AI agent capability gaps in infrastructure reasoning stem from missing training data, not model limitations.
“like with everything in NML, uh the gap in models is usually a gap in data”
Latest signals
Every signal tracked from the world's AI conferences — 4,665 talks, quotes & announcements, newest first.
The traditional base-model paradigm of web-scale pre-training is being displaced by post-training
“RL was mostly just a cherry on top, shaping the flavor of the interactions more than conferring extra knowledge or quality onto the base model itself.”
LatchBio operates as a vertical AI lab building benchmarks and agents for biology's data explosion
“We are basically a vertical AI lab for benchmark and agent engineering.”
AWS launches an AI agent in Amazon QuickSight that auto-inherits enterprise catalog metadata for grounded analytics
“Today, we're announcing the Agentic Catalog Experience in Amazon Quick, an AI-powered workflow that helps data curators rapidly define their context boundary, inherit upstream semantics, and enable end users for grounded Q&A and trusted dashboards at scale.”
Taste Labs exits stealth to build data infrastructure ending AI slop in subjective domains
“our whole mission is basically how do we end AI slop?”
LangChain's open-source Deep Agents adds production harness primitives to any tool-calling agent with one function swap
“if you're not the model, you're the harness”
Google DeepMind launches Gemini Robotics 2 with whole-body intelligence, dexterity, and multi-robot collaboration.
“What we're building here is the intelligence layer to power any robot to do a broad range of useful tasks.”
Patrick Collison suggests AI may make source code obsolete like compilers made assembly irrelevant
“maybe in the same way we shouldn't mourn source code and we should just transcend the plane of uh instructions to Claudius at all”
Agent loops amplify LLM hallucinations by compounding errors across downstream steps
“one hallucinated fact in step two can poison step three, step four and everything downstream”
Apple WidgetKit enables interactive, personalized app widgets on iOS home screen
Core Automation aims to maximize human agency, not remove humans from AI research loops
“most automated lab in the in the world”
Google ADK voice agents using Gemini Live can be traced with LangSmith for production observability
“To take that agent to production safely, you need visibility into what your agent is doing and the ability to test its behavior in a number of different scenarios that it might encounter with real end users.”
AI benchmark prompts are so unrealistic that experienced engineers would never write them in practice
“No one writes prompts like these ever.”
AWS Bedrock AgentCore Observability enables systematic diagnosis of production agent performance bottlenecks
“Slow is subjective. What's acceptable for a batch processing agent is unacceptable for a customer service chatbot.”
NVIDIA Video Codec SDK 13.1 adds zero-copy transcode, AV1 B-frames, and frame-accurate seek
Prime Intellect is extending RL to real-world tasks lacking verifiable reward signals.
OpenAI outlines safety and transparency practices aligned with EU AI Act compliance
OpenAI frames its strategy as 'abundant intelligence' via a full-stack AI approach
“A full-stack approach to making advanced AI more capable, more affordable, and more widely useful.”
Decagon runs 90% open-source models because they outperform frontier models on specific enterprise tasks
“An AI agent should just be the front door of your business.”
Hugging Face releases Tau, a Python port of its Pi coding harness for AI workflows
“coding harnesses are one of the most important parts when creating reliable AI workflows”
Netic builds AI to autonomously run essential services businesses between enterprise and end customer
“Ned builds AI to run millions of real world businesses that keep the world running.”
Univé deployed ChatGPT Enterprise via leadership-driven, employee-led AI transformation at scale
Power Pages external-user authorization now enforces natively in Dataverse via mapped identities and record filters.
“It's a meaningful step forward for how customer-to-business access is enforced across Power Platform.”
GPT-5.6 recursively self-optimized OpenAI's own serving infrastructure, cutting costs 20-80%
“GPT-5.6 Sol was actively used to analyze production traffic, tune load balancing, and autonomously rewrite production kernels in OpenAI's own Triton and Gluon languages.”