Christine Lemmer-WebberCan't tell you how many times I have heard about a friend's company needing to send an apology email to customers about downtime and flakiness due to AIgen commits that were poorly reviewed and misunderstood
Hell even Microsoft did it https://blogs.windows.com/windows-insider/2026/03/20/our-commitment-to-windows-quality/
The slow part of software is NOT the initial generation of software. It's the maintenance and review of it.
If your management is pushing for 10x programmer output, hell even 40% more programmer output, what they're asking for is a stability crisis. There's no way around it. That's how it is right now.
You can use these tools for red teaming (caveat: you will get a lot of false positives also). You can sort of use them for prototyping (though a lot of the value of understanding building through the prototyping process may be lost during that time; still, it is one place where things can increase). Those two categories don't create huge and unresolved copyright output questions in your codebase, and I think you can justify them.
But if you're using them to actually write the software itself, you're borrowing against the future, against stability, and against institutional understanding of your own stack.
Oh yeah the other caveat about using them for prototyping, as @tamzin highlights below, is that "quickly thrown together prototypes" often become production code, to their authors' dismay.
In many ways, whiteboard prototypes are much better, as they are immune from this problem.
Kaye@cwebber We can deploy whiteboard prototypes directly to production. We can do this with the power of AI
Softwarewolf
Mx. Eddie R
Not a Spring Onion@silvermoon82 @faoluin @kaye @cwebber
Deploying new versions fast enough that "uptime" loses any meaning.
@kaye curse you someone has a startup for this right now don't they
Michael Bacon
Matt Maurer
Wayne Myers@tir3ur4m @MichaelTBacon @cwebber @kaye Why stop there? Why not just *generate* the whiteboard and see what happens? WCPGW etc
Eloy.
Ellie
Dan Sugalski@cwebber I've described previous attempts at work to Increase Velocity as strapping on rocket skates so we can careen headlong into a brick wall faster.
With AI codegen I think we've decided the rocket skates weren't fast enough or the brick wall big enough, and are going full Saturn-V-Into-The-Sun.
I'm sure it'll be fine though. Really. What could possibly go wrong?
Thommy
Luci Bitchface Angerfoot@cwebber i had said it before about LLMs in other contexts, but a few videos i watched today made me realise this absolutely applies to the LLMs writing software case: It’s gambling addiction
Richter
prom™️
Little Detritus@cwebber imho we are still lacking is a good taxonomy for maintenance.
Whilst "new code" can be easily measured by "lines of code" or through "new features" there is no metric for maintenance.
Because maintained code is a non-functional feature.
@d3sre did some amazing work on the other non-functional feature info-sec, to make the work of SOCs visible, see:
https://github.com/d3sre/IntelligentProcessLifecycle
Would you happen to know if anyone works on this?
Human Brain Enthusiast@cwebber I posit that writing code itself is never the bottleneck (otherwise it would have been solved with cheap offshore programmers long ago).
The hard work making software is designing it, including from a user experience point of view, your business needs and operational constraints.
Using generative AI to add code you don’t understand (or worse, features you don’t know why you add them) will make all of these things cumulatively harder.
Tamzin Hadasa Kelly@cwebber AI has proven very good at fixing two problems humanity didn't have: a shortage of labor and a shortage of cobbled-together first drafts of things being used in production.
Steve Hersey@cwebber
I want you to tattoo this onto some corporate executives...
I want you to tattoo this onto some corporate executives...
ǝʌɐp@cwebber There's the GitHub nines, but I'm wondering when we will start seeing more hard numbers that things have generally gone to shit. The data is probably out there but siloed and guarded in hushed tones.
@cwebber People would be right to point out that the start of the quality decline seemed to precede coding agents, but I think it's a bit of a perfect storm. My personal theory has been that it started with US interest rates going back up and austerity measures being introduced across the industry. Coding agents really meshed with that because everybody has been looking for areas to cut cost.
Craig S. Kaplan@cwebber Happened to read a nice essay on this topic, which I saw shared on Mastodon this morning (thanks, @inthehands): https://andrewmurphy.io/blog/if-you-thought-the-speed-of-writing-code-was-your-problem-you-have-bigger-problems
If you thought the speed of writing code was your problem - you have bigger problems | Debugging Leadership
AI coding tools are optimising the wrong thing and nobody wants to hear it. Writing code was already fast. The bottleneck is everything else: unclear requirements, review queues, terrified deploy cultures, and an org chart that needs six meetings to decide what colour the button should be.
Sharlatan@cwebber It feels like nobody in a corp world gives a shit to think about - code reuse, code maintenance, code legacy, time spent on support - what they need is a feature generator machine without any logic
I wonder which type of swards Japanice smiths had if they allowed to change the technology each time the shōgun ordered a new feature for the war the plaid
Krzysztof Sakrejda@sharlatan @cwebber that definitely don't, ask anybody who writes code for a bank
Paul Wilde 
@cwebber I was literally trying to put this into words today.
Thank you for writing it in so much of a better way than I ever could!
Aedius Filmania ⚙️🎮🖊️well, the mantra at work is "move fast and break things" but for years some peoples were able to make a decent job and slow down just a little bit to not break everything ...
Now the tide is too high, the CTO is pushing in production untested code, i have coworker that run scripts on their personal computer opening chrome login with their account and updating things.
Everyday now, slack is about broken pages, broken features, action made "'by mistake".
We are so doom.
T2R
William B Peckham@cwebber we had an instructor once who gave us a quote, and I don't remember where the quote came from, but it was that "coding is the smallest part of software. It's 50% design and 30% coding and 99.9% long-term maintenance, changes, updates, and bug fixes." he maintained that it doesn't have to be that way. If you take ten times as long on design and development and do it right, then your long term maintenance can be cut to one tenth of that. Of course, almost no company does it that way because they see the initial cost, and make no plans to avoid the long-term cost. Everyone wants it faster and cheaper, and as a result, they get crud. I took this to heart in my career, and when I was doing development, I always quoted a time that was ten times what I thought it would take. They didn't like it, but they accepted it and were happy when I got it down a little bit early. In the process, I avoided having to do a ton of maintenance on my old code.
@cwebber AWS too.
