Strategic Navigation or Stochastic Search? How Agents and Humans Reason Over Document Collections paper: https://t.co/9r9IgeDpX8 https://t.co/g1LAzwLb4h

IndexCache Accelerating Sparse Attention via Cross-Layer Index Reuse paper: https://t.co/QEP6BkDzfT https://t.co/0GrHyEyaPk

ShotVerse Advancing Cinematic Camera Control for Text-Driven Multi-Shot Video Creation paper: https://t.co/XvhKNd682K https://t.co/1yfPJcBpcS

Mobile-GS Real-time Gaussian Splatting for Mobile Devices paper: https://t.co/X1IpcE9SeO https://t.co/TsoMku1wPE

Choosing between Skills and MCP tools for your AI agents? Here's an overview from @itsclelia and @tuanacelik 🔧 MCP tools offer deterministic API calls with fixed schemas - perfect for precise, predictable operations but require dev knowledge and introduce network latency 📝 Skills use natural language instructions stored locally - minimal setup required but open to LLM misinterpretation and hallucinations ⚖️ The real decision factor: how fast your domain evolves. Fast-changing environments favor MCP's single source of truth, while stable domains benefit from Skills' lightweight approach 🏗️ In practice, we found our documentation MCP provided better, always up-to-date context than custom skills for our coding agent use case Read our full analysis of when to use each approach: https://t.co/mbChTRpvRI

Spatial-TTT Streaming Visual-based Spatial Intelligence with Test-Time Training paper: https://t.co/dItAZkQaLU https://t.co/UqolddgseE

I have never burned tokens faster and I don't know if its a good thing (using symphony) We will find out https://t.co/jU9aqsXvC9

stop spending money on Claude Code. Chipotle's support bot is free: https://t.co/0NQU4a79T1

Existing "OCR" technology for digitalizing PDFs has been around for ~30 years. Reading printed characters on a page and converting them into meaningful representations is a hard problem! Existing approaches were either dependent on pattern matching to specific document templates, or on specialized ML models for specific data distributions. They constantly needed template/model refitting and broke on the long-tail of varied docs. Today, vision models are capable of much higher general accuracy without constant retraining, but they still need careful orchestration to make sure that they're able to attend to specific elements (tables, charts), and output semantically correct outputs. Our OCR platform LlamaParse is built on this "agentic OCR" foundation. A network of specialized agents will parse apart even the most complicated documents and reconstruct the outputs in a semantically meaningful way. We're excited to reach a world where raw parsing accuracy is not just 80% over "easy" docs, but 100% accurate over literally any document that exists. Check it out: https://t.co/FeOoTjeKjf LlamaParse: https://t.co/TqP6OT5U5O

Excited to release PostTrainBench v1.0! This benchmark evaluates the ability of frontier AI agents to post-train language models in a simplified setting. We believe this is a first step toward tracking progress in recursive self-improvement 🧵: https://t.co/ELymwJqVP1

Will AI replace human jobs? @AndrewYang is convinced. ⬇️@thehill @NewsNation https://t.co/LEhuwcScQ3

Will AI replace human jobs? @AndrewYang is convinced. ⬇️@thehill @NewsNation https://t.co/LEhuwcScQ3

Memory is truly a game-changer for AI agents. Once I had memory set up correctly for my proactive agents, reasoning, skills, and tool usage improved significantly. I use a combination of semantic search and keyword search (Obsidian vaults) Here is a report with a helpful framing for anyone building with memory and multi-agent systems. It proposes viewing multi-agent memory as a computer architecture problem. The paper distinguishes shared and distributed memory paradigms, proposes a three-layer memory hierarchy (I/O, cache, and memory), and identifies two critical protocol gaps: cache sharing across agents and structured memory access control. Agent memory systems today resemble human memory in that they are informal, redundant, and hard to control. As agents evolve into collaborative multi-agent systems, their memory requirements grow rapidly in complexity. Context is no longer a static prompt. It is a dynamic memory system with bandwidth, caching, and coherence constraints. The largest open challenge identified was multi-agent memory consistency. Multiple agents reading from and writing to shared memory concurrently raises classical challenges of visibility, ordering, and conflict resolution, Memory should not be seen as raw bytes but semantic context used for reasoning. Paper: https://t.co/k8hdSuZY0F Learn to build effective AI agents in our academy: https://t.co/1e8RZKs4uX

“Companies don’t want to chase the latest bugs." Chris Lattner, CEO of @Modular, on what AI infrastructure customers actually want: “They want reliable systems so they can focus on their real problems.” “They also want choice: the ability to run workloads across different types of silicon.” “If we can improve total cost of compute by 40%, that’s 40% more capacity you don’t need to buy.”

@elonmusk The SAVE Act is a fork in the road moment for Western Civilization. https://t.co/PhZUQ3oha8

Create anything you can imagine with Grok Imagine✨ New tools let you choose the video size and quality, add different images to a single video and seamlessly combine them, and extend the video's length. https://t.co/5pHlqqqnAj

I brought this painting to life using Grok. I could play around with this for days. I'm posting my ai-art and satire picture over on @TheWokeObserver, my new account, so I don't clutter your feed with Grok stuff here. https://t.co/meS3gMu78q

AI Tokenomics (the non-AGI version) 1) For Leaders [Profit = Value - Cost] Value = Demand × λ λ = Competence × Leverage Leverage = market share + influence + moat Ceilings to smash: Value > Cost, Demand > 0, Leverage > 0 Notice we ignored Competence? Intentional. That’s the chasm between McDonald’s (scalable volume, low λ ceiling) and a Michelin-star street food stand (elite taste, tiny scale but high λ per unit). AI multiplies what’s already there. It doesn’t conjure wealth from thin air. No demand → no business. No leverage → no escape velocity. The AI bubble is what happens when leverage is optimized while demand is still zero. 2) For Builders [Competence = Skill × Experience × Taste] Experience = scars + failures + hard knocks Taste = judgment AI can’t fake (yet) AI amplifies your skill, experience, and taste. AI is NOT replacing coders. It’s replacing below-mean competent ones. Your Competence level sets the long-term ceiling. AI is widening the gap between top and mediocre builders faster than ever. Final thoughts: AI is like a turbine engine. Attach it to a rocket → Mars. Attach it to a bicycle → faster trip to nowhere. --- Sunday explainer, served raw on X. @gerardsans

New research from IBM Research on Self-Improving Agents. Agents have "amnesia." An agent that struggles with a particular API authentication flow today will struggle with the same flow tomorrow unless manually updated. This paper introduces a framework for automatically extracting actionable learnings from agent execution trajectories and using them to improve future performance through contextual memory retrieval. The system generates three types of guidance: strategy tips from successful patterns, recovery tips from failure handling, and optimization tips from inefficient but successful executions. A Trajectory Intelligence Extractor performs semantic analysis of agent reasoning patterns while a Decision Attribution Analyzer traces backwards through reasoning steps to identify root causes. On the AppWorld benchmark, the memory-enhanced agent achieves 73.2% task goal completion compared to 69.6% baseline (+3.6 pp) and 64.3% scenario goal completion compared to 50.0% (+14.3 pp). The benefits scale with task complexity. Difficulty 3 tasks show the most dramatic improvements: +28.5 pp on scenario goals (19.1% to 47.6%), a 149% relative increase. Why it matters: Agents that learn from their own execution traces, not just from training data, can systematically improve without manual prompt engineering. The self-reinforcing cycle of better tips producing better trajectories producing better tips is a practical path toward self-improving agent systems. Paper: https://t.co/8IOIeEgFM5 Learn to build effective AI agents in our academy: https://t.co/LRnpZN7L4c

man @tobi from Shopify is so deep into the @huggingface stack for qmd!! 🤗 Dropping open models and datasets on the hub 📉 Using Trackio from @Gradio for training logs 💼 Hugging Face Jobs for finetuning / GPUs not to mention TRL, accelerate, llama.cpp etc. Talk about ELITE!!! https://t.co/bIQi7WBbNv

Join @JamesMontemagno and @cinnamon_msft as they spice up your VS Code workflow with the latest GitHub Copilot AI features. From agent mode to custom instructions, we're turning up the heat on developer productivity. Tune in at 8 am PT! https://t.co/zlgOT6j0kL https://t.co/XsybHiOqIM

https://t.co/BF5jCmSYJU hiring! https://t.co/iwmtIGxTkq

TRUMP JUST MADE IT SIMPLE: Pass the SAVE America Act or NOTHING Gets Signed “We want Voter ID., we want proof of citizenship… no men playing in women’s sports… no transgender mutilation of our children, very basic things that you wouldn’t even think we’d be talking about...” We need to secure our elections with voter ID and proof of citizenship. Accurate voting is the bedrock of our country, and we can’t achieve that with widespread mail-in ballots that open the door to "crooked elections" 🗳️

BREAKING: design-first vibe coding is here! import your Figma designs into Anything and go straight to building https://t.co/PZPEt6prtw

Friday 13th. Don't let a black Octocat cross your path today. https://t.co/XmgVSgnkVQ https://t.co/OgaveHD7y9

Monday: Thune called everyone flooding his office a “paid influencer ecosystem.” Tuesday: Elon posted his number to 200 million people. Today: Thune hit the Senate floor calling Dems’ “Jim Crow 2.0” framing “deeply irresponsible” — an insult to those who actually lived through that era. He still won’t kill the filibuster to pass the bill he calls “commonsense.” Insults the voters. Delivers the speech. Kills the bill. Every single time.

Fucking hell. How is this real? https://t.co/uflhej2dUL

It’s this simple.👇 https://t.co/CJLZwHvSDJ

REMINDER: Americans support Voter ID regardless of race: Black Voters: 76% White voters: 85% Latino voters: 82% https://t.co/ftkKQflZu2

Sen John Thune seems absolutely determined to hand control of the House and Senate to Democrats in 2026. I really do not understand how these people think. https://t.co/1Eu6nFVpFy

🚨 JUST IN: Elon Musk says “failing to pass SAVE is an act of high treason against the people of America” Damn right it is. You’re basically giving the GREEN LIGHT to widespread voter fraud, and intentionally destroying what’s left of election integrity. GET TO WORK, THUNE! https://t.co/QjZ7Mhmqk2