“Engineers are becoming sorcerers” | The future of software development with OpenAI’s Sherwin Wu - Lenny's Podcast: Product | Career | Growth Recap

Podcast: Lenny's Podcast: Product | Career | Growth

Published: 2026-02-12

Duration: 1 hr 20 min

Summary

In this episode, Sherwin Wu discusses the transformative impact of AI on software engineering, highlighting how tools like Codex are reshaping the role of engineers into tech leads and wizards of code. He emphasizes the need to adapt to the evolving landscape of AI and its second-order effects on startups.

What Happened

Sherwin Wu, the head of engineering for OpenAI's API and developer platform, shares insights into how AI tools like Codex are revolutionizing software development. He notes that a staggering 95% of engineers at OpenAI use Codex daily, with 100% of production code being reviewed by it. This has led to a shift in the engineering role, where engineers are transitioning from writing code to managing AI-driven processes, akin to 'wizards casting spells'. The integration of AI tools has resulted in engineers opening 70% more pull requests, showcasing their growing trust and proficiency with these technologies.

Wu points out a key challenge faced by teams relying entirely on AI-generated code, as they encounter issues without the option to manually intervene. He stresses the importance of anticipating the evolution of AI models, advising developers to build for future capabilities rather than current limitations. Wu quotes Kevin Whale, OpenAI's VP of Science, who states, 'this is the worst the models will ever be,' underlining the rapid advancement in AI technology and its implications for software engineering in the coming years.

Key Insights

Key Questions Answered

How is Codex changing the role of engineers?

Codex is fundamentally altering the traditional role of engineers by allowing them to focus on management and oversight rather than manual coding. Sherwin Wu indicates that engineers are becoming tech leads who manage fleets of AI agents, transforming from code writers to orchestrators of AI-driven processes. This transition marks a significant shift in responsibilities, as the act of writing code is now largely handled by AI.

What challenges do teams face with AI-generated code?

Wu mentions that teams using a 100% Codex-written code base encounter unique challenges, particularly when issues arise. Unlike traditional coding environments where engineers can troubleshoot and manually intervene, these teams lack that escape hatch, which can lead to significant stress when the AI tools do not perform as expected. This highlights the need for engineers to adapt to new workflows that incorporate AI-generated solutions.

What is the significance of the 70% increase in pull requests?

The 70% increase in pull requests among engineers using Codex reflects a major shift in productivity and engagement with the coding process. Wu points out that as engineers become more adept at using Codex, they open more pull requests, indicating a growing trust in the AI tool. This trend emphasizes how integrating AI tools can enhance collaboration and output within engineering teams.

How should developers prepare for the future of AI in software development?

Wu advises developers to focus on building for the future capabilities of AI models rather than their current state. He emphasizes the importance of understanding that the models are continuously evolving and improving. As Kevin Whale stated, 'this is the worst the models will ever be,' suggesting that developers should anticipate rapid advancements and adapt their strategies accordingly.

What does Sherwin Wu imply about the future engineering landscape?

Wu implies that the engineering landscape is on the brink of a significant transformation, where the role of engineers will continue to evolve rapidly. He describes the current moment as an exciting time, filled with potential for engineers to reimagine their work. As AI tools become more integrated into the development process, engineers will need to adapt their skills and approaches to keep pace with ongoing changes in technology.