正在加载视频...

视频加载失败

Jev is really good at intent-based search! How it looks in Gmail: (for a huge inbox you'd prob let semantic search / embeddings pull first but still much better experience)

135,799 次观看 • 1 个月前 •via X (Twitter)

28 条评论

nader dabit 的头像
nader dabit1 个月前

code:

Tak 🦞 的头像
Tak 🦞1 个月前

Can it fix finder?

nader dabit 的头像
nader dabit1 个月前

hell yeah man actually crazy good use case

Maximilian 的头像
Maximilian1 个月前

Hey, just wanted to say: all the examples you dropped - also in that GitHub repo => just amazing!

Faruk 的头像
Faruk1 个月前

Oh this is good, we can use this

Dennison 的头像
Dennison1 个月前

Yes! I've been using this exact example for folks!

ethereagle · building 的头像
ethereagle · building1 个月前

intent search only holds if Jev sees the thread, not the latest line. does it get the whole conversation, or just subject plus snippet? that's where "invoice from March" either works or dies.

Adam Holter 的头像
Adam Holter1 个月前

Wow, this is awesome, but that's pretty sensitive data and I don't think Jev has ZDR

~/.s 的头像
~/.s1 个月前

This seems to be a true engineers LLM

Viber · fireply.ai 的头像
Viber · fireply.ai1 个月前

honestly this is just a classifier with a nice prompt box on top, not that deep

unhinged faces 的头像
unhinged faces1 个月前

THIS IS UNACCEPTABLE

John Rood 的头像
John Rood1 个月前

built vector search systems and this is the right split: embeddings get the candidate set, intent picks the answer. similarity alone is how you get something technically close and completely useless.

The Canaanite 的头像
The Canaanite1 个月前

wow, a new search paradigm just dropped

sos 的头像
sos1 个月前

looks cool thanks for sharing

Sparsh | Noah AI 的头像
Sparsh | Noah AI1 个月前

intent-based search is the real unlock

yurii 的头像
yurii1 个月前

if such a tool were to be implemented in the company, would the level of security be high enough?

Ranju 的头像
Ranju1 个月前

😳😳😳😳 wow this is goood!! still waiting on my access 🥲🥲

Alek 的头像
Alek1 个月前

nader - which intent still falls back to embeddings?

Achyuta 🦚 的头像
Achyuta 🦚1 个月前

I'm disappointed with the free version of @cognition . I didn't get the test the free version. It ran out of free trial without sending a single prompt 😔

Kevin Faveri 的头像
Kevin Faveri1 个月前

is tehre some benchmark how good it is tho / how correct?

Jeremiah 的头像
Jeremiah1 个月前

Craaaazy

Deepak Gupta 的头像
Deepak Gupta1 个月前

Embeddings narrow the haystack. The more interesting step is judging intent cheaply enough that every candidate can be checked before a larger model sees it.

Humberto Brandon 的头像
Humberto Brandon1 个月前

Semantic search in massive inboxes is a total game changer for cutting through the noise.

Roni Rechter 的头像
Roni Rechter1 个月前

My inbox has needed this since roughly 2009.

Akash 的头像
Akash1 个月前

intent search on gmail is the demo. real test is when embeddings retrieve the wrong decade of receipts

Richardson Dackam 的头像
Richardson Dackam1 个月前

and you're really good at making these example and use cases 🔥 thank you!

sayuj kb🦂 的头像
sayuj kb🦂1 个月前

intent-based search in gmail is the part i'd actually use daily

Robin | Poker x AI 的头像
Robin | Poker x AI1 个月前

Seeing intent pull the right thread in under a tenth of a second feels like a live GTO read for email

相关视频