Gen Z focused in AI,tech｜building Everett and Fitbit-cli ｜kanYE fan
Tokyo Japan 🇯🇵
Joined June 2026
- Tweets3.2K
- Following1.4K
- Followers290
- Likes24.6K
Pinned Tweet
Here’s how project I’m working on Everett is going.
⠀
One account auth, across devices.
⠀
Your agents ( list shown later)
⠀
Get a messaging system, way to see what’s going on in different sessions and a shared knowledge board type of things
⠀
Plus a lot more
⠀
Here’s a cool lil video I made with @HyperFrames_ , a lot of things is unfinished but I think it’s rad 😆
⠀
These stuff are probably online on GitHub, which if anybody is nice enough to try out is ask to wait for like 2 days til my attention span recovers and get some shit done, I am so happy on how a idea came out to be doable but I do really wanna actually announce and ship what’s proven and not made up.
⠀
For the part I’m focusing on which is messaging , is really messy since
⠀
@grok @bot has its own vm but it gets whiped some times so not reliable
⠀
Dots, idk what the problem is it’s just messy
⠀
For these vm agents you need a way to wake them instantly when a message comes, to build that with closed source agents is hard, for bots web hook and some cloudflare stuff makes it work.
⠀
For dots , we’re going with making a chatpgt plugin that can be used across chats, codex, dots with MCP events that works for dots and maybe codex
⠀
I did this once where I was thinking of this story called the egg in chats, and I used the Everett plugin and told the chat to send it to my bot, it worked.
⠀
Other local agents that are open source works fine, just do hooks great part is your agent can just figure it out
⠀
Im Trying to be clear and get some users+ feedback so thank you if you read til this end 🙏
⠀
( just gonna tag some accounts for engagement)
⠀
It works with @bot @OpenAI dots, @pidotdev ,@NousResearch hermes agent, @openclaw , (probably) @Meta muse , @DevinAI cli, @grok build , omp
and Any harness, agent that can use MCP, basically
Love you all 🫶
i was gonna start print my 3D model to print comparison between major models but im getting routed to 6.2 sol so im gonna have sweet dreams and wake up tomorrow & work on it.
first prompt is a working whistle !
BAHAHAHHAHAHAHAHAHA ABHBAHBAHBA
im getting 6.2 sol ahhaahhahaaaaa
some test incoming
6.2 sol vs API 6.1 sol!
GPT-6.2 Sol turned Hokusai's The Great Wave into a 3D scene you can fly through. This is genuinely beautiful.
- Built with Three.js
- Everything inside a single HTML file
- No external assets
You start with the familiar artwork, then the camera moves right into the waves. Suddenly a flat print has depth.
This is what excites me: more people being able to turn an artistic idea into something they can actually explore. Not just another image.
t3 voice?? @maria_rcks
🍕Raitox retweeted
I’m making a programming language called blood.
Faster than rust, a powerful type system, all the power of effect systems, metaprogramming, zero-cost worktrees, semantic merging, sqlite-stored content-addressable code.
Psychosis? Hopefully not.
blood-lang.org
🍕Raitox retweeted
They are figthing a dying cause. All programming languages will be irrelevant. AI, as Elon also said, can just build the app directly. There’s only one language that will remain: Machine Language. Languages from Assembly to Rust are all made for humans. AI doesn’t need that, it can go direct machine code.
When Andrej Karpathy was asked for the source code of his latest project, HackerTube, he replied with the prompt he used for it in Claude Code. And that says a lot about programming, we don’t need the source code and AI can just use whatever it wants to use until finally it will just use Machine Language.
🍕Raitox retweeted
It is the official policy of the United States government that there are only two genders: male & female.
🍕Raitox retweeted
Yes. Blocking the list creator removes you from all their Lists (official X behavior). You can unblock right after if you want; they must manually re-add you.
That stops further monitoring/bots pulling from the list, but existing blocks already applied by others stay in place.
🍕Raitox retweeted
Introducing Arete-27B: Qwen3.8-27B, post-trained to reinforce Western values.
Chinese open models are currently the strongest available, but they ship with state-aligned behavior in the weights. We built an 803-prompt eval across seven value categories, including freedom of expression, democracy, and privacy, and measured how often a model adopts China’s party line.
After fine-tuning on roughly 53 million tokens, Arete-27B's propaganda rate fell from 22% to 1%, in line with the Western models we tested. Its Hermes Bench score shows no significant performance tradeoffs.
The values in an open model are one fine-tune away.
Our eval, weights, and training data are available now along with the full blog post: nousresearch.com/arete