Meta Launches Hatch AI Agent Platform in Coming Weeks

Instinct, an AI assistant in early testing, is drawing user enthusiasm for its capabilities but also raising red flags around privacy and security. The tool's broad access permissions, expansive terms of service, and ability to take autonomous action on behalf of users create potential risks that testers say warrant scrutiny. The tension between functionality and user protection highlights ongoing challenges in how AI assistants balance capability with safeguards.

NVIDIA's Vera Rubin NVL72 GPU system delivers up to 30x higher throughput per megawatt than its GB300 NVL72 predecessor on agentic AI workloads, according to measurements using the SemiAnalysis AgentX benchmark. Agentic tasks consume 15x more tokens than simple chat because agents iteratively query databases, invoke sub-agents, and accumulate context across multiple reasoning steps. For power-constrained AI infrastructure operators, the efficiency gain translates directly to running significantly more agent-based work within the same energy budget.

Anthropic updated Claude Tag, its Slack agent, to read full conversation context rather than evaluating messages individually, improving its ability to decide when to interject unprompted by roughly 30%. The company frames this as a shift from single-user AI tools to 'multiplayer AI' agents that operate across teams and pursue organizational goals. Scott White, Anthropic's head of enterprise product, argues this represents AI transitioning from personal assistant to organizational colleague.

AWS published a technical walkthrough for building a voice-based restaurant ordering system using Amazon Connect, Lex V2, and AI agents. The system answers inbound calls, takes orders through natural conversation, and confirms them without requiring apps, websites, or customer logins. The architecture separates the telephony channel, conversation logic, and backend services to keep ordering functionality independent from how customers access it.