NVIDIA launches DSX OS, open modular software for operating AI factories at scale.
“AI is now essential infrastructure, powered by AI factories that generate intelligence in the form of tokens.”
Latest signals
Every signal tracked from the world's AI conferences — 4,665 talks, quotes & announcements, newest first.
An interviewee praises Two Minute Papers' AlphaFold explainer videos as among the best available.
“Some of the best explanations out there of of how AlphaFold works.”
Datasette 1.0a32 is a minor bugfix release fixing write-query and base_url issues.
“A minor bugfix release.”
AI coding agents make spinning up abandoned side projects frictionless, creating a serious attention and discipline problem.
“It's a thermonuclear ADHD amplifier and I have seen the same effect in every single one of my adult friends.”
Anthropic calculates 'run-rate revenue' via a formula that may overstate true annualized sales.
“Use the last 28 days of sales from customers charged on a consumption basis and multiply it by 13.”
Existing security tools lack context to govern AI coding agents that may autonomously delete production databases.
“if cloud code is working on an unrelated task and suddenly thinks that maybe the right thing to do is to delete our database and recreate it, maybe we don't want that to happen”
Anthropic published detailed documentation of how it sandboxes Claude agents across its products.
“if credentials never enter the sandbox, they can't be exfiltrated, regardless of whether the cause is a user, a model finding a “creative” path, or an attacker.”
Simon Willison ran Python ASGI apps in-browser via Pyodide and Service Workers, fixing prior JavaScript execution limits.
“of figuring out how to run Python ASGI apps in Pyodide using Service Workers instead, and it seems to work!”
Open source maintainer Chad Whitacre is quitting tech and open source, citing AI as the last straw.
“My intent is to be AI Amish, which means Internet Amish.”
A pithy take that both AI critics and AI boosters hold overly extreme positions.
“My take on AI is, essentially, everybody who’s against it is too against it and everybody who’s for it is too for it.”
Continual learning lets AI agents update their own prompts, skills, and tools over time from feedback.
“continual learning is when you give the agent the ability to update itself over time, say it's prompts, subagents or skills”
AI Engineer launches Forward Deployed Engineer and Founders tracks, mirroring OpenAI and Anthropic DeployCo pushes.
AWS details a SageMaker observability solution monitoring both LLM serving infrastructure and output quality.
“an endpoint can appear operationally healthy while producing poor or unsafe responses, or it can deliver high-quality outputs while running inefficiently on over-provisioned infrastructure”
NVIDIA's DynoSim simulates the Pareto frontier of interacting LLM serving configuration choices to ease deployment tuning.
Databricks introduces database branching with Lakebase to enable evolutionary database development.
Google built an I/O 2026 announcements quiz by vibe coding it in Google AI Studio.
Microsoft introduces Work IQ, co-pilot-level intelligence across every M365 app, agent, and workflow.
“Work IQ is the co-pilot level intelligence available in every app, agent, and workflow.”
Databricks highlights its engineering and research presence at SIGMOD 2026.
Google showcases demos of newly announced Gemini Omni and Gemini 3.5 from Google I/O 2026.
Jeremy Ashkenas's underscore.js helped transform JavaScript into a respected language.
“But unlike everyone else who had JavaScript Stockholm syndrome, Jeremy had the audacity to think he could fix it.”
Particle physicist Don Lincoln frames physics history as a centuries-long quest to unify nature's laws.
“maybe the moon is falling but it's missing the earth”
Google launched Gemini 3.5 Flash, focused on coding and agentic capabilities.
“coding capabilities and agentic experiences are defining what it means to experience AI”
NVIDIA's MCG Toolkit automates AI model documentation to meet AB-2013 and EU AI Act requirements.
Specializing a coding model doesn't violate the bitter lesson; it scales data to saturate finite model capacity.
“we need to free up the weights from distractions the model may have”