Computer use

unironically if those engineers were considered to be 'bloat' its better to have them label data because they are smarter and vetted

basically a soft layoff

"It’s literally the gulag" - okay this was a funny comment

its unclear to me why they need their model to be the best at coding (maybe to build an internal technical moat?)

Because coding appears to be the most monetizable use of AI, for some time to come.

It's also not like Meta could train it for content moderation, translation or judicial expertise even if it wanted to. They have software engineers idling, not translators.

But Meta's had so much success in this area by copying the leading lans already!

but who’s going to pay Meta for it?

??? The same market that is paying for Codex or Claude Code, if Meta offers competitive services.

I find it pretty hard to see how Meta could possibly catch up to the existing frontier labs. Distillation, I guess, if you want to go that route, but between the fact that they're already way behind and the constant deluge of negative press like this, it strikes me that they would straight up need an architectural miracle in order to actually be competitive.

its fantasy

scale ai's value prop was catching people like this

It’s just advertisement/SEO. It’s basically guaranteed, just not the way you think

what about just... becoming mediocre? engineers are already infamously lazy at reviewing PRs. how is Meta incentivizing these Data Labelers to give a shit and actually scrutinize the AI-generated code they're supposed to be reviewing? what's the reward structure? what prevents engineers from flagging minor nitpicks all day while they look at LinkedIn?

Probably forcing them to review each other's work to panopticon "quality," and keeping track of the average throughput per engineer so if people fall behind the taskmasters can pay them a visit.

dont think the value of btc comes from its utility anymore

stables are much better for transactions anyway

its for the distribution, talent, proprietary data. spacex has the compute, and elon wants to win in AI.

its very hard for most businesses, especially large ones, to build good agents (not the kind that does rag on a faq) that complete actions autonomously and cannot be jailbroken

demand for ai support vendors is going vertical this year

+1

its surprising to HN tech crowd but a well implemented support agent gets higher reviews and successful resolutions than humans

Can you cite a source or study for your assertion?

personal info

maybe will be public soon with earnings calls? We’ll see!

yes, it counts as unresolved

c'mon listen to yourself

the bulk of the context engineering for users of these ai support platforms is done in the platform

and the amount of context needed to automate f500 is non trivial, plus you usually cant use reasoning because latency would blow up and you get escalated on

if this was so easy as you claim theres many millions for you to be made selling it to enterprises, but you wont

F500 is exactly the kind of scale where I fully expect support agents to be developed in house. They'll try Fin. Then one day, a single dev inside the company demos a custom agent that outperforms Fin and cost almost nothing.

tech forward f500 is possible (but.. even anthropic used fin)

most f500s are going to outsource. i think sierra is already at 40% of the f50? and each of those deals they have to compete against teams of inhouse engineers and convince execs to buy instead of build

the reality is agents at scale is a hard problem and most f500 engineers are not equipped to solve it

fwiw i heard Fin and Agentforce frequently lost to deca and sierra, their product / team / funding weren't as good

Seems like a catchup play by Marc