Arman Hezarkhani challenges conventional engineering compensation, arguing that paying AI engineers like salespeople—based on delivered value—is the only way to maximize their full potential and accelerate AI adoption.
The AI Productivity Chasm & Incentive Failure
- Hezarkhani, a Carnegie Mellon alumnus and former Google engineer, observes a stark divide in engineer productivity. He contrasts an engineer manually typing code with his own workflow, leveraging 45 AI agents for tasks like lunch ordering, code generation, and research. This disparity highlights a fundamental incentive problem.
- Traditional compensation models (hourly, project-based, salary+bonus, equity) fail to motivate engineers to adopt cutting-edge AI tools.
- Founders, driven by direct ownership, actively seek efficiency gains through AI.
- Employees, however, often lack direct incentives to innovate or integrate AI, leading to a significant productivity gap.
- “My belief is that this is an incentive issue.” – Arman Hezarkhani
The Evolution of Engineer Compensation Models
- Hezarkhani traces the historical shortcomings of various compensation structures, arguing each model eventually breaks down in the face of evolving work dynamics.
- Hourly Pay: Incentivizes slow work; offers no upside for efficiency.
- Project-Based Pay: Engineers inflate estimates to mitigate personal risk, leading to higher costs for buyers.
- Salary + Bonus: Fosters minimum effort, with employees performing only required tasks ("punch in at nine, leave at five").
- Equity: While effective for high-risk founders (e.g., Larry Page at Google), it holds less appeal for risk-averse engineers who increasingly prioritize cash over speculative upside.
- “My contention is that this model needs to be reinvented in the age of AI.” – Arman Hezarkhani
TenX's Output-Driven Compensation Framework
- TenX, Hezarkhani's company, implements a novel compensation model directly linking engineer pay to delivered value, mirroring sales commission structures.
- Clients pay TenX based on "story points" (a unit of work estimation in agile development) delivered for custom AI builds.
- Engineers receive a flat base salary plus quarterly bonuses tied directly to the number of story points they complete.
- This model attracts top talent, including ex-founders, NASA rocket scientists, and world-class machine learning and AI researchers.
- The process involves strategists distilling product requirements, engineers crafting architecture design documents, and then implementing tickets graded by story points.
- “When that ticket is accepted, the engineer gets paid a fee per story point that they complete.” – Arman Hezarkhani
Real-World AI Implementation & Impact
- Hezarkhani showcases TenX's successful AI projects, demonstrating the model's efficacy in delivering rapid, high-quality solutions.
- Billboard Ad Moderation: TenX developed an AI model in two weeks for a billboard company, achieving 96% accuracy compared to human moderators. This reduced operational costs and increased revenue by minimizing billboard downtime.
- Retailer Edge AI: For a global retailer, TenX built five parallel AI models for low-power edge devices. These models perform heat mapping, queue detection, and theft detection, significantly expanding the devices' utility beyond initial single-model capabilities.
- Each project adheres to a structured workflow: product requirement definition, architecture design, and story point-based implementation.
- “We did it in two weeks and we got to 96% accuracy when compared to the human moderator.” – Arman Hezarkhani
Mitigating Risks in Output-Based Pay
- Hezarkhani addresses the inherent risks of an output-based compensation model, outlining TenX's countermeasures.
- Inflated Story Points: Strategists scope tickets, and internal reviews, alongside client approval, prevent overestimation.
- Quality Degradation: Multi-round internal QA and client approval for every delivered ticket ensure high code quality and functionality.
- Internal Competition: Strategists are compensated based on customer happiness (NR), providing a counterbalance to engineer incentives and fostering collaboration.
- Rigorous hiring is paramount; AI amplifies inherent engineer qualities, making strong engineers exceptional and poor ones less effective.
- “AI makes people look like one of those crazy mirrors where any one of your attributes, it makes it 10 times larger.” – Arman Hezarkhani
Investor & Researcher Alpha
- Capital Movement: Investment will increasingly favor firms and internal structures that directly incentivize engineer output and AI tool adoption, moving away from traditional salary/equity models that fail to capture AI-driven productivity gains.
- New Bottleneck: The primary bottleneck for AI integration is shifting from talent scarcity or compute power to the incentive structures within engineering teams. Companies must address this to maximize existing talent's AI potential.
- Research Direction: Future research into AI-assisted engineering workflows should prioritize incentive alignment and performance measurement, not just tool efficacy. Models that quantify and reward AI-driven productivity will gain significant traction.
Strategic Conclusion
- Traditional compensation models hinder AI adoption and engineer potential. The industry must transition to output-based incentives, mirroring sales structures, to accelerate AI development and maintain competitive advantage.
- The next step involves redefining engineering value through measurable, AI-augmented delivery.
