TurboQuant compresses agent retrieval embeddings to 3-4 bits, cutting memory cost 5x without degrading search quality.
“Today, we will see how you can cut memory cost of agent retrieval five times without breaking your search.”
Latest signals
Every signal tracked from the world's AI conferences — 4,665 talks, quotes & announcements, newest first.
Spec-driven development—writing specs and design docs before code—produces faster, higher-quality output from LLM coding assistants.
“Because if you give them just a little bit of leeway, they will go off the rails.”
The open model ecosystem is diversifying globally, with more orgs like Zyphra, Cohere, and Poolside releasing models beyond the early Chinese-dominated landscape.
“Attempts to slow or ban this ecosystem are not only futile, as the history of tech-related bans has shown, but also unsafe and anti-freedom.”
Agent failures stem from static retrieval that never learns from eval and observability signals.
“We made wrong answers appear faster and cheaper, but we forgot to make retrieval learn.”
Open-source tool Docling extracts structure from unstructured enterprise documents to power RAG and agent systems.
“unstructured data is becoming this new context layer for AI”
NearForm uses AI agents and spec-driven development to iteratively build more reliable, secure AI agents.
“an agent is an LLM inside a an agentic loop, which is connected to tools and can retrieve context. That's it.”
Browser agents fail due to poor infrastructure around models, not model intelligence; better page representation makes cheap models fast and reliable.
“models are pretty smart, but it's the infra around them that sucks”
Giving AI agents every tool at once degrades speed, cost, and accuracy as the catalog grows.
“It fails because the design does not fail because one tool is badly written.”
An investor argues India is best positioned to build the largest AI companies in this technology wave.
“This is about do you understand this technology 10x better than everyone else and I think nobody does that better than in India.”
Use MCP servers for known, enterprise-auth, human-in-the-loop tools; use CLIs for model-familiar, light-surface tasks to save tokens.
“the MCP server for GitHub is about 45,000 tokens because it's giving every single API name, description, and schema to the model. But with the CLI, running the help command where it can see all of the tools, there's only about 1,500 tokens”
AWS on Air episode previews a demo of AI agents working with Amazon WorkSpaces desktops.
“we're going to talk a little bit about um having some AI agents actually work with your desktop”
OpenAI previewed GPT-5.6 Sol, a coding-capable model released only to trusted partners.
“GPT‑5.6 Sol does not cross the Cyber Critical threshold under our Preparedness Framework.”
AI labs face a narrow post-release window to recoup frontier model costs, and US data center buildout assumes a global market for US AI services.
“No one is building $100 billion dollar data centers to serve frontier models to whatever 100 companies the US government will allow access.”
Critic argues LLMs do have a learning curve, rebutting claims they take no skill to use.
“This is like saying there's no learning curve to being a manager because your employees will just do whatever you tell them to do.”
Databricks unveiled capabilities to turn video into searchable, actionable intelligence for enterprise use cases.
Databricks offers a decision framework for migrating legacy ETL stored procedures and schedulers to its platform.
England's Office for Students cut a 300-million-record data job from 8 hours to minutes using Databricks.
Brand voice in AI agents fails on edge cases unless instructions are layered into a four-tier architecture instead of one prompt.
“I'm not programming a robot, I'm managing a brilliant intern with an incredibly high IQ and a terrible EQ.”
New media has displaced legacy media's prestige, and creators should embrace novel formats like virtual production.
“It can't be half and half. The whole motion.”
NVIDIA releases an open-source AI-Q Blueprint for deploying production-ready long-horizon AI agents on Oracle Cloud Infrastructure.
“Today, we have long-horizon agents. Systems that plan many steps, split work between sub-agents, keep context across a long task, and run tools in a safe sandbox.”
Apple introduces a new Evaluations Framework at WWDC26 to measure model outputs, including model-judge evaluators.
“if models are judging other models, who's judging the model judges?”
Google demos Gemma running on-device on Hugging Face's open-source Reachy Mini robot for local AI interaction.
“Faster processing means quicker reactions, which is great for robots navigating the real world.”
Frontier model anti-prompt-injection training held up against 6,000 attempts to leak secrets from an AI assistant.
“after 6,000 attempts (and $500 in token spend and a Google account suspension triggered by too many inbound emails) nobody managed to leak the secret”
Cerebras launches an Ambassador Program funding community-led AI builder meetups worldwide.
“We cover your food and venue budget so you can focus on bringing in the best AI builders.”