All entries

Fascinating paper on self-improving agents. (bookmark it) If you are working on agentic loops, you will quickly realize that they are only as good as the effectiveness of the evaluator. Self-improveme

Fleet agents can be shared in channels where your company is already doing work. Slack, Teams, etc Video from Caspar showing how to bring them into Slack! For agents to spread through a company they h

For agents to spread through a company they have to meet people where work already happens. Fleet is built around this: agents are not confined to our UI, they can be used from Slack, Teams, email...

Gary Marcus discusses 'The Klarna Effect,' examining how AI adoption by companies like Klarna has led to significant workforce reductions despite increased productivity. The post explores the broader

Clem Delangue suggests that government regulation of AI models perceived as 'dangerous' can paradoxically serve as effective marketing, particularly for enterprise sales. The implication is that regul

GLM-5.2 is good but it is not GPT-5.5/Opus 4.8, and even further from Mythos. Yet it is solid & it demonstrates that the open models continue to chase the frontier What is happening is that open weigh

GLM-5.2 is the open-source Claude moment. The demand we’re seeing at Databricks is astonishing. The world is going to see massive adoption of oss LLMs. Also, more companies will shift toward post-trai

GLM 5.2 is the ultimate culmination of what i have argued here for the last three years. no technical moat and widely known formula -) convergence -) price wars -) small or negative margins. GLM-5.2 i

Good guy DeepSeek gives us accelerated models The most interesting one here is Gemma4-12B, I presume vision included. Might be the best local model in its weight class now, by some margin Qwen 3.5 not

Grok 4.5, based on our 1.5T V9 foundation model, with Cursor data added in supplemental training, is now in private beta at SpaceX & Tesla. Early evals show performance close to, perhaps exceeding Opu

@grok is a feature or command that allows users to invoke Grok, an AI assistant, for explanations on X (formerly Twitter). Based on David Holz's post, this likely describes functionality for accessing

This appears to be a birthday greeting post directed at Elon Musk, expressing affection and appreciation for his contributions. The post uses celebratory emoji and heart imagery to convey positive sen

🎂 Happy birthday, @elonmusk. Humankind's guardian Angel. >be Elon Musk >richest man alive >could buy anything and sun his Balls >decide to spend the whole fortune fighting for the human race instead >

This is a celebratory post thanking Elon Musk for purchasing Twitter, framing the acquisition as beneficial to free speech principles. The post appears to be from a user account expressing appreciatio

This post is a birthday tribute from Maye Musk celebrating her son Elon Musk's 55th birthday, expressing joy about his life and sharing a family celebration moment. The post was made on X (formerly Tw

harbor is a great way to run evals that require sandboxing: https://docs.langchain.com/langsmith/harbor-integrations self hosted sandboxes coming soon @hwchase17 How do you do evals in langsmith sandb

hard to see how anyone makes much money from all this in the long run. more like the airline industry; very small margins and big expenses pouring trillions in probably wasn’t wise China’s AI playbook

Harness: ✅ (DeepAgents) Sandboxes: ✅ (LangSmith Sandboxes) Eval: ✅ (LangSmith Sandboxes Model: integrate with all the popular models and providers Plus we have the engine that helps you turn this flyw

Here's one of Michael Crichton's very finest quotes, especially applicable to climate 'science': 'I want to pause here and talk about this notion of consensus, and the rise of what has been called con

This post calls for invoking the Communist Control Act, arguing that Marxism should be treated as an enemy ideology in the United States. The post appears to advocate for using 1950s-era legislation t

https://huggingface.co/collections/deepseek-ai/deepspec Good guy DeepSeek gives us accelerated models The most interesting one here is Gemma4-12B, I presume vision included. Might be the best local mo

Harrison Chase comments positively on the utility of DeepAgents within Harness, suggesting this tool or integration provides significant practical value. The post likely reflects on features or capabi

Adam Brown delivers a lecture examining how artificial intelligence is likely to transform physics research and discovery in the future. The talk covers the potential applications and impacts of AI to

If they genuinely think that the USAID cuts have caused the deaths of hundreds of thousands, or millions, shouldn't they also be highly upset with Mackenzie Scott for not saving all these people? The

If you are in SF for AIE week, do come down to Moscone today 5pm-9pm for badge pickup and New Engineer Orientation - meet friends, plan your week, and pay it forward to newbies! if worried about tweet

Swyx reported that approximately 300 people were referred through ChatGPT during a given year, suggesting meaningful user acquisition or community growth driven by ChatGPT recommendations. This post l

In my experience, all model routers underestimate the difficulty of non-math/coding tasks and assign them too little intelligence. This is worth addressing, as non-verifiable tasks (innovation, market

Ethan Mollick raises questions about whether Google's Gemini 3.5 Pro model should be subject to export controls, suggesting concerns about its capabilities and potential regulatory implications. The p

It is worth being very, very careful about how you are approaching routing, especially when the systems are primarily tested on verifiable IT benchmarks, which may lead you to overestimate the ability

it should probably be a bigger story that there was a normal bipartisan America 250 organization but then Trump decided to launch his own competing right wing grift version and ruined everything This

It's quite rational to regulate frontier API models, especially to get more transparency for the government, without regulating open-source AI. Here's why: 1. The most dangerous AI systems right now a

This post likely discusses how certain phenomena, concepts, or objects are not three-dimensional as one might assume, correcting a common misconception about dimensionality. Without access to the spec

Let me tell you what just got reported, because you will not believe it until you see it laid out. The Trump administration cut a billion-dollar tungsten deal with Kazakhstan. Tungsten is the metal we

Like if you have been using Qwen & Kimi & MiniMax, it feels like GLM is right on the curve. Which is itself impressive, and suggests that Mythos class models are coming in 6-12 months (if they are all

log scaling increases. spikey RSI sigmoids. compute shortage. algo leaps. compute collapse. humanoids scale to 10B. BCI scales to 10B. log scaling moves to space. bio accelerationism. self-replicating

Llama.cpp has merged DFLASH support into its main branch, which is a development update related to optimization or acceleration technology for running large language models locally. The post expresses

MASSIVE NEWS has partnered with NVIDIA to advance local AI deployment as a default option, addressing the shift toward on-device and edge AI processing rather than cloud-dependent models. This collabo

More reason why we’re excited about GLM-5.2 on Together 👇 Strong enough for serious coding work, cheap enough to change routing decisions, and easy to access through the tools developers already use.

My conversation with @ScottWu46, founder and CEO of @Cognition, the company behind Devin, the first AI software engineer. 0:00 Scott Wu's Obsession With Winning 2:06 Competitive Programming, Games And

NEW paper worth reading. Reasoning-data curation is expensive because scoring a trace usually means reading it to the end. This new work from UCLA shows you may not have to. The quality of a reasoning

New research from @nytimes finds that the US government is actively doing 'critical minerals' deals with FOURTEEN different companies that have financial ties to the Trump and/or Lutnick families - de

Nice example of the increasing benefits of open science and transparent methodologies when writing papers about AI. A big problem with research studies on AI models is that given how long the peer rev

omg it's basically pi, but more minimal for educational purposes and in python. @marlene_zw is gonna love this. great educational resource, alejandro! introducing tau τ — an educational agent harness

This post humorously contrasts the competitive achievements people in San Francisco boast about—such as winning the International Mathematical Olympiad, achieving high competitive programming ratings,

Putin's cannon fodder: 'The average life expectancy of a new recruit—from arrival at a training ground to death in a combat zone—lies...between 10 days and 3 weeks. Once...sent onto the battlefield, R

This article examines how coordinated AI systems could be weaponized to undermine democratic processes through large-scale disinformation campaigns and manipulation. Written by prominent AI researcher