Will AI replace software engineers?
One guest says the agent now writes the code. Another says writing code was never the job.
One view
The agent writes the code now — the old workflow is going away.
Anush Elangovan, AMD EP 56 · 250,000 Lines of Code/Week: Inside an AMD VP's Agent-First Workflow | Anush ElangovanSure, that's my agent's problem. And I'm not opening a file.rs ever.
The other
Coding was never the point — engineering is solving problems, and that part stays.
Charity Majors, Honeycomb EP 18 · AI Won't Solve Your Toughest Engineering Problems | Honeycomb’s Charity MajorsSoftware engineering is not about writing code. It is about solving business problems with technology, which has historically included a lot of writing code.
Elangovan describes handing a coding task to an agent, while Majors defines engineering as solving business problems with technology. Both positions can hold: an agent can write code while the engineering job remains centered on the problem it solves.
What should your team own?
Own the problem: Laurie Voss, at 13:04, argues that product sense means deciding what to build, not just how to build it. Before delegating implementation, write down whose problem the change solves, what success looks like, and who can decide whether it worked. This is an editorial recommendation drawn from his argument, not evidence that engineering jobs have disappeared.
Own the acceptance criteria: Anush Elangovan, at 18:39, describes putting correctness and performance requirements into a test harness that gates his agents' changes. Ask which objections can become executable checks, and which still require a person to judge. His workflow is a useful example, not a universal replacement for human review.
Own the understanding: Greg Foster, at 19:17, warns that teams can lose context when they neither write nor absorb generated changes in review. For the next agent-written change, ask someone to explain what changed, why it is safe, and how the team would diagnose a failure. Passing tests and retaining that understanding are different responsibilities.
Own the apprenticeship: Charity Majors, at 9:06, describes engineering as an apprenticeship industry where people learn at work and from each other. If agents take over tasks that used to teach context, decide how newer engineers will practice specifying, reviewing and operating real changes. These interviews leave an open question: what evidence would show that an agent-heavy workflow develops judgment as well as shipping software?
Beyond the podcast
Controlled studies point in different directions, which is part of why the argument persists. In a GitHub experiment, developers using Copilot implemented an HTTP server in JavaScript 55.8% faster than a control group (Peng et al., The Impact of AI on Developer Productivity). In a 2025 randomized trial, 16 experienced open-source developers working on 246 tasks in mature projects they knew well took 19% longer when AI tools were allowed, even though they had forecast a 24% speedup and afterward believed they had gained 20% (Becker et al., Measuring the Impact of Early-2025 AI on Experienced Open-Source Developer Productivity).
Usage keeps rising while trust does not. In Stack Overflow's 2025 Developer Survey, 84% of respondents were using or planning to use AI tools, up from 76% a year earlier, and 51% of professional developers used them daily. But more developers distrusted the accuracy of AI output (46%) than trusted it (33%), the most common frustration was solutions that are almost right but not quite (66%), and 76% did not plan to use AI for deployment and monitoring (Stack Overflow, 2025 Developer Survey: AI). Writing code is changing fastest; checking it and owning what ships is where developers are most cautious, which fits both positions above.
Conor Bronsdon's September 2026 essay, Software Engineering is Dead. Long Live Product Engineering, develops this distinction through Laurie Voss's argument for product engineering: agents can write code, while understanding customers, translating their needs into requirements, and deciding what to build remain valuable skills. Bronsdon also raises an apprenticeship problem: if agents take on the implementation and review work that helped junior engineers learn, teams need new ways to develop their judgment.
Is AI taking software engineering jobs? covers the hiring side, who reviews AI-generated code? covers verification, and the AI coding topic collects the episodes.
More debates
- Is the agentic AI boom a bubble? Spicy take versus the long view on whether agents are hype or here to stay.
- Where do AI agent frameworks add value? Two founders emphasize different parts of the problem: operating the agent stack and improving the context it uses. These positions can coexist.
- Do you need to fine-tune a model? Habib says Writer's models spare customers fine-tuning. Srinivasan says most open-model use cases she sees involve fine-tuning.
- Should AI companies build in the open? One guest hopes more companies take an open source route with AI. Another believes in open source but says it is hard to turn into a profitable business.