scripod.com
Head of Claude Code: What happens after coding is solved | Boris Cherny

Overview

Shownote

Highlights

Transcript

Chapters

Pins

Head of Claude Code: What happens after coding is solved | Boris Cherny

Lenny's Podcast: Product | Career | Growth

Feb 19
Lenny's Podcast: Product | Career | Growth

Lenny's Podcast: Product | Career | Growth

Feb 19
OverviewShownoteHighlightsTranscriptChaptersPins

Shownote

Boris Cherny is the creator and head of Claude Code at Anthropic. What began as a simple terminal-based prototype just a year ago has transformed the role of software engineering and is increasingly transforming all professional work. We discuss: 1. Ho...

Highlights

Boris Cherny, head of Claude Code at Anthropic, shares insights from the rapid evolution of AI-augmented software development—how a scrappy terminal prototype became foundational to engineering workflows and catalyzed broader agentic tools like Cowork.
00:00
Coding is largely solved by AI
03:49
He missed Anthropic's safety-focused mission, which originally attracted him to the company and is essential for his job satisfaction
05:35
Claude Code authored 4% of all public GitHub commits in its first year
08:41
The model demonstrated surprising autonomy in using tools independently
13:29
By November, the speaker personally reached 100% AI-written code
15:01
In the early days of Claude Code, they didn't think it would be useful, but gradually its capabilities improved until it reached 100% of their coding workflow
16:17
100% of Anthropic's code has been written by Claude Code since November
19:58
Since introducing Claude Code, Anthropic has seen a 200% increase in engineer productivity
22:25
A newer engineer solved a memory leak faster using Claude Code than the speaker did with traditional debugging
24:02
Giving engineers unlimited tokens to experiment with AI leads to better use of AI tools and more productivity
26:49
Be liberal with token usage as interesting ideas come from pushing the limit
30:20
Programmers have always adapted to new tools—from pen-and-paper to AI—so skill atrophy fears are overblown
34:52
The fun part is determining what to build, engaging with users, and collaborating
38:23
AI has made coding more enjoyable and the Claude Code team is hiring
40:42
In the future, distinctions between engineering, design, and product management will become murkier, possibly with titles like 'builder' becoming more common
44:37
70% of engineers and PMs enjoy their jobs more since adopting AI tools, versus 55% of designers
51:26
For Claude Code, the product is the model—exposed with minimal scaffolding so it decides tool usage and order
51:54
Cowork was built in 10 days and quickly used by millions
54:04
Releasing AI early enables discovery of latent demand and real-world safety learning
1:01:59
His mother drew on the punch cards as a childhood memory
1:02:26
Their families now toast to America—sometimes still with vodka
1:08:09
Models can now run unattended for hours or longer
1:08:39
Using Opus 4.6 can be cheaper and less token-intensive than smaller models
1:11:19
Their team focuses on solving user problems, not spending much time on competing products
1:12:13
If not at Anthropic or post-AGI, Boris would make miso
1:20:57
Cowork is Anthropic's new AI agent that can perform tasks like filling out forms and is growing faster than Claude Code did a year ago

Chapters

Introduction to Boris and Claude Code
00:00
Why Boris briefly left Anthropic for Cursor (and what brought him back)
03:45
One year of Claude Code
05:35
The origin story of Claude Code
08:41
How fast AI is transforming software development
13:29
The importance of experimentation in AI innovation
15:01
Boris’s current coding workflow (100% AI-written)
16:17
The next frontier
17:32
The downside of rapid innovation 
22:24
Principles for the Claude Code team
24:02
Why you should give engineers unlimited tokens
26:48
Will coding skills still matter in the future?
27:55
The printing press analogy for AI’s impact
32:15
Which roles will AI transform next?
36:01
Tips for succeeding in the AI era
40:41
Poll: Which roles are enjoying their jobs more with AI
44:37
The principle of latent demand in product development
46:32
How Cowork was built in just 10 days
51:53
The three layers of AI safety at Anthropic
54:04
Anxiety when AI agents aren’t working
59:35
Boris’s Ukrainian roots
1:02:25
Advice for building AI products
1:03:21
Pro tips for using Claude Code effectively
1:08:38
Thoughts on Codex
1:11:16
Boris’s post-AGI plans
1:12:13
Lightning round and final thoughts
1:14:02

Transcript

Boris Cherny: 100% of my code is written by Claude Code. I have not edited a single line by hand since November. Every day I ship 10, 20, 30 pull requests. So I got the moment I have like five agents running. Lenny Rachitsky: While we're recording this? Y...