Coding is solved. Now what?

Three Lenny's Podcast episodes dropped recently that, taken together, paint a pretty clear picture of where software is heading. I keep coming back to them.

Boris Cherny: "What happens after coding is solved"

Boris is the head of Claude Code at Anthropic. His claim: coding is now "solved" for most use cases. He hasn't written a single line of code by hand since November 2024, and 100% of his work is authored by Claude Code. Meanwhile, he's still one of the most productive engineers at Anthropic. The team has seen a 200% increase in engineer productivity since adopting the tool.

The part that stuck with me: he said the title "software engineer" is going to start going away by the end of the year, replaced by "builder." Product managers, designers, data scientists will all see similar transformations as agentic AI expands beyond coding. Any job where you use computer tools is next.

Jenny Wen: "The design process is dead"

Jenny is the head of design at Anthropic (Claude). Her argument: the traditional design workflow of research, mockups, and iteration is functionally obsolete. When an engineer can spin up multiple AI agents and ship a working version before a designer finishes exploring options, the old discover-expand-narrow framework doesn't make sense anymore.

The title sounds scary on purpose, but the real point is that we have to change the way we work to go faster, and we can all do a bit more. The proportion of time on mockups has shifted from 60-70% down to 30-40%, with more time spent jamming and pairing directly with engineers in code. Design isn't dead. The process is just different.

"Engineers are becoming sorcerers"

This framing from another Lenny's episode captures something I've been feeling. AI specifically flattens the curve to learn technologies. A lot of the way team structures used to be created was based on how many different pieces of software you could learn. Now that you don't need to learn as much to use those tools, some of those roles are getting consolidated into builders with different focuses, all embedded on a team, all heavily supporting developers, and many pushing code themselves.

I believe this is a good thing. Not because it eliminates jobs, but because it removes artificial barriers. The skill ceiling isn't "can you learn React and Terraform and Kubernetes." The skill ceiling is "can you think clearly about problems and taste the difference between good and great."