AI Agents & MCP
The AI Agents & MCP stories that matter — sourced, dated, and explained in plain English, without the hype.
The lead
Latest first
Latest in AI Agents & MCP
NAVI-Orbital Deploys First Zero-Shot Vision-Language Model In Orbit for Earth Observation
An arXiv preprint details NAVI-Orbital's April 2026 in-orbit demo of the first zero-shot vision-language model for autonomous Earth observation, which achieved…
Allen AI Releases olmo-eval for LLM Development Loop
Allen AI's olmo-eval extends OLMES with modular benchmarking, checkpoint analysis, and agentic evaluation for iterative model development — Allen AI Releases…
Self-Evolving Agent Boosts Legal Case Retrieval via Rule Learning
self-evolving agent legal case retrieval: New arXiv paper shows LLM agent that iteratively learns query-rewriting rules beats human-designed baselines on…
GitHub Copilot Cuts VS Code Token Waste via Context, Routing
Copilot context handling: GitHub’s latest Copilot for VS Code update introduces prompt caching, on-demand tool loading, and HyDRA-powered model routing to cut…
DivInit Fixes First-Turn Redundancy in Agentic Search
diverse query initialization: New DivInit method boosts multi-hop QA by 5-7 points across five models by diversifying initial queries in parallel agentic…
GitHub Copilot Cuts Token Waste With Caching, HyDRA Routing
GitHub updated Copilot for VS Code with prompt caching, on-demand tool search, and expanded Auto model routing to cut redundant token use…
OpenAI Deployment Simulation Catches Model Misbehavior
OpenAI's Deployment Simulation replays real conversations to predict GPT-5-series behavior, catching novel misalignment patterns that traditional evals miss.
Android 17 Launches on Pixel With Mandatory Resizability
Google releases Android 17 on Pixel devices June 16, introducing mandatory large-screen resizability for API 37 apps and AppFunctions for on-device AI…
Microsoft, OpenAI, Google Outline Three-Layer AI Stack
enterprise AI adoption: Microsoft's Intelligence+Trust framework, OpenAI's deployment simulation, and Google's $1.5B Alabama buildout reveal converging…
OpenAI Simulates Deployments to Predict Model Behavior
OpenAI Simulates Deployments to Predict Model Behavior: OpenAI's new Deployment Simulation replays real conversations to catch undesired GPT-5-series behavior…