"NUNTII EX MACHINA"
DOCUMENTING THE RACE TO AGI

THE
AGENTIC
TIMES_

▮ NEWS_TICKER.LIVE REC
▸ Ryan Serhant Reveals How He Turns AI Agents Into Million-Dollar Sales Forbes · 21.07.2026▸ Chinese Tech Firms Pitch AI Agents as the Future of Smartphones citynewsservice.cn · 21.07.2026▸ Powerful AI Models Happily Being Given Away for Free RealClearMarkets · 21.07.2026▸ GitLab 19.2 Puts AI Agents to Work on the Security Backlog infoq.com · 21.07.2026▸ OpenAI’s GPT-Red: How AI Models Are Now Training Each Other to Be More Secure quasa.io · 21.07.2026▸ MCP update prepares AI agents for widespread deployment Techzine Global · 21.07.2026▸ New ENCFORGE Ransomware Targets AI Model Files in Langflow RCE Attack The Hacker News · 21.07.2026▸ In Depth: Chinese Smartphone-Makers Bet on AI Agents Caixin Global · 21.07.2026▸ WebMCP: Bridging the Gap Between AI Agents and the Web devmio · 21.07.2026▸ Boomi study finds AI agent trust lags enterprise adoption InsiderPH · 21.07.2026▸ Rapidus and Cadence partner to advance AI-driven SoC design New Electronics · 21.07.2026▸ Agentic AI and Smart Data: The Architecture of the UK’s New Open Finance Framework… Finextra Research · 21.07.2026▸ Databricks: US$188bn Valuation, Genie One and Agentic AI AI Magazine · 21.07.2026▸ Nvidia targets simulation bottlenecks with AI agent expansion InsiderPH · 21.07.2026▸ China Weighs Export Controls on AI Models, Including Open Weight LLMs trendingtopics.eu · 21.07.2026▸ Research says JadePuffer Ransomware wipes off data on AI Model Infrastructure Cybersecurity Insiders · 21.07.2026▸ Reported US push to ban Chinese AI models reflects anxiety over eroding tech hegemony… Global Times · 21.07.2026▸ Webinar: Can AI agents finally automate data testing? QA Financial · 21.07.2026
▮ MODEL_FEED.LIVE REC
▸ Nemotron-Labs-Audex-30B-A3B NVIDIA · 30B · MoE · 06.07.2026▸ Nemotron-Labs-Audex-2B NVIDIA · 2B · 06.07.2026▸ DeepSeek-V4-Flash-DSpark DeepSeek · Open language model · 27.06.2026▸ Qwen-AgentWorld-35B-A3B Qwen · 35B · MoE · multimodal · 22.06.2026▸ GLM-5.2 Zhipu · Open language model · 16.06.2026▸ North-Mini-Code-1.0 Cohere · coding · 05.06.2026▸ DeepSeek-V4-Pro DeepSeek · Open language model · 22.04.2026▸ DeepSeek-V4-Flash DeepSeek · Open language model · 22.04.2026▸ granite-4.1-8b IBM · 8B · 06.04.2026▸ granite-4.1-3b IBM · 3B · 06.04.2026▸ Mamba2-primed-HQwen3-8B-Instruct Amazon · 8B · instruct · 31.03.2026▸ Falcon-OCR TII · Open language model · 22.02.2026▸ tiny-aya-base Cohere · Open language model · 13.02.2026▸ tiny-aya-global Cohere · Open language model · 13.02.2026▸ GLM-4.7-Flash Zhipu · Open language model · 19.01.2026▸ Falcon-H1R-7B TII · 7B · 29.10.2025
▮ DISPATCHES / 16.04.2026 · 2 MIN READ

NVIDIA Wants to Be the Operating System for Enterprise AI Agents

NVIDIA Wants to Be the Operating System for Enterprise AI Agents

NVIDIA has never been content to sell picks and shovels when it can build the mine. At GTC 2026 in San Jose the company made its most significant software announcement in years — not a new chip, but a new platform: the NVIDIA Agent Toolkit, an open source suite designed to become the default infrastructure for building and running enterprise AI agents.

At its core is OpenShell, an open source runtime that enforces security, network, and privacy guardrails around autonomous agents. The pitch is straightforward: enterprises want to deploy agents, but their legal, compliance, and security teams are terrified of what an unsupervised AI might do when given access to internal systems. OpenShell is NVIDIA’s answer — a managed container for agent behaviour that lets organisations define precisely what an agent can and cannot do.

The partner list announced alongside the toolkit reads like a who’s who of enterprise software. Adobe, Atlassian, Cisco, CrowdStrike, Google, Microsoft, and LangChain — whose open source frameworks have been downloaded over a billion times — are all building compatibility with the platform.

Jensen Huang’s framing was characteristically bold. He described Claude Code and OpenClaw as having sparked the agent inflection point, extending AI beyond generation and reasoning into action. Employees, he said, will be supercharged by teams of frontier, specialised, and custom-built agents they deploy and manage.

What NVIDIA is really announcing is a bid for the enterprise agent stack — the layer of infrastructure that sits between the frontier models at the top and the business workflows at the bottom. If OpenShell becomes the default runtime, NVIDIA’s position in the agentic economy extends far beyond the GPUs that train and run the models.

For enterprises evaluating their agentic AI strategy, the toolkit is worth examining closely. The security and governance layer it provides addresses the single biggest objection most IT and compliance teams raise when agentic deployment comes up in conversation.

The infrastructure of the agentic era is being laid right now. NVIDIA intends to be a very large part of it.

END OF TRANSMISSION ▮ ◂ MORE DISPATCHES