←back to thread

324 points bilsbie | 4 comments | | HN request time: 0.238s | source
Show context
dcastonguay ◴[] No.44974574[source]
> At the end of it, they were sketching a completely different architecture without my "PMing". Because they finally understood who was actually using our product.

I cannot help but read this whole experience as: “We forced an engineer to take sales calls and we found out that the issue was that our PMs are doing a terrible job communicating between customer and engineering, and our DevOps engineer is more capable/actionable at turning customer needs into working solutions.”

replies(41): >>44974602 #>>44974635 #>>44974655 #>>44974660 #>>44974676 #>>44974814 #>>44974873 #>>44975042 #>>44975156 #>>44975182 #>>44975196 #>>44975269 #>>44975293 #>>44975666 #>>44975685 #>>44975856 #>>44975925 #>>44975972 #>>44976091 #>>44976207 #>>44976426 #>>44976440 #>>44976835 #>>44976924 #>>44977035 #>>44977052 #>>44977553 #>>44978517 #>>44978620 #>>44978689 #>>44979587 #>>44979694 #>>44979713 #>>44980051 #>>44980093 #>>44980149 #>>44980874 #>>44981249 #>>44981402 #>>44982096 #>>44982636 #
1. gedy ◴[] No.44975293[source]
I wonder if LLMs might be replacing these type of PM jobs where they gather up feedback (usually it's mostly in text form anyways), and translate and summarize so engineers can cut out some noise and confusion from PMs.
replies(2): >>44975515 #>>44976321 #
2. deepsun ◴[] No.44975515[source]
Ok, LLM translated and summarized. Then what?

Someone needs to look at it and push important points. Sometimes it's hard to push engineers, until they visit some calls and push themselves.

replies(1): >>44976175 #
3. gedy ◴[] No.44976175[source]
Sure, I know there's companies like that, but just as often in my experience engineers are spoon fed tickets without broader context. In many cases are also treated like an interruption if you want to discuss to root issues etc with PMs
4. zamadatix ◴[] No.44976321[source]
I'd say it's about as likely as LLMs actually replacing the engineers in implementing the code in the next couple of years. I think it's more likely LLMs end up being like every other tech advancement: a way to increase the total amount of stuff being done, but not actually lower the need for people to use them.

Or maybe the next thing after LLMs arrives in 2026 and it's actually better than everyone at everything and can feed itself in a loop, but I doubt it.