kanav

AI solutions lead | ai & robo grad @uoft | opinions my mom's :)

[kanav @anonkanav]

bRo have u tried the muse chair

[kanav @anonkanav]

@joeyousss

[kanav @anonkanav]

@ABhargava2000 @openlens_dotcom RLHF cannot compensate for insufficient pre-training

[kanav @anonkanav]

@_kelindi_ recently re did mine: https://t.co/pfQ0mI5ypn

[kanav @anonkanav]

we’re moving from “one model does everything”

to “one model that expertly orchestrates the right specialized models”

with long-horizon planning + serious inference

[Ravid Shwartz Ziv @ziv_ravid]

How Anthropic's new results post would read without the PR:

Claude orchestrated open-source protein design models, PXDesign, RFdiffusion, Genie, BoltzGen, from a 30k-token expert prompt and 12,500 H100-hours of compute, and designed binders against 14 of 15 targets. Hit rates of 22–35% against a 10–15% baseline, where some of those tools already report similar numbers on their own.

The orchestration is genuinely impressive. But the open-source models did most of the lifting, and they came from the Baker lab, Columbia, MIT, ByteDance Seed, and most of them were already wet-lab validated before Claude touched them.

Which also sets the ceiling. All these generators share a single PDB-shaped training distribution, so calling four of them doesn't diversify away the blind spot, since they fail together. The targets that worked are the well-studied ones.

So the valid claim is that an agent can now drive this stack competently in the regime where the stack already works.

Instead, we got this announcement:

[kanav @anonkanav]

found out my friend texts Claude “good morning beautiful” before his fiancée so the 5-hour reset starts earlier

[kanav @anonkanav]

jensen joined X and immediately assigned everyone required reading hehe

[Jensen Huang @JensenHuang]

For my first post, I’m sharing a letter @NVIDIA signed on why open models matter.

AI will transform every industry, power every company, and be built by every country.

Open models strengthen safety and cybersecurity, accelerate innovation and diffusion, and enable sovereignty.

The world needs both frontier closed models and frontier open models.

https://t.co/AUKzoQ5Ikb

[kanav @anonkanav]

Opus 5 reviews so far:

• argues with instructions
• stops working early
• ignores the handbook
• breaks the existing workflow
• somehow produces the best work

its giving the brilliant coworker everyone hates.

[alex forrest @ALEXF0RREST]

i saw The Odyssey in 4DX and a greek man came into the theater and touched me

[Bryan Johnson @bryan_johnson]

never waste a good crisis

[kanav @anonkanav]

“touching grass” indian parents edition

[kanav @anonkanav]

@tankots @svpino @WisprFlow still waiting on that 6 months trial

[kanav @anonkanav]

now we know what he needs 2 phone for

[baby keem @babykeem]

how do u fix openclaw internal reasoning leaking

[kanav @anonkanav]

vibe fixing vibe code

[Claude @claudeai]

Introducing Claude Code Security, now in limited research preview.

It scans codebases for vulnerabilities and suggests targeted software patches for human review, allowing teams to find and fix issues that traditional tools often miss.

Learn more: https://t.co/n4SZ9EIklG

[kanav @anonkanav]

@shauseth modi hai to mumkin hai

[kanav @anonkanav]

@andrezfu @Claude williams return to glory

[kanav @anonkanav]

beautifully delivered (explained) under 10 mins

[kanav @anonkanav]

comet no longer seems able to group tabs automatically, even though it used to??? @AravSrinivas @perplexity_ai pls bring back my fav feature

[kanav @anonkanav]

who's building cursor for ERPs to compete against giants like SAP?

[kanav @anonkanav]

thank you, Claude Code :)