Yohei Nakajima

remember, remember, the sixth of November…

RT POLITICO Silicon Valley plots Jensen Huang’s comeback https://politi.co/4iU3r8Q

RT Raymmar going live with @yoheinakajima next Tuesday for a deep dive to discuss agents on the frontier.

RT Shengkun Claude + Monid = 8M+ impressions 🤯 A Monid user let an agent run his entire SEO loop. Six months in: 49K clicks from search, about 100K impressions a day. No SEO hire. No agency. Every prompt is in the guide ↓

RT Swann Bessa Introducing e-means, an algorithm beating k-means in 10x fewer epochs. With order of magnitude wall-clock speedups, e-means is released in FAISS and needs no hyperparameter tuning beyond k. It is robust to various scales and datasets. Still skeptical ? Just try e-means [1/10]🧵

RT andrew pignanelli New Blog Post: The Era Of Fully Autonomous Companies. Fully autonomous companies are now possible. By this time next year we will see millions of them, and many will be profitable. In early December, we'll show you some. https://www.generalintelligencecompany.com/writing/the-era-of-fully-autonomous-companies

got this working with Qwen3.5-4B this weekend better reasoning solved most of the unsolved tasks from the 2B test. so yeah, we can compile agent loops into model weights i guess next: harder unbounded tasks? even bigger models?

i’m a bit gentler w my tone like “i don’t think this is it man”, but +1 on this

RT Marko Kraemer Introducing the open-source @kortix CompanyOS: Your entire company as code: agents, skills, connectors and rules in one repo you own. A workspace for your team, a framework for your agents. Sovereign, multiplayer, extensible AI.

my most recent ArXiv paper took 68 days from submission to acceptance, which is the longest it's taken (Aug 28 -> Nov 5) my soft-to-rigid only took 4 days, but i'm guessing because it was routed to soft condensed matter [cond-mat.soft] which has less submissions (Oct 9 -> Oct 13) can't blame them though with this uptick in research https://x.com/rizbicki/status/2117594875449487643

evil qwen, weight-space programming, and qwenputer went viral in a 10–day window

RT Thomas Wolf 18 of the 20 best open models today are Mixture-of-Experts. Dense is basically over. And yet the know-how to train a good MoE still lives mostly in the heads of a handful of teams. So we spent 800k GPU hours running the experiments ourselves and wrote everything down. Super happy to finally release "The Ultimate MoE Training Handbook", free and open, organized around the questions training teams hit: - how many experts, how big, shared or not? - why is half my step spent in all-to-all, and how do I hide it? - why is my run 25x slower with no error in the logs? - softmax, sigmoid or sqrt(softplus) router, and why keep it in fp32? - aux loss or DeepSeek's loss-free bias? - upcycle a dense checkpoint or train from scratch? - can I serve with fewer experts per token than I trained with? ⚠️ Warning: it's long (~70k words, budget a few days, sorry) but with 55 interactive figures, code you can run and numbers from our own runs. Read it here: https://huggingface.co/spaces/huggingface/moe-blogpost

an agent loop compiled into model weights

RT Advaith Sridhar Today, we announce five Grand Challenges in Material Science. We’re also committing $1M in prize money to anyone who can solve these challenges. AI can solve Millennium Prize problems in math. Can it do the same in material science? To find out, we reached out to 10 leading professors to identify the hardest open problems in the field.

RT Vasek Mlejnsky We hit an annualized run rate of $120M in October. This year alone, we’re projected to add around $150M in net new revenue. This is an amazing milestone and a testament to the work of the @e2b team. I’m particularly excited about two things: 1/ This showcases how strong the tailwinds of AGL (Agent-Led Growth) can be. If you build a tool for agents and optimize your whole growth engine to serve them, you can see tremendous growth. This isn’t something you can just throw money at and buy. Agents don’t decide based on ads. You need a sophisticated growth engine that fine-tunes things like your AEO. I think this is what all future B2B devtool sales will look like. Increasingly, decisions about tool adoption will be made by agents, not humans. 2/ The use cases we’re serving. The majority of our revenue and usage comes from agents running in production. Not RL. When we started, we had two theses: (i) code is all you need—all agents will use code to complete tasks (which has turned out to be true); and (ii) sandboxes for agentic workloads will be a bigger market than sandboxes for RL workloads, but takeoff will be slower (still playing out). In the past 30 years, every company has become a software company. In the next 10 years, every company will have become an agent company. We’re building an infrastructure layer to power all these agents.

helping banks adopt agents safely

siri ai tool call seems to be getting an upgrade and looks like it’ll handle multi step workflows too

okay that's confusing but i guess makes sense that https://console.x.com is for x api and https://console.x.ai is for grok api but it's still confusing

RT AI.Engineer 🔜 Shanghai We're excited to announce our AI Engineer Code Summit speaker lineup! → https://ai.engineer/code VP Gemini Post-Training Lead, @GoogleDeepMind CEO, @e2b CTO, @ExaAILabs CEO, @warpdotdev CEO, @aviator_co_ CEO, @fivetran CTO, @Lovable CTO, @iamovercut CTO, @earendilworks (Pi) CEO, @diffusionio CPO, @cognition CTO, @GoodfireAI MTS, @AnthropicAI CTO, @composio GPM, @Cloudflare MTS, @databricks CEO, @AmpCode Co-Founder, @gimletlabs CEO, @mixedbreadai CEO, @humanlayer_dev Head of AI, @MonacoGTM Lecturer, @UCBerkeley Principal Engineer, @SpaceXSI CEO, @TypeSafeAI (Jev) Founder, Bun at Anthropic VP Product, @reflection_ai Research Lead, Reflection AI Staff Software Engineer, @SonarSource Principal Software Engineer, @tryramp Senior Engineering Manager, @netflix Forward Deployed Engineer, @baseten Also joining us: @kevinhou22 — Google DeepMind Antigravity @Steve_Yegge — Creator of Beads & Gas Town @lucasmeijer — http://AgentsInTheCloud.com @kentcdodds — Software Engineer & Educator @trq212 — Independent @simonw — Django, Pelicans, Lethal Trifecta Presented by Google DeepMind. November 10–12 in San Francisco. Attendance is by application. → https://ai.engineer/code

looks like we’re about to start benchmarking loyalty

happy 20th galactic birthday, earth! 🌎 🎂 🎉

i love how fast viral projects can trigger new projects into existence

get your tool on monid

i just caught myself referring to activegraph as “our project” to codex

RT Endowment Eddie It’s hard out there for unicorns. Just 14% of the universe is growing >60%. Very few of these businesses meet the bar to go public or can justify their marks. With 57% growing <20% and cash runway appearing limited, the power law will play through.