𝗿𝗮𝗺𝗮𝗸𝗿𝘂𝘀𝗵𝗻𝗮— 𝗲/𝗮𝗰𝗰

@elonmusk If you want to understand the algo, I have generated one for you. Check here.

The latest 𝕏 algorithm has been published to GitHub github.com/xai-org/x-algo…

The latest X algorithm has been pushed to github. Every like, reply, watch time, repost, mute, and report becomes a signal inside a massive ranking system deciding what gets pushed and what disappears. X is no longer just rewarding clicks. It’s prioritizing replies, long-video content, trusted social graphs, and topic relevance. brand-safe media while aggressively filtering spam and unsafe content. What do you think, guys?

@_jaydeepkarale They are cutting down vloggers mostly.

How much had they over hired ovee the years ???

Watch the full video here: youtu.be/4QXtObc61Lw?si…

Jeff Crume explains the critical roles of security & AI governance in protecting systems from vulnerabilities and ethical lapses. Learn how to safeguard AI systems with best practices in governance, compliance, and risk management. Read the blog quoted about the agent governance layers.

@yyz81681981 Thanks for your feedback.

@techwith_ram 这篇好 把治理拆成五层递进的结构讲清楚了，不是泛泛说"要加 guardrail" 但有个地方我想补一句 他五层都是"人的设计"——intent、permission、audit、escalation、feedback，全是人在定义规则然后让 agent 遵守 漏了一个方向：agent 自己学的行为怎么治理？

@_jaydeepkarale Great share, Jaydeep.

I was working on Newsletter automation when I found this Github Repo Claude's training cutoff is months old. When you ask it to write a tweet or a Reddit reply on a topic, it has no information about what people are actually saying on that topic right now. The output usually invents examples and references trends that do not exist. notslop fetches that missing context. Fresh social context for AI agents. Reddit, Hacker News, blogs and X, reranked and piped into Claude Code via 19 skills.

We spent the last 2 years asking: “Which AI model is the best?” Turns out, we were asking the wrong question. The same model can perform wildly differently depending on the system wrapped around it. Same brain. Different behavior. That’s why some AI agents feel incredibly smart in demos… & completely fall apart in production. This video explains the hidden layer nobody talks about enough: Harness Engineering, the memory, tools, orchestration, and control logic shaping how agents actually behave. Also check out my blog attached as well.

Full video here: youtu.be/A0xu44a1BHE?si…

Unacademy co-founder talking about how "Founders are not innovative." 🫢

A 10M document corpus eats 31 GB of RAM as float32 Most teams hit that wall & reach for a managed vector database. $400 a month or $800 a month. Scaling with every document you add. Then a solo engineer shipped turbovec turbovec is the open-source Rust vector index built on Google Research's TurboQuant algorithm. It fits that same 10M corpus in 4 GB, & searches it faster than FAISS. Here is what makes it different: → Zero training. Zero data passes → 16x compression on 1536-dim vectors → Beats FAISS IndexPQFastScan by 12–20% on ARM → NEON (ARM) + AVX-512BW (x86) SIMD kernels baked in → IdMapIndex for stable IDs that survive deletes → Python bindings, LangChain, LlamaIndex, and Haystack integrations → Fully local. No managed service. Rust core. Python bindings on PyPI. Built on Google Research's TurboQuant. SIMD-optimized for both ARM and x86. Integrations for every major RAG framework.

@nghoihin Great. Thanks for sharing here.

@techwith_ram Team chat is a wiki waiting to happen. We open-sourced the runnable version: Beever Atlas. github.com/Beever-AI/beev…

Github Repo here: github.com/RyanCodrai/tur…

@davetweetlive Exactly. And with money he has why as a founder he is not going for something innovative now?

@techwith_ram Funds release karo then talk shit. Innovation or implementation of a new idea requires money.

@ThedatagGuy Haha. Their whole business model is copy paste.

@techwith_ram 😭 irony Bro made teaching platform and bitching about innovation