Jack Smith

Writer. Husband. GTM @withcascadia

[Jack Smith @jackwsmth]

https://t.co/rLi2NmjdCd

[Jack Smith @jackwsmth]

@JasonGlynnLive can't argue with that, just odd that it manages to ignore its own settings to do so

[Jack Smith @jackwsmth]

Claude co-authored a bunch of commits even though the setting was disabled... hmm 😅

[SergeB @SergiiioBS]

Qwen3.8 Flash Next IQ3_S on 2x B70 on XeStrata.

Keep all your results coming!

https://t.co/s8A8K3KJwK

[Jack Smith @jackwsmth]

@happysmash27 @nishnishnishh @JustMicrock I think it's great, but the UI could use some improvements imo

[Jack Smith @jackwsmth]

Some thoughts on why AI may create specialists instead of generalists:

https://t.co/mRjEVB9d5U

[Max Spero @max_spero_]

it should tell you something that the most bleeding edge ai-pilled technologists are not letting AI write for them

[Jack Smith @jackwsmth]

@inkdrop_app me three :)

[Jack Smith @jackwsmth]

@aramh I'm kind of not surprised actually, seems like a lot of their team is led by people who are big into CS theory etc

[Jack Smith @jackwsmth]

@AdamHoltererer What about @NousResearch?

[Jack Smith @jackwsmth]

alex is a great writer.

going to be trying out my own hermes workflow soon 😎

[alex @afmedia_]

Computing is a deeply spiritual practice

[alex @afmedia_]

Computing is a deeply spiritual practice

[Jack Smith @jackwsmth]

big if this works.

would love to see some more OSS video editing software, too!

[Eric S. Raymond @esrtweet]

This is the doom I predicted a few days ago, coming for Photoshop. A clean-room open-source reimplementation.

No prizes for guessing that they decompiled Photoshop to source code, processed that to some kind of non-code specification language, then fed the spec to an LLM with an instruction to generate Rust.

Adobe just got nuked. And closed source is dead, dead, dead.

https://t.co/0sxtGNaPtq

[Tate @TateBerenbaum]

We ran @thinkymachines’ Inkling 975B across eleven Intel Core Ultra PCs using @withcascadia.

Peak aggregate throughput: 60.29 tok/s at 88 concurrent streams. Single stream clocked 5-10 tok/s.

Almost a trillion parameters on consumer hardware :)

[Jack Smith @jackwsmth]

975B open-weight model served on-prem, across commodity hardware.

[Cascadia @withcascadia]

Cascadia is now serving Inkling (975B) from @thinkymachines across eleven Intel Core Ultra PCs, in a deployment with @IntelBusiness.

Quantized, Inkling sits at ~475GB in size, which we split into shards and distributed across the eleven machines. Each one runs a custom MoE engine built on top of OpenVINO.

Peak aggregate throughput reaches 60.29 tok/s at 88 concurrent streams.

[Jack Smith @jackwsmth]

I'm starting to think AI won't make you a generalist, but will make you more of a specialist

You can use AI to do all of the busywork surrounding your domain, and focus everything on your own specialization

[Jack Smith @jackwsmth]

@hope_rythmn @Apple I've been saying for so long that Apple's software experience is degrading :(

[Jack Smith @jackwsmth]

People blame technology for many problems, when in reality the problems are created by humans.

Greed, lack of empathy, etc. These are the things that make software (among other things) worse

[Jack Smith @jackwsmth]

Constantly amazed at the work being done on @withcascadia

[Jack Smith @jackwsmth]

Codex limits seem way worse than Claude (anecdotal)

[Jack Smith @jackwsmth]

@nic_carter I'd argue this is less to do with prompting and more to do with domain knowledge