Video wird geladen...
Video konnte nicht geladen werden
Jev Engineering transforms a collection of AI agents into a full-fledged management system while eliminating heavy, expensive models from every decision-making loop Speed: +19,000% Costs: -99.7% How this architecture works: Gatekeeper Request Mini-Model Validation Tool Router State Worker Management layers avoid generating detailed text and instead focus on directing,... show more
13,950 Aufrufe • vor 1 Monat •via X (Twitter)
48 Kommentare

this could make multi-agent systems way more efficient

Yes, this increases efficiency manifold

the speed is absolutely insane

It really is a rocket

In general, I’ve been thinking for a long time that it’s time to stop riveting models, it’s time to make infrastructure for them

Yes, infrastructure is far more important

jev is really cooking

A real alpha right now

@Bober_smart interesting setup! one thing i'd watch for is over-relying on the mini-model for decisions. it can sometimes bottleneck if it handles too many complex tasks at once.

Clean terminal setup with real style points

Thanks, I put a lot of effort into it

jev reduces costs by 99.7%?? So, it's practically free?

Yes, it really costs next to nothing

Cool

jev meta in progress

True men

bro, is jev cooking?

Yeah, buddy, this is a new era

this looks like a logical evolution of agent systems

That is actually the case

appreciate the setup guide still on it

Good luck with the results

I’d love to see what this looks like at scale

I'm sure this will impress you

Heavy models only when needed makes sense

Yes, this significantly reduces costs

99.7% cost reduction is crazy but the design is what caught my attention

It really looks attractive

This speed is beyond human limits

It really is incredibly fast

the managament layer is really smart

This shi looks clean

Thanks for sharing this

I've a question buddy i did you generate/make this video 🥹

this is a really clean AI architecture

Is JEV a full-fledged LLM?

The biggest win is making the management layer measurable

I bookmarked this guide made. Its very useful for everyone AI enjoyer

how does the mini-model fit into the overall decision-making process?

the gatekeeper layer is the part i’d instrument first. log every route decision with the token spend it avoided, then you can see which rules are actually earning their keep.

把复杂决策拆解给轻量模型才是降本增效的正道

this is lowkey relaxing

Routing tasks through a lightweight gatekeeper changes everything about running agents at scale.

So real

いいね

This is a strong example of AI agents moving toward complete decision-making loops

how I can make it too

nice breakdown

