Today in AI: Agents Go Mobile and Autonomous
July 07, 2026
Anthropic is pushing Claude onto smartphones, signaling a major shift toward persistent, mobile-first AI agents. Meanwhile, real-world autonomy is advancing from chip manufacturing to actual battlefields, highlighting the dual-track progress in consumer and industrial AI.
🚀 Claude Cowork Agent Expands to Mobile and Web
Anthropic is rolling out its Claude Cowork platform to mobile devices and the web, starting with Max subscribers. This allows tasks to continue running even after you close your laptop, marking a crucial step towards always-available, persistent AI agents. It's a direct challenge to the desktop-bound model of AI assistance. (The Verge)
💰 China's DeepSeek Plans Its Own AI Chips
Facing US export controls, Chinese AI firm DeepSeek is developing its own AI chips to reduce dependency on Nvidia and Huawei. This move underscores the intense geopolitical battle over AI hardware sovereignty and could accelerate a fragmented global chip ecosystem. (Ars Technica)
💰 NVIDIA Touts CPUs as Critical for AI Agents
NVIDIA argues that high-performance single-threaded CPUs are the unsung heroes of the 'agentic AI era,' handling the reasoning and task execution that models command. This reframes the AI hardware conversation beyond just GPUs, highlighting the full-stack compute requirements for complex agents. (NVIDIA)
🚀 Savi Raises $7M to Fight AI Scams
The startup Savi launched an app to protect consumers from realistic AI scams like fake kidnappings, backed by a $7 million seed round. As generative AI makes deception terrifyingly easy, consumer-focused defense tools are becoming a necessary and investable category. (TechCrunch)
🚀 First US Autonomous Ground Vehicles Fight in Ukraine
Forterra has deployed over 100 self-driving ATVs in Ukrainian conflict zones. This isn't a simulation; it's a live, large-scale deployment of autonomous systems in warfare, marking a sobering milestone for military AI. (TechCrunch)
🛠️ Google Expands Managed Agents in Gemini API
Google is adding capabilities like background tasks and remote Model Context Protocol support to its Gemini API for building production-ready agents. The platform wars are heating up, with every major player racing to offer the most robust agent development suite. (Google AI)
🔬 Space Lab to Beam Protein Data for AI Aging Research
A British startup launched a lab into orbit to study proteins related to age-related diseases, with data being used to train AI models. This is a fascinating example of using unique, hard-to-get data (microgravity effects) to power next-generation biomedical AI. (Wired AI)
The takeaway: AI agents are breaking free from the desktop, becoming persistent mobile companions and autonomous battlefield operators simultaneously.
