AI's 2025 coding inflection point reshapes software development and workforce dynamics

Apr 2, 2026 · Lenny's Podcast
🎧 PodShort 100 min squeezed to 2 AI SprinklerAS AI / ML
Episode artwork
Simon Willison
Independent AI Expert/Engineer at N/A
Lenny's Podcast
100 min squeezed to 2
Full episode from Lenny's Podcast
Quotable Moments

Today, probably 95% of the code that I produce, I didn't type it myself.

Simon WillisonIndependent AI Expert/Engineer
00:23

What gets interesting is when if you ask them for 20 more [ideas], and now by the sort of end of that list, you're beginning to get things which are not good ideas, but they point you in interesting directions.

Simon WillisonIndependent AI Expert/Engineer
02:59

The problem is the people in the middle [of their career]. Like if you're mid-career, if you haven't made it to sort of super senior engineer yet, but you're not sort of new either, that's the that's the group which Thoughtworks resolved were probably in the most trouble right now.

Simon WillisonIndependent AI Expert/Engineer
03:14
Key Insights
  • AI's ability to generate code became a major focus for leading AI companies in 2025, driven by products like Anthropic's Claude Code, which saw rapid adoption and high user willingness to pay.
  • A significant 'inflection point' occurred in November, when GPT 5.1 and Claude Opus 4.5 were released, models that were incrementally better but crossed a critical threshold, enabling almost flawless code generation.
  • Vibe coding, defined as building applications without looking at or understanding the code, empowers non-programmers to create functional apps, democratizing software development.
  • Using AI-generated code responsibly is crucial, especially when it's for others, as undetected bugs could harm users. This necessitates expert judgment to ensure ethical deployment.
  • The 'Dark Factory' pattern describes a futuristic approach to software development where code is built and maintained without direct human review, similar to fully automated factories operating in darkness.
  • AI tools significantly amplify the skills of experienced engineers and accelerate onboarding for new engineers, but mid-career engineers face challenges as their current roles may be most impacted by automation.
  • AI excels at generating initial, obvious ideas in brainstorming, allowing humans to focus on refining, combining, and exploring more unconventional and interesting directions.
  • The 'Lethal Trifecta' is a critical security vulnerability where an AI agent has access to private information, is exposed to malicious instructions, and possesses a mechanism to exfiltrate that data, posing severe risks.
Metrics Mentioned
  • 1000 interns (Cloudflare and Shopify hired 1000 interns over the course of 2025, with AI reducing onboarding time from a month to a week.)
  • $10,000 a day on tokens (The cost of simulating end-users for rigorous security software testing using AI agents.)
  • 100 potential vulnerabilities (Anthropic's specialized security AI models discovered 100 potential vulnerabilities in Firefox.)
  • 95% of the code (Simon Willison states that he personally produces 95% of his code without typing it himself, leveraging AI.)
  • 10,000 lines of code in a day (Simon Willison claims he can churn out 10,000 lines of code in a day using AI.)
  • 250 Kakapo parrots (The total number of rare Kakapo parrots left in the world, with a fantastic breeding season in 2026.)

RevBots.ai View:

  • AI Sprinkler teams are bolting on code generation tools without rethinking workflows, creating technical debt.
  • The 'Dark Factory' concept mirrors ARM's vision of full automation, but requires responsible implementation.
  • Revenue teams should track how AI-driven product development cycles impact competitive positioning.
🎧Full Episode:Lenny's Podcast →