Episode Summary
Executive Summary: Alex Lieberman and Armand Hezarcani, co-founders of 10X, discuss their AI-first engineering consultancy that compensates engineers based on output rather than hours. They share how Armand's experience downsizing his engineering team by 90% and rebuilding with AI led to the insight that AI enables 10x productivity, but traditional hourly billing disincentivizes efficiency. 10X hires top engineers, pays them for story points, and has seen rapid prototyping results, like building a mobile app that hit #20 globally in a month. They are currently human-capital constrained and use a rigorous take-home interview to filter candidates.
Main Topics: Origin of 10X: From AI-Driven Efficiency to Output-Based Compensation: Armand downsized his engineering team by 90% and rebuilt with AI, achieving 10x output. This led to the insight that hourly billing disincentivizes AI-driven efficiency, so 10X compensates engineers based on output (story points) to align incentives. Output-Based Compensation Model and Its Challenges: 10X pays engineers per story point, not per hour. They mitigate gaming by hiring long-term selfish engineers and using technical strategists who are incentivized on retention to sign off on engineering plans. They have not yet faced client disputes over story points. Rapid Prototyping and Client Success Stories: 10X built a prototype for a retail camera system in two weeks (previously quarters), a mobile trivia app that hit #20 on the App Store in a month, and a fitness coach app prototype in four hours to win a client. Technology Stack and AI Agent Selection: Default stack is TypeScript (frontend and backend) with React/Express. They do not have a favorite coding agent; they switch based on daily performance (e.g., Claude Code vs. Codex). They emphasize 'feel' over formal evals. Hiring Process and Constraints: 10X is human-capital constrained. They use an unreasonably difficult take-home interview that 50% of candidates fail to complete, but it shortens the overall process to as fast as one week. They ask questions about building an AI engineer to gauge depth. Debate on MCP (Model Context Protocol): Armand is skeptical of MCP, calling it 'a three-letter word for API' and criticizing hype-driven fundraising. Alex is more neutral, noting that MCP has useful spec elements beyond APIs. They plan a pro/con debate at the conference.
Key Arguments: Hourly billing perversely incentivizes engineers to work slower, especially when AI boosts productivity; output-based compensation aligns incentives with efficiency. Hiring 'long-term selfish' engineers who understand that maintaining client relationships yields long-term rewards prevents gaming of story points. Technical strategists incentivized on retention and account growth serve as a quality gate, ensuring engineering plans are robust before clients see them. Rapid prototyping with AI can win clients who initially decline, as demonstrated by building a working app in four hours. The limiting factor for scaling 10X is finding enough top-tier engineers, not technology or capital. Context engineering (getting the right context into LLMs) is the key bottleneck to building a fully autonomous AI engineer, not just model intelligence.
Data Points: Engineering team downsizing: 90% - Armand downsized his engineering team by 90% at Parthian and rebuilt with AI. Output increase after AI-first rebuild: 10x - Output of production-ready software increased 10x after the shift. Engineer compensation: Over $1 million cash per year - 10X expects multiple engineers to earn over $1 million cash next year based on story point compensation. Prototype build time (retail camera system): 2 weeks - A prototype that previously would have taken several quarters was built in two weeks. App Store ranking: #20 globally - A mobile trivia app built by 10X hit #20 on the App Store globally in one month. Prototype build time (fitness coach app): 4 hours - An engineer built a working version of a fitness coach app in four hours to win a client. Take-home interview completion rate: 50% - 50% of candidates do not even respond to 10X's unreasonably difficult take-home interview. Interview process duration: 1 week - The fastest interview process can be completed in one week if the candidate passes the take-home.
Pivotal Quotes: "We will probably have more than one engineer make a million dollars cash next year based on this model." — Alex Lieberman: Describing the financial upside of 10X's output-based compensation model for top engineers. "I just think that MCP is a three-letter word for API." — Armand Hezarcani: Expressing skepticism about the hype around Model Context Protocol, which he sees as rebranding existing concepts. "The thing that keeps us up at night is how can we hire enough good engineers fast enough?" — Alex Lieberman: Identifying human capital as the primary constraint on scaling 10X.
Implications: Output-based compensation could disrupt traditional engineering billing models, especially as AI boosts productivity. Companies may need to rethink hiring and incentive structures to attract top AI-native talent. Rapid prototyping with AI will become a competitive advantage in sales and client acquisition.
From the Episode
Work well with story points. I think one thing I'm trying to get my guests to do a better job of is just brag. Could you brag a bit? Yeah, just like some really interesting project that you accomplished, just opens people's minds. Let's get specific without maybe naming the exact client, unless you can. And then also, what's the highest hourly rate that one of the engineers has made since you're technically young Pat? Yeah. So I'll answer the last one or the second one first. We will probably have more. Than one engineer make a million dollars cash next year based on this model. And that is just with story point compensation. It's very likely that we will have more than a handful of folks make more than a million dollars next year. The answer to the first question, like, for example, one project we built, so we work with this company that's a, they, they build, they work, they partner with retailers to basically make cameras in the business more valuable. And the way that they do that is they deploy what was historically like a Gen 4.
Of this kind of conference would be sort of stewing around those ideas. Like MCP, the last time we were here in New York, MCP was kind of just taking off, and we did the workshop, and that really blew up MCP. And I think, like, that is something that you will see a little bit of. Like, just like, think about the way Armand Grimms grins because he has very strong feelings about MCP. Very strong feelings. Prolo, Anti. We're hosting a debate. I just think that MCP is a three-letter word for API. And like, Alex Oliver. Every time he hears someone say the word the letters MCP in that order, he tells them that I hate MCP and starts a war, a religious debate. Well, I will say though, do you think a few of our engineers have warmed you up to it more with specific use cases, Armand? Yeah, I mean, like, are MCPs useful? Like, of course, I use all the MCPs with Claude Code. I just think that there's like what bothers me is when people create a new name for something and then use that to raise some inordinate amount of.
Like, what is your limiting factor in 10X becoming a bigger business than either of you have run before? Today, it's human-bound 100%. You're recruiting. Yeah. We are. The thing that keeps us up at night is how can we hire enough good engineers fast enough? And then the second thing that keeps us up is how do we match those, the great people within the business with the right process such that delivery doesn't suffer. As we scale. And I think more and more as we build this business, like technology is going to be an enabler of the work we do. And I think long term, if we're to talk about the long term of the business, there are ambitions of this business beyond just acting as a transformation and engineering partner for companies. There are ambitions to build our own technology. But today, and probably for the foreseeable future, we are human capital constrained. How do you interview? You don't have to give the exact interview.
About Latent Space: The AI Engineer Podcast
The podcast by and for AI Engineers! In 2025, over 10 million readers and listeners came to Latent Space to hear about news, papers and interviews in Software 3.0. We cover Foundation Models changing every domain in Code Generation, Multimodality, AI Agents, GPU Infra and more, directly from the founders, builders, and thinkers involved in pushing the cutting edge. Striving to give you both the definitive take on the Current Thing down to the first introduction to the tech you'll be using in the next 3 months! We break news and exclusive interviews from OpenAI, Anthropic, Gemini, Meta (Soumith Chintala), Sierra (Bret Taylor), tiny (George Hotz), Databricks/MosaicML (Jon Frankle), Modular (Chris Lattner), Answer.ai (Jeremy Howard), et al. Full show notes always on https://latent.space
View all episodes from Latent Space: The AI Engineer Podcast