Jev Engineering ran 5,400 agent executions and sent exactly 0 decisions back to me. that’s the number i care about here. and up to 193x faster and 444x cheaper in tests. 18 turns → 300 agents → 5,400 runs → 43 Jev calls → 16 outputs kept → 0... human escalations the loop handled three decisions around every turn: Choice → Score → Noul then it stopped itself when the goal crossed 0.93 vs 0.85 required. that’s what Jev Engineering looks like when it works: not “more autonomous agents.” fewer reasons for the human to come back into the loop.show more

Actually crazy bro

useful for you?

yeaah g, thanks!

This should reach 1m

i hope it will

this is what autonomy should actually look like

wow who's the pink one in the middle?

Looks wild from first second

try my best

damn, thanks a lot of agents

这才是自动化该有的样子省心太多了

5400次跑完0次回人工 这个数比那个193x有说服力 说白了前面把筛选和验收卡死 后面才敢让它自己跑 不然省下的时间全花在收拾烂摊子上

0 escalations but 16 kept, whats the eviction criteria for the other 5384?

5,400 runs and 16 outputs kept is a 0.3% keep rate. that's either a very picky scorer or a lot of wasted compute

0 escalations in 5,400 runs reads clean until you ask what the escalation threshold was. what triggered a human send vs skip, and what % of runs got flagged. that's the number that decides if the loop is safe or just quiet

bro, this is both terrifying and exciting at the same time

5,400 executions and zero decisions coming back to you is the whole point. Escalation count is the real KPI — nobody puts it on a slide because it's always embarrassing.

0 human escalations这个数我最不信…验证环节去哪了。193x/444x这种测试数字看看就行，真跑起来瓶颈从来不在调用速度上

Interesting perspective

running 5,400 executions with 0 decisions shows real trust in the system

This design idea is really good, efficient and cuts a lot of redundant costs.

really appreciate you taking the time to write this out, it crystallized something for me

This is so true

zero human escalations with 5,400 runs is impressive, what’s next for scaling?

It all looks insane