Gmux is a GPU resource management tool that enables spatial and temporal sharing of GPUs across multiple jobs, similar to tmux for terminal multiplexing. It allows users to run GPU workloads from any machine without local GPU hardware, with automatic resource allocation, monitoring, and usage tracking across NVIDIA, AMD, and Intel GPUs.
Large-scale agent systems face distributed computing challenges similar to traditional operating systems. Agents need resource management, scheduling, and context persistence rather than running indefinitely, while multi-agent coordination introduces overhead that limits scalability depending on task structure.
Google has expanded CC, an AI agent originally designed for personal task management, into a family-focused tool that helps households coordinate schedules, manage tasks, and handle logistics. CC now operates as a shared agent with its own Google account, allowing up to six family members to collaborate while controlling what information each person shares, integrating with Gmail, Calendar, Tasks, and other Google services.
Floria is a high-throughput serving engine for large language models that eliminates GPU starvation by replacing Python-based scheduling with native hardware dispatching. Running on a single vCPU, it achieves 191k tokens/sec and 100% GPU utilization, compared to conventional systems like vLLM that leave GPUs idle 25-40% of the time due to host scheduling latency.
Chutes released a year of production LLM serving request traces covering 6.12 billion requests across 9,174 models, enabling research into batching, scheduling, and GPU optimization. Key findings show high temporal locality in user requests, LRU cache effectiveness, and cache-aware routing improvements.
Xiaomi's MiMo Desktop is a smart scheduling system that automatically assesses tasks and routes them to appropriate frameworks, dynamically selecting models and tools. It splits complex work into parallel subtasks handled by multiple agents and features a Self-Evolve engine that continuously improves its own code.
Google has expanded CC, an AI agent initially designed as a personal assistant, into a family-oriented tool that helps households coordinate schedules, tasks, and logistics. CC operates as a shared Google Account with permission controls, allowing up to six family members to collaborate on managing school schedules, activities, bills, and meal planning while maintaining individual privacy over what information each member shares.
Woven Matter v0.2.0 introduces workspace integration tools enabling agents to coordinate across conversations, schedule follow-ups, access notes and calendar events, and manage document versions. New features include inter-agent communication, file transfer to remote agents, and improved permission controls, plus performance improvements for large documents and background operations.
Mise is a web-based meal planner that schedules multiple dishes to finish simultaneously by working backwards from the desired eating time. It requires no login or app, works offline after loading, and provides alerts for each cooking step.
OmniFire is an all-in-one platform that manages social media accounts, business listings, and mobile apps through a unified interface for publishing, analytics, inbox management, and team permissions. Users can compose once and adapt content per platform, schedule by timezone, and track engagement and performance across social accounts, app stores, and business profiles in one dashboard.
An open-source AI-powered tool that transforms articles and projects into on-brand LinkedIn posts via OpenRouter, manages approvals and scheduling through Convex, and syncs performance metrics back to a dashboard. Built with Next.js, shadcn/ui, and LinkedIn's OAuth and Posts APIs.