Naval

waste tokens save time

Naval·May 30, 2026

OVERVIEW

This episode discusses the profound impact of artificial intelligence on software engineering, shifting the role of engineers from direct coding to designing "software factories" and architecting systems. It explores how AI models are enabling unprecedented productivity gains, changing how software is built, and redefining the skills most valuable in the tech industry.

KEY TOPICS

  • The concept of "software factories" and multiplicative output from AI.
  • The evolution of engineer productivity from 10x to 100x or 1000x with AI.
  • The utility and limitations of AI models (e.g., Claude, ChatGPT) in development.
  • The importance of human judgment and architectural decisions in AI-assisted development.
  • The practice of "wasting tokens to save time" when interacting with AI models.
  • The potential obsolescence of traditional software engineering and the rise of new specializations like model training and building AI infrastructure.
  • The concept of "building blocks" and reusable components for AI agents.
  • Personal experiences of building complex software without writing traditional code.

MAIN TAKEAWAYS

  • AI models are transforming software engineering by enabling engineers to achieve 100x or 1000x productivity, shifting the focus from writing code to designing systems and providing strategic guidance.
  • The models, while increasingly capable of complex reasoning and suggesting tradeoffs, still benefit significantly from human judgment, domain expertise, and effective prompting, though this reliance may diminish as they become smarter.
  • The act of writing code by hand is becoming less necessary for building functional software, with AI tools handling the implementation details, allowing humans to focus on higher-level architectural decisions and problem-solving.
  • Waste tokens to save time: It is often more efficient to "brute force" AI models with repeated prompts and tokens than to meticulously craft perfect prompts, given the models' low cost relative to human time and the continuous improvement of the models themselves.
  • The future of software engineering may lie in "training models" and creating reusable "building blocks" (libraries and dependencies for AI) rather than traditional coding, potentially making "pure software engineering" as we know it an obsolete skill for many tasks.

NOTABLE QUOTES

"The way that I'm judging you as an engineer is like, are you producing the factory that will produce multiplicative outputs B through Z, right?"
"I just assume the models are just going to get better faster than I would figure out how to use it. It would figure out how to use me faster than I would figure out how to use it."
"There's something really cool is that you understand how the pieces click together. Like, I feel like anyone that understands what an API is, and how data flows, inputs and outputs, performance... that's always been infinitely more useful than writing code."
"The thing that really changed... with the agents what happens is you just don't get stuck anymore."

Summarized with DriftNote — AI-powered podcast summaries

Try it free