DOP 375: AI Writes the Code. Who Understands It?

November 4, 2026 Episode 375

Show Notes

#375: An agent can crank out 200 or 300 lines of code a minute. You can maybe review 40. So who’s actually supposed to understand what just got merged?

Maybe nobody ever did. Did you really read every line of every PR before AI showed up? Or did you just have the time? Now the junior has an agent, the senior is buried in review, and the job looks a lot more like management than coding.

In this episode, Darin and Viktor talk about who’s supposed to understand the code once AI writes it, and why that’s now the senior engineer’s job.

Frequently Asked Questions

Do you need to review every line of AI-generated code?

No, according to Viktor Farcic on DevOps Paradox episode 375. He says you should review AI-generated code the same way you should always have reviewed human pull requests, focusing on what matters. He spends most of his review time on interfaces, whether an API or a user interface, because a well-built interface shows the author understood the problem. In his view, people reviewed every line of a diff in the past because they could, not because they should.

Should a company hire a junior developer or spend that money on AI agents?

Viktor Farcic says on DevOps Paradox episode 375 that he would put the money into agent usage for an existing senior without doubt, but only if he ignores the future, because companies still need future seniors. He still thinks larger organizations need juniors to take on work that would waste senior time. Darin Pope argues that junior work should mix grunt tasks with validation work, roughly three days of one and two of the other, so juniors grow instead of leaving.

Are developers or managers more at risk of losing their jobs to AI?

Viktor Farcic argues on DevOps Paradox episode 375 that traditional managers should be more afraid than developers. In his view, developers are only under threat if they fail to take over management work. He thinks a coder is better qualified than a non-technical manager to become the very technical product manager the job now needs. He describes running five agent teams in parallel, 25 agents in total, doing work that once took six managers and 25 coders.

Does AI brain fry get easier with experience?

Viktor Farcic says on DevOps Paradox episode 375 that five parallel tasks caused far more brain fry six months earlier than they do now, because his agents need fewer interventions. By his own rough estimate, he once stepped in around 50 times per feature and now does so about five times at the start and a few more at the end. His agents now run autonomously for hours, so he takes on more parallel work and finds the overall load equally hard.

Do developers understand their systems less when AI writes the code?

Viktor Farcic says on DevOps Paradox episode 375 that he understands more than ever in absolute terms, but less as a percentage of the system, because he now builds several features a day. He adds that agentic AI makes learning a codebase easier than it has ever been. He can ask an agent to explain something and get an answer in a minute, where before only the person who started the project could explain anything beyond trivial code.

How should you evaluate a new AI model?

Viktor Farcic explains on DevOps Paradox episode 375 that when a new model arrives, he takes a task he is about to do that day and gives it in parallel to the model he used yesterday and to the new one. He then compares the time, the error rate and the rest of the results. Darin Pope suggests a second test: point the new model at your own application, ask it to find the problems, and walk away.

What is the DevOps Paradox podcast?

DevOps Paradox is a weekly podcast co-hosted by Darin Pope and Viktor Farcic, covering DevOps, platform engineering, and modern software delivery. Episode 375 asks who understands code once AI writes it. It covers how to review agent output, whether to hire juniors or spend on agents, why traditional managers face more risk than developers, and how AI brain fry changes with experience. Every episode page carries audio, video, and a full transcript.

Episode Transcript

Read the Full Transcript

Hosts

Darin Pope

Viktor Farcic

Viktor Farcic is a member of the Google Developer Experts and Docker Captains groups, and published author.

His big passions are DevOps, Containers, Kubernetes, Microservices, Continuous Integration, Delivery and Deployment (CI/CD) and Test-Driven Development (TDD).

He often speaks at community gatherings and conferences.

He has published DevOps Paradox and Test-Driven Java Development.

His random thoughts and tutorials can be found in his blog The DevOps Toolkit.