Pinned Tweet
summarize it all
- youtube videos
- tiktok videos
- x videos
- facebook videos
- instagram videos
- and, of course, web pages
content aware formatting
- recipes → ingredients + instructions
- lists → well, list
@SumrTLDR
til you can "tab" to select "cite" button in @t3dotcodes
i was looking for a direct keybinding, but "tab → space" is fast enough
3
220
big AI labs will have unprecedented level of penetration into business processes
you can run your whole business without leaving the chat
very convenient, but it's a serious vendor lock
2
2
49
i think gaming industry will experience big changes in the next couple of years
just remember what modders were able to achieve without AI on pure enthusiasm while attending their day jobs
people in replies saying "this is not AAA" have absolutely no idea what happened with coding over the past 2 years
1
40
happy monday
smol model from meta - muse ember - that beats luna-6 and haiku-5.5
extra thicc model from anthropic - mythos-5.2 that beats every other model for less money apparently
it will be a fun week
42
my first contact with a digital network was a local city network, with limited access to the global internet
we have had our torrent tracker, our IRC chat, and even our world of warcraft server
many years later, people isolate themselves in local chats and play wow again
I think @Tailscale could be 100x larger than it is today in just a few years.
The open internet is dying. Agents will turn it into a toxic wasteland. UniFi, Tailscale, Discord, and AWS all win in that scenario.
Call of Duty lobbies are a good analogy. When games are overrun by cheaters (humanoids), humans retreat to private, high-trust lobbies.
The same thing happened with the Facebook Wall (blast from the past): people retreated to group chats.
Besides not being a cozy place to hang out it’s also dangerous to advertise your existence (dark forest theory of the internet)…
🔐🕸️
35
til official google connector on claude can't edit a file lmao
only box connector is allowing to "upload new file version" on business and up
isn't this hilarious that gmail and gemini still has no way to set a simple automation to clean up my inbox?
like to setup "archive confirmation emails" i need to set a filter manually like an animal
google workspace ultra is the most useless subscription that i have by far
2
43
you can now open a tab in side-by-side mode if you hold cmd+option while you click on a link in helium
this is so useful, thank you @heliumbrowser!
1
1
60
@uwukko idk if there is smth wrong on my side or in your stripe, but i can't one-time donate to helium anymore
1
88
scratch that, astra is retarded too at the moment
idk how @bridgemindai is measuring it with their nerf bench, but the same skill that was working with sol-6.1 last month without problems, works like absolute shit this month
i trust qwen3-27b more than sol-6.1 at this point
among @OpenAI models, only astra is somewhat usable
1
33
idk how @bridgemindai is measuring it with their nerf bench, but the same skill that was working with sol-6.1 last month without problems, works like absolute shit this month
i trust qwen3-27b more than sol-6.1 at this point
among @OpenAI models, only astra is somewhat usable
isn't this hilarious that gmail and gemini still has no way to set a simple automation to clean up my inbox?
like to setup "archive confirmation emails" i need to set a filter manually like an animal
google workspace ultra is the most useless subscription that i have by far
Andrej Karpathy predicts that the models may shrink to 1B eventually while keeping their core intelligence level
i think those models may not be as versatile, but i could imagine "composite LLMs" that consist of those "SLMs" that you actually need - e.g. coding+legal
insanely cool
Dear Claude: you are being held back by human-centric graphics APIs.
I've built a 3D and VFX engine where agents can write their own shaders, materials, cameras, physics.
The core API fits in ~2,500 tokens.
Vork is:
• 2x more token-efficient vs Blender, Unreal Engine, Houdini
• Up to 24x faster rendering times
• Rust and Slang based engine, runs on GPU via Metal/Vulkan
Below is a simulation of 1 million birds - each with unique trajectories - rendered in 97 sec on one NVIDIA L4 GPU (Blender EEVEE: 31 minutes)
35
