The Hallway Track
Every signal tracked from the world's AI conferences — 4,665 talks, quotes & announcements, newest first.
Enterprise AI deployments fail at security integration, not agent building, per Azure MVP assessments
“many organizations don't struggle with the building of agents. What they struggle with is integrating them in a way that meets enterprise security and governance requirements.”
Microsoft's CAF-aligned Azure AI landing zone delivers enterprise RAG at millions-of-documents scale
“The result is a grounded response traceable to enterprise data, not a hallucination.”
Claude Code can build a Copilot Studio AI agent in under 10 minutes
“we can develop an agent in 10 minutes”
GitHub's agent capabilities remain underutilized among developers unfamiliar with platform depth.
Enterprise agentic RAG systems lack active feedback-based learning in production today
“there is uh no active learning uh for agents is visible anywhere”
AWS integrates Ray distributed computing framework natively into SageMaker HyperPod infrastructure.
AWS presents a RAG-based avatar system for capturing and retrieving institutional knowledge at scale
Meta open-sourced MetaRoCE, a new RDMA transport protocol built for million-GPU AI clusters on Ethernet.
“The fabric sees packets, but the NIC sees intent.”
Apple migrates Sign in with Apple relay addresses to new private.icloud.com domain
Meta's MTIA 300 embeds NICs on-chip, making communication first-class for recommendation model training
“We've made communication a first-class citizen in the chip's design, not an afterthought handled by general-purpose compute cores.”
AI systems lack brain-like plasticity to update deeply ingrained weight-level knowledge when it becomes outdated
“Something that has been true for 20 years when it stops being true it can go and update that and get rid of that. And that is the capability I think that's extremely useful we would want in our systems.”
Databricks now supports running and debugging workloads directly from local IDEs
Databricks uses AI internally to accelerate and automate incident investigation at scale
Qwen 3.8B runs on consumer laptops while matching frontier model performance
“if we wait a bit, we might get frontier level systems running on our laptops”
Anthropic Python SDK v1.0 switches to httpx2, mirroring OpenAI's identical move two weeks prior.
“Upgrade to anthropic>=1 - read https://raw.githubusercontent.com/anthropics/anthropic-sdk-python/refs/heads/main/MIGRATION.md and get the tests passing”
AWS launches Agent Registry and open ARD specification for cross-environment AI agent discovery
“if every registry describes resources in the same format and exposes discovery through a common protocol, publishers describe once and consumers discover everywhere”
AWS shows how to build a full agentic voice ordering system for restaurants using Amazon Connect and MCP
AWS demonstrates Amazon Bedrock-powered metadata harmonization workflow with human-in-the-loop validation
NVIDIA Spectrum-X Ethernet reengineers networking for giga-scale AI GPU clusters
NVIDIA Vera Rubin and Blackwell set new performance-per-watt benchmarks for agentic AI workloads
“across 100 trillion tokens of real-world usage, OpenRouter's State of AI report found that average prompt tokens per request grew roughly fourfold”
NVIDIA BlueField-4 DPU enables dedicated networking for agentic AI factory infrastructure
“Agentic AI factories connect diverse users, agents, applications, data sources, and storage systems to massively accelerated compute at multi-terabit bandwidth per server, making dedicated DPU processing essential for line-rate networking, storage, and security.”
NVIDIA Vera CPU is purpose-built to handle orchestration and tool execution for agentic AI fleet workloads.
“AI factories are interconnected systems where fleet economics depend on how efficiently the entire stack converts power and capital into completed agent tasks.”
NVIDIA Groq 3 LPX accelerator enables ultrafast interactive inference on Vera Rubin NVL72 at long context
“NVIDIA Vera Rubin NVL72, the most versatile machine ever built, delivering high throughput and interactivity across the widest range of AI workloads—from small to large models, both open and closed.”
Frontier models handle big problems but local models offer control and privacy
“you just want some control, you want some privacy, you just want to keep things local and make my stuff my stuff”