Can definitely attest to this. The frequency of outages at my company have increased drastically the past year, especially ever since incorporating agentic development. I’m seeing all of the dev best practices go out the window. We have a few vibe coders that are posting 15-30 PR’s per day. It’s way too much for us to review. We’re not a big shop. I think we’re going to have to hire more people just to review code across the industry. And those people will have to know how to actually write software otherwise what are they even reviewing. Maybe the models will get so good they never make a mistake. Doubt it.

The proposed industry solution is to use agents to review PRs, as not to slow down the velocity of delivery...

My current workplace is going through a major "realignment" exercise to replace as many testers with agents as humanely possible, which proved to be a challenge when the existing process is not well documented.

The fact that anyone in leadership would ever think this is even remotely possible - given my experience in the general state of requirements / contracts / integrations / support - makes me bleed from my earholes just a little bit.

It's starting to just feel a little like an excuse to call everyone on deck for "a few weeks trying 9-9-6". But even then the lack of traction isn't between the eyeballs and the deployment. You'll still be spinning wheels in that slippery stuff between what a customer is thinking and what the iron they bought is doing.

So you essentially trust the output of the model from beginning to end? Curious to know what type of application you're building where you can safely do that.

Edit: to clarify, I know these models have gotten significantly better. The output is pretty incredible sometimes, but trusting it end to end like that just seems super risky still.

I guarantee you it's nothing quantifiable.

LLMs can't be responsible for deciding what code you use because they have no skin in the game. They don't even have skin.

If you type fast, well then it takes just as long to code it yourself as review it. Plus you actually get flow time when you're coding.

For heaven's sake people have the robot write your unit tests and dashboards, not your production code. Otherwise delete yourself.

"Hey Claude, did Claude do a good job?"

I did an experiment today, where I had a new Claude agent review the work of a former Claude agent - both Opus 4.6 - on a large refactor on a 16k LOC project. I had it address all issues it found, then I cleared context, and repeated. Rinse and repeat. It took 4 iterations before it approached nitpicking. The fact that each agent found new, legitimate problems that the last one had missed was concerning to me. Why can’t it find all of them at once?

You're expecting it to be a person. It's not.

It is more like a wiggly search engine. You give it a (wiggly) query and a (wiggly) corpus, and it returns a (wiggly) output.

If you are looking for a wiggly sort of thing 'MAKE Y WITH NO BUGS' or 'THE BUGS IN Y', it can be kinda useful. But thinking of it as a person because it vaguely communicates like a person will get you into problems because it's not.

You can try to paper over it with some agent harness or whatever, but you are really making a slightly more complex wiggly query that handles some of the deficiency space of the more basic wiggly query: "MAKE Y WITH NO ISSUES -> FIND ISSUES -> FIX ISSUE Z IN Y -> ...".

OK well what is an issue? _You_ are a person (presumably) and can judge whether something is a bug or a nitpick or _something you care about_ or not. Ultimately, this is the grounding that the LLM lacks and you do not. You have an idea about what you care about. What you care about has to be part of the wiggly query, or the wiggly search engine will not return the wiggly output you are looking for.

You cannot phrase a wiggly query referencing unavailable information (well, you can, but it's pointless). The following query is not possible to phrase in a way an LLM can satisfy (and this is the exact answer to your question):

- "Make what I want."

What you want is too complicated, and too hard, and too unknown. Getting what you are looking for reduces to: query for an approximation of what I want, repeating until I decide it no longer surfaces what I want. This depends on an accurate conception of what you want, so only you can do it.

If you remove yourself from the critical path, the output will not be what you want. Expressing what you want precisely enough to ground a wiggly search would just be something like code, and obviates the need for wiggly searching in the first place.

I wonder if the PR workflow is just unsustainable in the agentic era. Rather than review every new feature or bug fix, we would depend on good test coverage, and hold developers accountable for what they ship.

The result might be more faulty code getting merged, but if you already have outages and can't review every PR, is there currently a meaningful benefit to the PR workflow?

This is the "if you're already letting faults through, why not give up trying to stop faults?" approach.

The alternative might be "what if we could get the genie back into the bottle?"

We know some people are using LLMs to evaluate PRs, the only question is who, and how strong the incentive is for them to give up.

> I wonder if the PR workflow is just unsustainable in the agentic era. Rather than review every new feature or bug fix, we would depend on good test coverage, and hold developers accountable for what they ship.

I think what you're describing is setting up the human as the fall guy for the machine.

Diogenes carrying a lamp, looking for good test coverage

Copy-pasting screenshots of red lines.

This reminds me a bit of monoliths vs microservices. People would see microservices as the next new shiny thing and bring it with them to their next job, or read a great blog post that sounds great in theory, but falls apart in practice. People would see it as as purely architectural decision. But the reality was that you had to have the organizational structure to support that development model or you'd find out that it just doesn't scale the way you expect and introduces its own sets of problems. My experience is that most teams that didn't have large orgs got bogged down by the weight of microservices (or things called "microservices"). It required a lot of tooling and orchestration to manage. But there was this promise that you could easily just rewrite that microservice from scratch or change languages and nobody would notice or care.

LLM-generated code feels the same. Reviewing LLM-generated code when it's in the context of a monolith is more taxing than reviewing it in the context of the microservice; the blast radius is larger and the risk is greater, as you can make decisions around how important that service actually is for system-wide stability with microservices. You can effectively not care for some services, and can go back and iterate or rewrite it several times over. But more importantly, the organizational structures that are needed to support microservice like architectures effectively also feel like the organizational structures that are needed to support LLM-generated codebases effectively; more silo-ing, more ownership, more contract and spec-based communication between teams, etc. Teams might become one person and an agent in that org structure. But communication and responsibilities feel like they're require something similar to what is needed to support microservices...just that services are probably closer in size to what many companies end up building when they try to build microservices.

And then there are majestic monoliths, very well curated monoliths that feel like a monorepo of services with clear design and architecture. If they've been well managed, these are also likely to work well for agents, but still suffer the same cognitive overhead when reviewing their work because organizationally people working on or reviewing code for these projects are often still responsible for more than just a narrow slice, with a lot of overlap with other devs, requiring more eyes and buy-in for each change as a result.

The organizational structures that we have in place for today might be forced to adapt over time, to silo in ways that ownership and responsibility narrow to fit within what we can juggle mentally. Or they'll be forced to slow down an accept the limitations of the organizational structure. Personal projects have been the area that people have had a lot of success with for LLMs, which feels closer to smaller siloed teams. Open-source collaboration with LLM PRs feels like it falls apart for the same cognitive overhead reasons as existing team structures that adopt AI.

Maybe it’s time to have multiple agents and models review the PRs and also provide context for easier human review. That and lots more focus on robust testing.

There’s no way velocity will decrease now that upper management is obsessed with AI.

Sounds like people need to speak up to management

Management doesn’t care. This sort of thing is becoming more common at my workplace too. More outages, more embarrassing bugs, even bugs that leak customer data. The solution is always more AI, and if you’re still shipping bugs and causing outages, it’s because you did’t use the AI correctly. Leadership makes all the right noises about quality and ownership, but when it comes down to it, the incentive structures clearly prioritize shipping things faster, all else be damned.

Sounds like a fast track to sinking their company into the ground

Management wants to get rid of people; they want to have their "wish-machine" that does what they say without any need to deal with nerds or ethical issues.

Management likes how fast features are getting deployed so they essentially told us to just deal with it.

I mean speak up to management in a way where they know it's stupid and they're stupid for pushing it

People pushing dozens of PRs per day need to learn to prioritize tasks, and balance a bit more towards quality over quantity.

This is the way. There's nothing inherently wrong with using AI as long as it's used responsibly.

I highly doubt there are any managers or executives who care how AI is precisely used as long as there are positive results. I would argue that this is indeed an engineering problem, not an upper management one.

What's missing is a realistic discussion about this problem online. We instead see insanely reckless people bragging about how fast they drove their pile of shit startup directly into the ground, or people in denial loudly banging drums to resist all forms of AI.

And maybe spend some time doing reviews for other developers. And if they aren't qualified to be, then maybe spend that time becoming qualified rather than pumping out more slop.

I love it. I was getting burnt out due to ADHD or autism burnout but with AI tooling I’m able to work a full week without burnout. I think the kind of burnout I get is helped with these tools, but since I’m not neurotypical it’s different from the burnout people are getting from doing too much.

I do see “task expansion” happening often though. If I can do the full feature rather than doing baby steps I’ll often do that now, because wrangling code is easier.

Using vibe coding for frequent PRs seems insanely reckless.

In my scientific computing environment, the majority of my vibe coded output goes to one-off scripts, stuff that is not worth committing (correcting outputs, one-off visualizations, consistency checks), and anything worth committing gets further refined to an extent that it pretty much can't be considered vibe coded anymore. It's simply too risky, any bugs would propagate down to decision making for designing new, expensive instruments.

I imagine that the cost and trust risks in enterprise environments are similar, so this seems very reckless.

AI Agents have helped up my productivity, but that's specifically because I can focus on the science, and delegate the auxiliary things to AI. I also believe I get this productivity out of them because my supervisor really drove home how hard I need to go on consistency checks and years of having my visualizations nitpicked (so I am able to do the same to AI and recognize when results are suspicious).

Most people don't care. Leadership is demanding feature, feature, feature. IC are worried about losing their jobs and outages rarely cost most business actual money. So garbage gets shipped, outages rise, everyone is burned out but since they can't find another job, they remain.

In this situation, you raise the issue with management, with a paper trail that Cover Your Ass, that the pace is unsustainable and bugs will continue to accumulate faster than it can be fixed. Make sure that you are not responsible for it and ensure this is known by all (including management).

You then continue to vibe code as instructed by management. No burnout because you are not responsible anymore.

People still get burnout if not from constant pages, the late nights and worry about their job.

I love vibe coding for little tools like that. Tools which can have their outputs quickly validated, and then throw them away. Like a jig in woodworking.

I'm a mostly solo dev, and I'm finding that being purely code-review for an AI is sub-optimal. Too often the AI runs off down bad paths which you only realise later, and unpicking the mess is most likely a productivity loss.

Working more as a pair, or essentially doing code review as you go, in small chunks, is significantly better.

I personally don't have the setup of tokens to spend to say "go build this entire thing" and then review 15k loc. I also find even opus is poor at coming up with tests to justify the business logic it's meant to be implementing.

Is there any publication which demonstrates that the improvement is really 10x?

It's like "decimate" -you would think 10x had literal force, but it's more figurative. It just means "moar"

(decimate had specific literal intent. Now it's just a force modifier like bigly)

> Removing 1/10

feels euphemistic for the original “colloquial” usage I have for it.

> The killing of one in ten, chosen by lots, from a rebellious city or a mutinous army was a punishment sometimes used by the Romans. The word has been used (loosely and unetymologically, to the irritation of pedants) since 1660s for "destroy a large but indefinite number of." [0]

[0] https://www.etymonline.com/word/decimate

Yup. What amuses me is that people think that decimate is to massively degrade something. I assume they're thinking "reduce to 1/10th" rather than "reduce to 9/10th". The effect is markedly different

A watched pot never boils. A watched vibe coder never 10x-es.

You can write your own linters for every dumb AI mistake, add them as pre-commit checks, and never see that mistake in committed code ever again.. it’s really empowering.

You don’t even have to code the linters yourself. The agent can write a python script that walks the AST of the code, or uses regex, or tries to run it or compile it. Non zero exit code and a line number and the agent will fix the problem then and rerun the linter and loop until it passes.

Lint your architecture - block any commit which directly imports the database from a route handler. Whatever the coding agent thinks - ask it for recommendations for an approach!

Get out of the business of low level code review. That stuff is automatable and codifiable and it’s not where you are best poised to add value, dear human.

… how are you getting actual usable output at that scale? I have to baby my AI in 1 minute increments or it just doesn’t arrive at the correct solution at all.

Using Codex 5.2

Perhaps the prompts you are using could do with some love. We're pretty consistently getting great results up to and beyond the 10 minute mark in a large monorepo.

We tend to use Opus 4.6 High and GPT 5.4 High.

I mean, why do you think people are burning out?

Due to prolonged stress, which lack of control is the main contributor e.g. you have expectations, you cannot control variable x,y,z, which leads to stress, which over long period of causes burn out.

In case it wasn’t obvious, I was being facetious. You can’t just let the AI rip without putting effort into constructing good input and verifying the output and expect anything good to happen, which is what the gp was asking.

There’s no secret into how people are getting “10x”, or at least claiming to, they’re just working more.

I can attest to this. Ultimately I dont think it is possible to 10x output systems with AI and actually keep the traditional quality controls (yet.)

IMHO you just need two stacks -- systems where you can play fast and loose and 10x output. And systems where quality matters where you can perhaps 1.5 or 2x. That is still a lot of output.

I feel this is not discussed enough. I can attest to this 100%.

Just the past weekend, I was talking with a very senior engineer (~distinguished engineer at a very large tech co) who basically said he's working 8-8-6 (8 am - 8 pm, 6 days/week), "writing code" (more like supervising 8-15 agents) for a product demo in 2 weeks, which otherwise would have taken at least 1 quarter's worth of time with a small team. He's zonked out, fwiw. There are no junior engineers in the team ¯\_(ツ)_/¯, most having been laid off a few months ago.

The toll it takes, and the expectations of AI-driven productivity, have only increased dramatically. At some point, the reality will hit the remaining engg team. Not sure if the company or its leadership realizes, but so far, it's all-AI, all-the-time, human cost of productivity be damned.

If if this person really is a distinguished engineer, then they are part of leadership and it's their responsibility to set realistic expectations. Leadership knows this, they just don't care and won't care until the job market improves.

> more like supervising 8-15 agents

How do they do it? (My own record is 5 agents, but it is not typical). Do they use gastown or something?

I often have 10+ running in parallel. I’m attacking parallel problems that aren’t interdependent. Sometimes adding additional products can bring me up to 15+.

Gotta have really good test harnesses so they can largely fix themselves.

But how do you cover such amount of multi tasking? Could you give an example? I mean what kind of tasks allow such a parallelization?

context switching across the entirety of the feature surface for an app

You could easily have agents to work on login page, messaging feature, database/data model update, recommender system, backend api, etc

But you have to keep it in your head, and remember all stuff at the same time. How is it possible to track, and do reviews one after another? Or are these pretty long running agents?

We have our doubts about this. Can you share your code or product? Anecdotally, my mistakes and lack of understanding exponentiate the more I try to parallelize.

Honestly, I dont know. I could be mistaken about the exact number of agents - but not wrong about fact of AI-driven workflows which is heavily automated, and goes on for hours.

He's one (small) step from distinguished engineer, with 20+ patents to his name, and is an embedded programmer (largely C/C++) with 30+ years of experience in the field; and I've known him for nearly as long, so I put a lot of credence to his words.

But we don't usually talk work; he's the guitarist in our band :) [I'm the bass] So we mainly chill over music + beer. And lately, it's been less chill ¯\_(ツ)_/¯

> The industry calls this “10x productivity.” I call it what it is: a system that generates output at machine speed and forces humans to process it at biological speed.

The question is can you tolerate the amount of PRs thrown at you per day on top of reviewing the exponentially growing mess of code that continues to double every hour and being paid less for it.

Just learn to say no and leave. Why do you tolerate the increasing comprehension debt that is loaded on to you.

You will never get that time back. Just give it to someone else that thinks it is worth maintaining that slop for less.

The job market under our Great Leader has taken away a lot of this agency. Software engineers have gone from having the pick of the market for themselves to becoming (perceived as) next to disposable.

That's a very American-centric point of view; the job market worldwide for developers is getting tougher and tougher.

And sometimes they do build interesting things but also leave a trail of destruction behind them. It reminds me of ‘moving fast break things’.

Given the research cited in the article it seems bigger than an anecdote about one guy who doesn't know how to do work life balance

70%+ saying that AI has increased their workload AND that they are burning out because is it.

Yeah, it's all well and good to say somebody doesn't know how to regulate their pace, and it's another thing for your manager to tell your team that you need to be using a squad of agents constantly. To have a weekly stand-up that is specifically and solely for the purpose of talking about your "AI wins" for the week. To be told that you will be evaluated on how much you're using AI for your job.

When your manager and your company regulate your pace for you with the understood threat that not using AI will risk your job, you don't really have much of an option.

That sounds horrible. I'm so glad I saved enough money to retire early right before this madness started.