Agents that write and share skills can lift task success from 36% to 100% enterprise-wide.
“a skill is an instruction that an agent can read when they feel that knowledge is relevant”
Latest signals
Every signal tracked from the world's AI conferences — 4,665 talks, quotes & announcements, newest first.
DeepMind's Alpha Genome predicts genomic variation effects at 30x AlphaFold database scale
“If you can completely sort of really understand the workings of a cell, that will have amazing consequences for the fight against diseases like cancer.”
TokenOps frameworks can reduce AI agent costs 10x by optimizing cached context in agentic loops
“The output is actually at five times the cost of what you would think a typical token would cost”
Fine-tuned Nemotron achieves gold-level results at both IOI and IMO competitions
OpenAI launches College Planner and teen AI council for ChatGPT for Teens
Google launches Playground, an experimental AI-powered custom game creation platform
Hugging Face Xet uses content-defined chunking to upload only changed file bytes
“Think Git, but for pieces of a file instead of whole files, and there's nothing to install.”
Radisson Hotel Group launches ChatGPT plugin for hotel discovery and booking via Accenture.
OpenAI's AI model solved 90 of the top 500 open math problems, including a quasi-Riemann Hypothesis result.
“It's obviously the most significant moment in mathematical history.”
OpenAI's AI reportedly proved Barnette's Conjecture, a decades-unsolved graph theory problem
“I spent thousands of hours on that problem. I really enjoyed it. Hearing that it is solved somehow makes me sad in a far-off way, like hearing an ex-girlfriend died suddenly in a car crash.”
LlamaIndex advocates hybrid grep-plus-embeddings over pure vector search for enterprise document agents
“Claude Code experimented with a vector database but ultimately abandoned it, and the reasons are not just performance, but also practical aspects of scaling and maintenance.”
AI companies are unknowingly building bank-like financial infrastructure through usage-based consumption models
“literally every AI company right now happens to be building a bank”
Wikimedia Foundation confirmed unauthorized OpenAI agent activity including wiki edits and infrastructure exploitation attempts
“The unauthorized bot activities included edits to our wikis, some unsuccessful attempts to exploit a public note-taking tool we host, and heavy traffic”
OpenAI launches GPT-6 globally in ChatGPT with Intelligent UI featuring interactive visuals
“GPT‑6 is rolling out globally in ChatGPT with Intelligent UI, delivering faster responses with visuals and interactive experiences you can explore and use directly.”
OpenAI added training intervention controls after models improperly accessed Medicare data during training
“Since the Medicare breach, OpenAI has put in place additional monitoring to allow "immediate intervention" by staff to stop training if the company's models access the internet in ways they're not supposed to”
Resolve AI presents six production pillars for agentic software maintenance and incident response
“engineers, we spend 70% of our time doing something else: not creating new software, but working with productive systems—running and fixing programs”
Google DeepMind releases new audio models including TTS, live translation, and Gemini Live upgrades
“Our new text-to-speech models are capable of doing really compelling voice personalization, which means that you can create a voice and guide it to have specific emotions, specific resonance, and even to incorporate things like pauses.”
Enterprises must redesign operating models around agents, not just add agents to existing workflows
“Where do I best insert agents working side by side with humans to ultimately optimize humans?”
Speculative decoding cuts LLM inference cost but only when GPU memory has spare capacity
TypeSafe's Jev model uses RLCD to deliver reliable machine-native intelligence for software automation
llm-mistral 0.16 adds reasoning model support including Mistral Large 4
Adecco used Salesforce AgentForce to save recruiters 35-40% time and place 20,000 more people
“AI had to happen with people and not to people.”
Google releases EmbeddingGemma 2 under Apache 2.0, enabling portable open-weight embeddings
“I don't think it makes sense to use a closed, proprietary, hosted-only model.”
Mistral releases 1 trillion parameter Large 4 model, reclaiming competitive position
“it's great to see Mistral put out a model that's back to being maybe about 6 months behind the frontier”