP server?
Mihail Eric: Yeah, I think there's-
Demetrios: That feels a little different.
Mihail Eric: It's, it's, it's somewhat different. I think that one ends up being more, um... I mean, that's an interesting category. I think that one ends up being useful for coding agent optimization.
Mihail Eric: Mm-hmm. Um, maybe less so on like product optimization. I've yet to see as much of that being applied to like product MCP categories. Uh, but definitely for, I guess it's this emerging trend of like Agent X, like agent experience. Yeah. You know, you maybe heard of this. Yeah. It's like how do you optimize your MCP tools and your MCP, uh, servers so that an agent, it's easy for them to identify what they need to do, and they can do it as quickly as possible in as few hops as possible, essentially.
Mihail Eric: Yeah.
Demetrios: Well, and the, the pitch that I've heard is that you generally don't understand why the agent is coming to your MCP server- Right ... because you [00:25:00] don't know what the prompt is.
Mihail Eric: Right.
Demetrios: You just know that it's trying to call some tools.
Mihail Eric: Right, right.
Demetrios: And so-
Mihail Eric: Right ...
Demetrios: that doesn't necessarily need to be fixed with bringing on a new SaaS provider-
Mihail Eric: Right
Demetrios: that captures that information. You can build your MCP server so that-
Mihail Eric: Mm-hmm ...
Demetrios: you ask the agent, "Hey, yeah, cool, you've got this tool call happening, but-" Yeah "... mind if I get a little more information here?"
Mihail Eric: Right, right. That's right, that's right. Yeah, I, I think there's, it- an evolving category, right? I mean, we'll, we'll see where it all kinda shakes out.
Mihail Eric: Uh, I know that some of these providers are in the business of, you know, effectively they just run these, like, these tests on your agents with your servers and say, like, how, how easy, how cost-effective is an agent when it's making MCP tool calls? Yeah. How many tool calls does it need to make to, to get to a certain answer?
Mihail Eric: Um, how many errors is it hitting when it's making those tool calls? You know, things of that nature. So I think that's, it's a cool, it's a cool concept. It's a cool, uh, it's a cool category to, I think, be in. We'll see what, [00:26:00] what shakes out as, like, a standalone product. But
Demetrios: do you guys... You have an MCP server, right?
Mihail Eric: Yeah, yeah.
Demetrios: Yeah, for sure. And so talking about Monaco now-
Mihail Eric: Yeah ...
Demetrios: what are some things that you're doing in the AI field or just, like, the agentic field?
Mihail Eric: Yeah.
Demetrios: You have the MCP server. You also have agents inside of-
Mihail Eric: Yeah, yeah ...
Demetrios: Monaco?
Mihail Eric: That's right. Yeah. That's right, yeah. So, so we've, we've built our own, like, agent, agent builder, and we are...
Mihail Eric: Our, our pitch is that our agents are the best in class for go-to-market use cases, right? So we're, we're a sales company. We're, we're in the business of helping, um, we have, like, world-class go-to-market team members, and we're in the business of trying to distill the subject matter expertise of these world-class go-to-market folks into our platform to then make that available to our customers so that on day one, anyone has, you know, world-class go-to-market at their fingertips, essentially.
Mihail Eric: So, so that's, that's, that's what we're working toward, and that's, that's the broader pitch. Um, we've built a version of an agent platform that allows you to, to build go-to-market specific agents, whether that's pipeline reviews, you [00:27:00] know, on the fly, whether that's, uh, TAM building on the fly, like, uh, TAM segmentation on the fly, um, you know, demand generation on the fly.
Mihail Eric: So, like, what kind of campaigns should you be running, and how do you actually optimize, uh- Like paid
Demetrios: ad stuff?
Mihail Eric: Uh, we do- we don't do paid ads, more like outbound related things. Oh, okay. Um, so all the things that I'm sure you loved doing when you were a salesperson and you're-
Demetrios: I love spamming people ...
Mihail Eric: and you're-
Demetrios: Better
Mihail Eric: watch
Demetrios: out
Mihail Eric: today is, yeah, it's why you're so, so good at building communities, is you're so good at spamming. Um, but- That's
Demetrios: all it is. That's 90% of my job. People are like, "This guy again?" Or, "Oh-"
Mihail Eric: Yeah "...
Demetrios: I guess I might as well respond."
Mihail Eric: And then you, yeah, you get them on, like, what is it? What is it, like, how many t- how many touch points?
Mihail Eric: Like, seven to- 12 ... 12, 12 before you give up or something and, and-
Demetrios: 12 nos.
Mihail Eric: 12 nos. Yeah. And so, but, but you break people down by, like, five or six. Yeah. You know what I mean? Like- Yeah, exactly ... you don't need to get to 12.
Demetrios: They're like, "God, I am so done with this. All right."
Mihail Eric: Yeah, yeah.
Demetrios: So let that be a lesson to everybody that's listening.
Demetrios: If some sales guy is hitting you up, just say no And then set up a reoccurring agent to say no every [00:28:00] day and uh-
Mihail Eric: Yeah ...
Demetrios: you should be fine.
Mihail Eric: Yeah. So 12 nos and then they'll probably... If they, if they, if they're reading the same book they will- Yeah ... they will I
Demetrios: think that did come from, like, this guy Chet Holmes.
Demetrios: Uh-
Mihail Eric: Oh,
Demetrios: that sounds really familiar ... I read this back in, like, 2009 or something.
Mihail Eric: Which one is this? This
Demetrios: is like The Sales Machine.
Mihail Eric: Oh, yeah, yeah. I think I've act- I actually have read this book. Yeah, yeah, yeah, yeah,
Demetrios: yeah. And he used to go for, like, 15 nos.
Mihail Eric: Yeah, yeah.
Demetrios: And he would tell people, like, "Until you get 15, don't even think about stopping."
Mihail Eric: Yeah. Yeah, yeah, yeah. Okay. So, so, um, look, there's a To the, to the recipient, that might seem incredibly annoying. However-
Demetrios: Aggressive ...
Mihail Eric: however, okay, like- Would be
Demetrios: another word.
Mihail Eric: H- yeah, aggress- aggressive and, and, yeah, just leave me the hell alone. But, but having worked with our go-to-market folks, um, seeing how they think about it is, like, pretty enlightening, right?
Mihail Eric: It, it's never... It's, it's, um, it's not just, like, bump, bump. Yeah, no. Like, like, you
Demetrios: know- It's no value. Yeah. You're not adding anything. It's just, [00:29:00] "Hey, I saw this in the paper." Totally. "Thought you might want
Mihail Eric: it." Totally. Totally, totally. Exactly that. And that stuff is, like... That, that makes sense, right? I mean, that, that actually, like...
Mihail Eric: If, i- if, if you treat sales as not a what do I stand to gain from you as a salesperson, but rather what, what value can I offer- Yeah ... and be more- How can I help you? But how can I help you? I mean, you, you know, you, you know this. You live this, right? Um, I, I think that's the most effective kind of sales. And so a lot of, you know, our, our salespeople do that, and, and then we try to capture exactly those kinds of insights in terms of how we he- help people do outbound and how we help people, um, get in front of their target buyer- Mm-hmm
Mihail Eric: and, and the target companies they wanna reach out to. Um, anyway, it's, it's, it's been a really fun product to build. Uh, and-
Demetrios: And how does AI get incorporated in that? Yeah. I imagine you're listening for things, and then- Yeah ... you can craft a-
Mihail Eric: That's
Demetrios: right ... message.
Mihail Eric: That's right. That's exactly right. Exactly. So, so-
Demetrios: So that's one way, but what else?
Mihail Eric: Yeah, no. Yeah. So we have a, a number of different things at our disposal. So, um, everything is agentic, of course. So we're, we're getting signals from a number of different places in, in the platform. I mean, we're an all-in-one, so we, we [00:30:00] don't just do, like, a CRM or, you know, just the outbound or just the data enrichments.
Mihail Eric: Uh, we've, we've deliberately gone and built everything from the ground up.
Demetrios: Oh,
Mihail Eric: wow. So we are like the one-stop shop, and because of that, we get access to everything from, you know, emails that have been sent, uh, meeting recordings and transcripts, Slack messages. Oh, nice. So, you know, all these things that come in.
Mihail Eric: And then we can do... We can use all that to infer, um, org-level tone, org l- org-level understanding about business context. So, like, what is... If I am, I don't know, if I am Uber, who am I s- actually Uber is not a great example 'cause it's, like, more consumer-facing. But- Sure ... if I am, if I'm a- I don't know ... like, a...
Mihail Eric: If I'm a Base10- Yeah ... and I'm selling to, you know, uh, I'm- Base10, '
Demetrios: cause you just saw one of their
Mihail Eric: Yeah. Like, we're in SF, people. Yeah. In case that wasn't clear.
Demetrios: I'm- We just saw a bus drive by-
Mihail Eric: Yeah. ...
Demetrios: said Base10. We have Andreessen Horowitz literally right here on the
Mihail Eric: corner. There you go. I think we just passed Pylon.
Mihail Eric: Uh- Yeah ... yeah, so, so, you know, any of these B2B SaaS companies- Yeah ... uh, there's a certain understanding about their ICP they have. There's a certain understanding of, like, their target, [00:31:00] uh, demographic, the target, like, size of company, the person at those companies they need to be targeting. Pain points. The pain points.
Mihail Eric: Objections. Totally. 100%, all that stuff. And so, so we can glean all that through the collateral they upload to the platform or the stuff that we implicitly learn through sales interactions, through meeting transcripts, through emails. We also can glean that because we're seeing them-
Demetrios: Base10 is- Yeah ... coming out with some kind of a announcement or some PR- That's right.
Demetrios: Yeah ... or whatever it may be.
Mihail Eric: Yeah.
Demetrios: And they're focusing on this, it's like, "Oh, hey."
Mihail Eric: Yeah. That's right.
Demetrios: Should mention something to your contact.
Mihail Eric: 100%, yeah. A- and so, so, you know, and that's something, stuff we capture through platform level interactions, and then we, we have this concept of, like, Monaco memory, which is effectively like a, um, kind of related to, if you remember Andrej Karpathy, like many months ago, brought this concept of like an LM wiki.
Demetrios: Yeah.
Mihail Eric: Yeah. I see. So, so like the LM wiki, like we have- it's like a version of that, this, um, this persistent, uh, store of everything that's happened and everything that we know about both the organization as well as the [00:32:00] user at the organization. Okay. So your tone when you reach out to people, "Hey," when Demetrios reach does- reaches out, he's gonna be super annoying, and it's like- It's like, it's like, it's like Oh, I should have thought that one through.
Mihail Eric: Like, you know, but that's your tone, right? That's like, that's that's- Oh, gosh We can learn something like that. Give me one final dude. Ouch. All right. Exactly. I, I only need like two emails to glean that. Yeah. You know what I mean? I don't need I see what you mean. All right. So- No, no ... you
Demetrios: have an agent-
Mihail Eric: It's effective
Demetrios: to superpower my skill. There you go. Yeah, augment my ability.
Mihail Eric: No, no, no. Demetrios is very effective at what he does, of course.
Demetrios: Yeah. Of course. Dial it up to an 11.
Mihail Eric: Uh
Demetrios: But I didn't realize, I didn't think about how many different insights and actions you could take from having- Yeah ... all that data. Yeah. And the salespeople that I know have cobbled that stuff together-
Mihail Eric: That's right, that's right
Demetrios: in a way that they're like, they're recording with one software- Mm-hmm, mm-hmm ... and then they have some kind of a automated pipeline where it will be-
Mihail Eric: Yeah ...
Demetrios: put into their [00:33:00] ChatGPT or Claude. That's right. And then from that, they get insights, and it gets thrown into HubSpot. That's right. And then-
Mihail Eric: That's exactly
Demetrios: right
Demetrios: they can write a report at the end of the month with all these calls that they've had- Mm-hmm ... or with all the, at the end of the week, whatever it may be, when they have the pipeline review. That's so correct.
Mihail Eric: Exactly
Demetrios: correct. So now you're saying, "Oh, cool. We've got your pipeline review already ready- Yep
Demetrios: because we have all of this stuff and we've set that up." How much work is it- Yeah ... on my side to put together all of these tasks or, they're not even tasks, they're like- Different lanes-
Mihail Eric: Yeah ...
Demetrios: that I want.
Mihail Eric: N- very, very little, honestly. Uh, and part of it is because, I mean, the platform is robust on its own, but the other thing that we've leaned into a lot from the get-go, which I think differentiates us from other products in, in the category, is we are...
Mihail Eric: We, we have... We're actually, like, a little bit more high touch on the human side than maybe would traditionally... So, you know, the, the PLG self-serve, like, you just log in, and you just start cranking. You start adding rows to your [00:34:00] tables and whatnot. It's, it's... That's not the experience here. The experience for us is every single person has a designated what we call an FDAE, so a forward deployed account executive.
Mihail Eric: Uh. So it's effectively like your sales consultant, um, that sort of guides you through the process for... 'Cause for a lot of, a lot of our early customers, you know, we, we have done really, really well with, like, early-stage founders that are all, like, very technically minded, very, very product minded, um, but maybe don't have as much experience on the go-to-market side, right?
Mihail Eric: And they're, they're learning these... They're kind of picking these skills up as they go, and so we're like, "Hey, we've... We have decades of experience doing this." So day one, you come on board to Monaco, and you have a built-out TAM for you, like three to 20,000 depending on your TAM size, companies that have been scored and ranked ready for you to go.
Mihail Eric: Whoa. Here are the buyers of those c- you know, and here's how you can get to those buyers based on, like, LinkedIn connections and all these things. All you have to do is- Handwritten
Demetrios: notes.
Mihail Eric: H- yeah, handwritten, all these things, yeah, and, and, and all you do is just hit go. And then, and then sequences are getting sent, getting followed up with.
Mihail Eric: You know, you're getting insights brought to your, to the platform as, as, [00:35:00] as things happen.
Demetrios: Don't sleep on the handwritten notes. Those are actually quite useful.
Mihail Eric: I... No, we, we do that. We... Like, I- Do you? No, yeah, of course. Like, yeah, when... I mean, we, uh, we had handwritten notes. We had this, uh, campaign where we were giving, you know, Monaco, so we, we do stuff in...
Mihail Eric: We have, like, a bit of a brand around gambling, and, like, the finer things, and F1. And so we, we had, um... We were... One part of our campaigns was- Tax
Demetrios: evasion.
Mihail Eric: No, no, no, no. Never. Whenever
Demetrios: I think Monaco, I'm just like, ooh.
Mihail Eric: No
Demetrios: taxes.
Mihail Eric: No, no, no, no, none of that. None of that. Um, none of that, none of that publicly. None of that on the record.
Demetrios: Yes.
Mihail Eric: Cut
Demetrios: that
Mihail Eric: out. Boss is gonna get pissed. No, no, no. None of that.
Demetrios: Um- So anyway, you got the brand around-
Mihail Eric: You got the brand around the gambling. And so we actually had, um... We had a... We do a few things that are kind of fun. Like, we, we gifted people, like, poker sets, that- The poker
Demetrios: night.
Mihail Eric: I know what you mean, yeah.
Mihail Eric: Po- poker night. We do the poker night. Um, that's happening, I guess, next, next week. Uh, I don't know when this is coming out, but this is October 29th. So we're doing one that's... I, I think we... Our claim to fame on, on that front is having the most, um Like the [00:36:00] biggest prize, biggest- Nice ... cash pool on, on poker.
Mihail Eric: $100,000 for the final table, so it's, uh- Wow. You know, let me tell you, somebody walking in, I've seen this twice already now, somebody walking in to just a random, like, Thursday night poker game and like, "Eh, I'm gonna eat some food, play some poker," and they're walking with $30,000. You're like- What? You're like, "What?"
Demetrios: That is VC money well
Mihail Eric: spent. That's wonderful. There you go. There you go. You know, it's, it's, it's about brand activations- Yes ... or brand awareness, right?
Demetrios: Um- Oh, that... And I think you guys do that really well. I saw the stuff where you're flying planes over SF- It's
Mihail Eric: fun,
Demetrios: yeah ... with the Monaco symbol.
Mihail Eric: Yeah, we-
Demetrios: I think just earlier today I saw somebody with a Monaco tote bag, and so you're doing well on that side of things.
Demetrios: Sure. When we talk about the tech-
Mihail Eric: Yeah ...
Demetrios: what are you actually using?
Mihail Eric: Mm-hmm.
Demetrios: Break down the stack, where are the harnesses, all of that stuff-
Mihail Eric: Yeah, yeah ... to
Demetrios: create.
Mihail Eric: Totally. Um- So a lot of... We, we've homegrown our own, our own harness, right? So a lot of... So there are, there are pieces of the harness that [00:37:00] we have used vendors, so for like Vercel sandboxes, right?
Mihail Eric: So we use them. We don't believe there's value in us building micro VMs- Mm ... you know, or like maintaining our own version of that. That, that would... That's what I would call undifferentiated heavy lifting. Mm. But, but at its core, we do have our own harness. We think that's like a competitive advantage, is to like re- lean very heavily into the primitives around a go-to-market, go-to-market specific ha- uh, harness, including the skills- And that's-
Demetrios: Yeah
Demetrios: that's the harness for the agents that- For the a- ... the end user is using. That's- Not your internal harness for coding and stuff.
Mihail Eric: No, no, yeah. No, not the... Yeah, no, not that. Yeah, exactly right. So, so good, good distinction. So we have our, we have our own internal coding system we've also built, but the, this one, the ex- the example that I'm giving is specifically around go-to-market.
Mihail Eric: The ones that produce the go-to-market agents- Yeah ... our customers- Got it ... uh, get to benefit from. And so, you know, at its core, we've, we've built effectively this like system that goes from natural language into this, uh, like intermediate representation. Like, you have, you can think of like a JSON of sorts, um, maybe, you know, JSON, like the YAML, whatever you want, that represents like the workflow of the agent.
Mihail Eric: Mm-hmm. And then that gets compiled down into like a [00:38:00] LangGraph, essentially. Um- Mm. Okay ... a LangGraph graph, uh, that you can then run and, and, you know, can loop and, and do, uh, the, the kinda self-healing that, that an agent will do. Mm. Um, you know, any- we can... The underlying... So effectively, it's like a compiler.
Mihail Eric: Like, we go from natural language to compile to an underlying engine that is the, the workflow builder, but we could swap it out into any other workflow builder if we wanted. Um, and then the, the, the secret sauce a lot of the times is like what have we built around the memory? What have we built around the tools that we've exposed?
Mihail Eric: Compaction. Compaction. Yeah. Like, how do we compact more intelligently? How do we... You know, all these things that are specific, you know, the, the data that we access, right? So of course, we have very particular data characteristics because we're dealing with, in some cases, like thousands of meeting transcripts- Yeah
Mihail Eric: and hundreds of thousands of emails for some of these orgs that get onboarded.
Demetrios: Yeah, that, that was what I was thinking about, is how you can avoid just bloat.
Mihail Eric: Yeah, yeah. It's, uh, it's... There, there, there's a sorta tips and tricks around this, um, but at its core, it's like you just have to be a little bit more intelligent around like the MCP tool, MCP server tool [00:39:00] design.
Mihail Eric: Like, you can't do, you can't like... I think the, the, the V1 vanilla way that a lot of people treat MCP is like- Here's an API, here's a one-to-one translation of the MCP tool for that API It's like the
Demetrios: worst way to do it.
Mihail Eric: Yeah, it's not. So you have to think about it in more terms like the semantics of what is it the agent needs to do?
Mihail Eric: Hey, I need to query meetings, and you can do query, query meetings either through, uh, like keyword searches or through like a more semantic relevant search based thing. And so you have to make, like, a single entry point for, for that tool call, for example, and then allow both modes as, as, as a way to, you know, query the actual data.
Mihail Eric: So, um,
Demetrios: the- Yeah, I've heard it said as doing the intent.
Mihail Eric: Yeah
Demetrios: Like really designing your server around intent-
Mihail Eric: That's right ...
Demetrios: versus tools.
Mihail Eric: That's exactly right. That's exactly right. Yeah, I think tools, yeah, the, the latter is, like, V1 APIs, right? This is like 2000s level API tool definitions where you have like get ID.
Mihail Eric: Well, I mean, you know, like, like this is... When we... And, and not in a, not to be derogatory, but, but in like a, uh, that's how we started when we were building web apps of the early 2000s- Yeah ... is like, [00:40:00] hey, get ID, you know, set ID, like all these- Yeah ... get and post style, uh, HTP methods that we defined. And now you're right, like agents actually can think at a higher level or like can operate at a higher level.
Mihail Eric: I don't wanna anthropomorphize here, but
Demetrios: like- But do you think that that is your, where your IP is happening, like for you building this product-
Mihail Eric: Yeah ...
Demetrios: which potentially can be headless completely in the future?
Mihail Eric: Mm-hmm. Mm-hmm.
Demetrios: And people interact with Monaco solely through the MCP.
Mihail Eric: Yeah, there you go.
Demetrios: Then your special sauce is how you're compacting, how your memory is- Yeah
Demetrios: how your search-
Mihail Eric: Yeah ...
Demetrios: functions work.
Mihail Eric: Yeah, I think that's right. I think it's a combination of the fact that because we're the all-in-one, we are both the, the system of record and then what I, I would call like the system of intelligence. So the system of record is we still are the place where you wanna store like CRM level data, right?
Mihail Eric: So the, the evolution of your deal pipeline, the accounts to reach out to, all these things. So, so that still is baked into the platform, and then you need abstraction layers, you know, tools in a lot of cases to interact with that data, reach from that data, write to that data. And [00:41:00] so you can do that either through like an external MCP or you can use that all internally and use our agents, which are baked into the platform.
Mihail Eric: The, the benefit of our agents are they're gonna be highly optimized to the workflows that we care about. They're going to, um, the defaults are gonna be... They're gonna capture a lot of the, the best, best in class practices from our own go-to-market teams. Like, it's remarkably difficult to build like a really good TAM, right?
Mihail Eric: If somebody comes in like, "Hey, I'm selling to a marketplace," but like they're like, think of them like Uber, but you know, like they're not too big and so, but it's in like, you know, it's in like- East
Demetrios: Asia ...
Mihail Eric: it's in East Asia, and it's for shipping companies. And you're like, "Okay, how many companies of those are there that you can target?"
Mihail Eric: Yeah, go do a LinkedIn search. Yeah. Yeah, exactly. Right. Exactly. So like you, you know, and so we've, we've, um, you know, the IP is like the data layer, the context layer of course, and then, and the fact that like- All that is, is resting on the bet that it is like the system of record, right?
Demetrios: And how much are you enabling the other software engineers at [00:42:00] your company-
Mihail Eric: Yeah
Demetrios: as your job versus building these externally facing agents?
Mihail Eric: It's both. It's both. Yeah. It's, it's, it's definitely both, right? So, so we have to do both. I think the, the... where we've really thrived as a company is domain experts working hand-in-hand with like subject, uh, with like software engineers, right?
Mihail Eric: That can like codify the workflows of the, of the domain experts. And so in practice, this, the software engineers have to be empowered to build very quickly, to ship really quickly- Yeah ... to, to have the primitives to, to build on top of the harness. And then the subject matter experts need to be empowered to, you know, know how to express their intent as well and the things they care about.
Mihail Eric: And then, 'cause the, the, the failure mode is like the translation, right? Is like how if technical person talking to a non-technical person, it, it sometimes they can like kinda miss each other and not really-
Demetrios: Yeah ...
Mihail Eric: not speaking the same language. Um.
Demetrios: Well, dude, we're arriving at our destination. This is fun.
Demetrios: Now I gotta kick you out.
Mihail Eric: This is fun. Oh gosh.
Demetrios: Oh gosh. As much as I wanna [00:43:00] keep talking. I'll see you on Thursday though.
Mihail Eric: Yeah, yeah, yeah.
Demetrios: You are a person of interest in the magazine we created.
Mihail Eric: Uh-oh. Uh-oh. That is, that sounds like that sounds like I'm, I'm being- You're about to be getting a lot of sales emails.
Mihail Eric: Well that or, or it sounds like I'm being tracked by the FBI. Yeah. Like, he's on a wanted list somewhere. Yeah. I, I'm cool, I'm cool. I'm cool. Exactly. Way, Waymo people, don't worry about me.
Demetrios: Yeah. Hey Waymo, don't-
Mihail Eric: D- don't- No
Demetrios: need to record this ...
Mihail Eric: don't need to record this. Lock the doors. Don't let him out.
Demetrios: Yeah.
Demetrios: But it's been cool, man.
Mihail Eric: It's fun, man. It's- I
Demetrios: appreciate you doing this.
Mihail Eric: This is a cool concept. I think this is, this is a... I, I'm a big fan of this. Hey man, it's so fun to see you, man. Let's go. Yeah, I'll see you on- Thursday ... Thursday. Yeah.
+ Read More
Watch More
Coding with AI // Chip Huyen
Posted Nov 21, 2025 | Views 862
# Agents in Production
# Prosus AI
# AI-assisted Coding
Building Cody, an Open Source AI Coding Assistant
Posted Aug 28, 2023 | Views 908
# Open Source AI
# Cody
# Sourcegraph