[archive]
295 stories · newest first · page 21
A new server chip, reportedly six to ten times more token-efficient than Google's current hardware, is planned for 2028.
The Model Context Protocol is moving to a stateless session model, reducing infrastructure headaches for companies deploying AI agents at scale.
The startup aims to democratize AI chip access by using an autonomous agent to generate low-level kernel code for any hardware.
The AI company is said to be developing a ChatGPT-powered companion device with a personality and autonomous movement, drawing on user data for personalized interactions.
The limited-run Codex Micro keyboard lets devs manage AI coding agents with dedicated keys and a reasoning dial, while a screenless smart speaker looms amid an Apple trade-secret lawsuit.
Researchers propose Oracle, a memory system that lets AI agents maintain persistent, structured user profiles across sessions without retraining.
Two technical decisions require Google to give competitors like ChatGPT and Perplexity comparable access to Android system features and Google Search data.
A new open-source tool lets agents claim tasks as dependencies resolve, reducing idle polling overhead.
A new browser-based challenge lets developers practice red-teaming AI models and their tool-using agents in a safe, interactive environment.
The data analytics giant raises another massive round, this time at a $188 billion valuation, while pushing open-weight models and cost-conscious agentic workflows for enterprise AI.
Moonshot AI’s latest model reignites debate over distillation, national security, and whether open-weight AI leads to a dystopian 'full AI communism.'
A new open-source tool lets developers boot MCP servers in ephemeral microVMs, inspecting behavior before any agent interaction.
