Discover real AI creators shaping the future. Track their latest blogs, X posts, YouTube videos, WeChat Official Account posts, and GitHub commits — all in one place.
leworldmodel
JEPA are finally easy to train end-to-end without any tricks! Excited to introduce LeWorldModel: a stable, end-to-end JEPA that learns world models directly from pixels, no heuristics. 15M params, 1 GPU, and full planning <1 second. 📑: http://le-wm.github.io
View quoted postthis @latentspacepod is getting lots of love
Must watch for anyone in love with Claude CoWork. Killer pod from the guys @latentspacepod and @felixrieseberg
No, my [ 📡 Situation Monitor ] pulls live revenue from Stripe API, more revenue and expenses from my Wise API (bank), etc. Then for historical data I imported a lot manually with CSV or just copy pasting into Claude Code on the server I also asked Claude in Chrome to download all the PDF invoices for lots of services like Linode, Cloudflare etc. because I don't have access to bank transactions before 2022 sadly
three layers of security for agents: machine isolation, capabilities limitation, and runtime validation
Maybe a fix for the AI bots here is a [ Report as AI ] button @nikitabier It would add a human input to the AI bot detection which then an AI could investigate and de-rank the account if there's enough reports Obviously need to build in failsafes so it's not abused to take down accounts but Community Notes shows you can have an algorithm that makes sure that doesn't happne
📊 Added live investment returns now to my revenue chart on my new 📡 Situation Monitor dashboard I built with Claude Code You can see the center bars are my projects revenue and then those green spikes are (mostly S&P500) investment returns Some months they very are up and I make more from investments than my projects (like 2025-05 about 3x more!) and other months they are down, meaning I actually lose money You can also see me start make money with Nomads .com (red), then Remote OK (blue), then that peaking during COVID, and then Interior AI (brown) and Photo AI (pink) take over and some white dots for X payoutsIf you look [ Yearly ] it gets better though, mostly up except 2022!
there are a lot of small, "unsexy" stuff like this in LangSmith Deployments even if they are small individually, as a collective makes deploying agents to production much easier!
@hwchase17 webhooks are the unsexy bit that make agents actually useful. otherwise they just finish and sit there like a demo
View quoted postLangSmith Deployments contains everything you need to run agents in production One of those things is "webhooks" - notify users (or kick off other actions) when an agent finishes!
⏳ Tired of constantly checking if your long-running agent is done? Use the LangSmith Deployment webhook feature to send a slack ping when your run finishes! Here's how to wire it up in a couple of minutes 👉 https://youtu.be/SR6eTBb0k2U Docs: https://docs.langchain.com/langsmith/use-webhooks
RT Sydney Runkle agents built w/ deepagents are often LONG running (they do a lot of work)! use LangSmith Deployment to deploy + manage your long running agents Original tweet: https://x.com/sydneyrunkle/status/2036128785825681604
⏳ Tired of constantly checking if your long-running agent is done? Use the LangSmith Deployment webhook feature to send a slack ping when your run finishes! Here's how to wire it up in a couple of minutes 👉 https://youtu.be/SR6eTBb0k2U Docs: https://docs.langchain.com/langsmith/use-webhooks
RT LangChain OSS How ServiceNow’s AI Agents Resolve 90% of Tickets Autonomously with NVIDIA AI-Q - built with LangChain Deep Agents The LangChain-NVIDIA stack enables developers to build agents at increasing levels of complexity. Built on top of LangChain Deep Agents, NVIDIA AI-Q Blueprint is a full production enterprise deep research system that ranks #1 on deep research benchmarks. Learn more about how ServiceNow builds agents: https://www.youtube.com/watch?v=ZPC6cfr1RVk Original tweet: https://x.com/LangChain_OSS/status/2036126090071101818