Do engineers still read the code agents write?

“PostHog is just a little further than that with agents writing about 70% of our PRs, but humans still skim at least 80% of those.”

Jina YoonAuthor, PostHog newsletter · Practitioner, PostHog

“Given zero hand-written code and zero traditional review, we required a system that could: Grow from cascades of natural-language specifications Be validated automatically without semantic inspection of source”

StrongDMPublished by the organization itself (documentation), no named author

“every engineer must be able to explain, even if an AI generated the PR, must be able to explain what they built.”

Ahmad Al-DahleChief Technology Officer · Executive, Airbnb

“We built a Software Factory: non-interactive development where specs + scenarios drive agents that write code, run harnesses, and converge without human review.”

StrongDMPublished by the organization itself (press report), no named author

“From June to September, merged pull requests to Claude Code's codebase more than doubled while the share that had to be reverted fell by more than 40 percent. We didn't remove review. But we did move most of the reading to machines, optimized for human time and attention, and invested heavily in making it fast and easy to undo mistakes.”

AnthropicPublished by the organization itself (post), no named author

“We think "did a human read the code?" is an imprecise proxy for "what could go wrong with this change, and which check—automated or human—is the right one to catch it?"”

AnthropicPublished by the organization itself (post), no named author

“Pulling a senior engineer out of deep work to catch a routine mistake costs far more than an automated review. People on the team are still responsible for: … Reviewing risky code such as billing, the auto-updater and CI workflows. … Owning the outcome. A person is accountable for every change that ships, whoever typed it.”

AnthropicPublished by the organization itself (post), no named author

“We do no code reviews. We require every engineer to work on things in parallel, and we have no standardization how people do AI.”

Vaibhav GuptaCo-founder and Chief Executive · Executive, Boundary (BoundaryML)

“So today I want to talk about what I think works in the real world and what we've started doing at HumanLayer, which, to be clear, is still building loops, right? I think loops are super powerful, but we can design loops and still read the code. In fact, we can design loops that make it easier to read the code because the loops are making the code better.”

Kyle MisteleCo-founder, HumanLayer · Executive, HumanLayer

“And so my advice is … instead of trying to automate everything end-to-end, build these small incremental loops throughout your system, and you will wake up one day and you will be moving two to three times faster while still being able to read the code, while still owning the architecture.”

Dex HorthyCEO, HumanLayer · Executive, HumanLayer

“while they're human-reviewed, they contain no human-written code.”

Alistair GraySoftware engineer, Leverage team · Practitioner, Stripe

“Now, you asked specifically, do I write the code? I read … and what this says, I'm doing more and more and more, is asking the AI to explain back what it did to me, which is a little bit of the job I've compared it to being an engineering manager.”

Guillermo RauchFounder and CEO of Vercel · Executive, Vercel

“right now we still require an engineer to review every PR … goes in. … Once the more confidence we get in our risk scoring and we can also back test it with years of data, right? We have all of this data and the more confidence we have in our AI code reviews, then maybe for some class of low risk PRs, we could enable, we call it green lane like auto landing.”

Uday Kiran MedisettyDistinguished Engineer, Uber · Practitioner, Uber

“You just tell the AI to go nuts. It makes a pull request. You just say, “Yep, that’s good.” You don’t really read it, and now it has to be reviewed by your colleagues.”

Tobi LütkeCo-founder and Chief Executive Officer · Executive, Shopify

“actually talked to Claude about the code and say, all right, these are the questions that I would have. Can you go investigate it? So it is kind of … code review, but still human-driven, and for the really important ones. And for the ones that are cosmetic visual changes, it's much more like we'll fix forward if we need to fix forward.”

Mike KriegerHead of Anthropic Labs · Leader, Anthropic

“So one of the reasons we built … artifacts that we shipped a couple of weeks ago was partially for that, which is you would send somebody a PR, and then they'd be like, I don't know, man, this is like 2000 lines of code. It looks like code to me. And what we started doing instead is sharing much more like, here's a … artifact. Here's the explanation. Here's the intention of the change. Here's the trade-offs that were made.”

Mike KriegerHead of Anthropic Labs · Leader, Anthropic

“I think we found that the state-of-the-art in December was you can probably, run this for about two weeks. By the end of those two weeks, you’d find that, hey, you want to, change the color of a button. Well, it turns out this button is implemented in, 10 different places, and they, have All these different variations, and oh, you forgot one of them, and actually it’s a slightly different color in one spot.”

Walden YanCo-founder and Chief Product Officer · Executive, Cognition

“And then you can see a whole bunch of these are merged and then we have some failures that say, hey, we've exceeded max rework count. And that's essentially an escalation to me saying, hey, you should probably look at this yourself as a human because you're responsible for the code that it generates.”

Shardul VaidyaSolutions Architect, Data and AI, AWS · Practitioner, Amazon