Yep requirement gathering takes forever Then validation takes forever

Writing code is lesser problem than figuring out what we want when we want, and to get stakeholders at one place.

Finally, a fella who gets it.

Apple has already shown this decades ago - they got the iPhone and iPod developed and out the door in relatively short-time scales given the impact of the products on the world. Once you know what you want, exactly what you want, things moves fast - really fast.

I've seen biz team taking 3 4 months to do UAT

But sure let's buy 200$ per month claude to ship things faster lol

Lol yeah. Much of the issues re. lack of productivity comes from management who dont have the ability to focus+clarity+confidence of where to go and also create the environment to get the max out of people.

If everyone was 10x productive then we would have had native Claude Code app for each platform.

Instead they are using Electron and calling it a day. Very ironic isn't it? If AI is so good then why don't we get native software from Anthropic?

Shift from quality to velocity started not yesterday but AI only accelerated this shift. Majority of comments here tell how fast Claude can generate code, not how good the result is. Electron is the prefect fit for move fast mindset. It is likely that Claude developers don’t see electron as problem at all.

I understand why they used Electron. But since AI is so amazing they should be using Claude to generate super optimized native software! They do claim that Claude is better than humans and AI is replacing programmers any day now.

So why aren't they using their own software to generate a linux optimized package for linux, a Swift software for MacOS and whatever windows uses.

That would be the best ad for AI. See, we use our own product!

But it doesn't happen.

So essentially by not generating custom binaries for every platform and using Electron they're doing one thing but saying something else. So maybe generating code isn't the #1 problem in the world!

Also I remember them saying that their engg write less code and they use Claude to write Claude. If Claude can be used to write Claude then why not use Claude to write OS specific binaries?!

GenAI currently has substantially more training from web tech than native.

Did you all read about the aws outage for 13hrs because their autonomous AI agent decided to delete everything and write from scratch?

and they blamed the engineer because insurance doesn't pay for AI failures

We're now in the post-scrutinize era.

I've seen some companies expecting 2x output from Devs because they now get Cusor licence

I've written an async agent. It's triggered by a http request. It does a specific processing and updates a database table regarding it's output

Has anyone noticed that in the past year we have seen a LOT of outages?

Yes. Feels like every other week.

That goes against all the gushing posts about how AI is great. I use all the frontier models and sure they're a bit helpful

But I don't understand if they're that good why are we getting an outage every other week? AWS had an outage unsolved for about 9+ hrs!

Correct me if I am wrong. But Claude has probably been trained on gcc, so why oh why doesn't it one shot a faster and better compiler?

I once had a PR. I told the dev that "LLM is ok but you own the code"

He told me "I spent n days to architect the solution"

He shows me claude generated system design .. and then i say ok, I went to review the code. 1hr later i asked why did you repeat the code all over at the end. Dude replies "junk the entire PR it's AI generated"

Has anyone who's familiar with compiler source code tried to compare it to other compilers? Given that LLMs have been trained on data sets that include the source code for numerous C compilers, is this just (say) pcc extruded in Rust form?