#Software Engineering

What the sources say

These analyses explore how artificial intelligence agents are reshaping software engineering from manual coding to architectural orchestration and the skills developers need to adapt.

Andrej Karpathy and Boris Cherny agree that software engineering is shifting away from manual typing toward autonomous workflows^[2][3]. Karpathy notes that developers will express their will to agents instead of writing lines of code^[2]. Similarly, Cherny explains that engineers using Claude Code transition from writers to reviewers and architects^[3]. Matt also discusses this shift to agentic workflows and notes that developers must learn to orchestrate these new systems^[1].

A key difference emerges in how developers should approach the underlying artificial intelligence. Matt argues that engineers should focus on the harness, which includes prompts and codebase architecture, rather than chasing new models^[1]. In contrast, Cherny states that Anthropic builds tools specifically for the models of six months from now because capabilities improve rapidly^[3]. A reader can conclude that while the role of a programmer is changing entirely, success requires both mastering current environments and planning for future automation^[1][2][3].

Written by an AI model from Google, from the analyses on this page.