BIG TECH

Tech giants play nice with agents

Major AI firms are joining hands on the industry’s biggest trend: Agents. 

On Tuesday, a coalition of companies, including Anthropic, OpenAI, Google and Microsoft, announced the launch of the Agentic AI Foundation, a group dedicated to developing open source standards for agents. The group is overseen by the Linux Foundation, the tech nonprofit in charge of the Linux operating system.

The Agentic AI foundation will initially focus on three existing tools, donated by major tech firms: 

• Model Context Protocol (MCP) is an open source standard developed by Anthropic that enables agents to communicate with applications. Anthropic said in a press release that it’s donating MCP to the foundation as part of its founding to “ensure these foundational technologies remain neutral, open, and community-driven;”

• Agents.md, an open format by OpenAI that provides agents with instructions and context for coding tasks; and 

• Goose, an open source AI agent framework invented by Block that runs locally on a user’s device.

“Bringing these projects together under the AAIF ensures they can grow with the transparency and stability that only open governance provides,” said Jim Zemlin, executive director of the Linux Foundation, in a statement. 

The foundation aims to solve a major problem that agents are facing as companies seek to scale them: fragmentation, Alex Salazar, CEO and Co-founder of Arcade.dev, a founding member of the Agentic AI Foundation, told The Deep View. 

With organizations building agents in entirely separate ecosystems, it’s difficult to make them work together, which in turn decreases the value an enterprise can get from them. While the popularity of Anthropic’s MCP was a step in the right direction, the foundation can capitalize on that momentum. 

“Enterprises to date have been forced to manually stitch together and create all the pieces necessary to build useful AI agents, which slows innovation and opens up risk,” Salazar said. “Standards removes that responsibility from the individual developers and means more development can be reused safely.”

The establishment of this foundation is a prime example of a rising tide lifting all boats. While several of these firms are in competition to get enterprises to adopt their agentic systems, solving one of the tech’s primary problems – a lack of interoperability due to disparate systems – is a win-win-win, for the model developers, the enterprises and open source technology advocates. Though open source tech typically comes with its own set of problems, like a struggle to keep up with proprietary tech and a lack of funding, the involvement of some of the biggest players in AI could potentially help alleviate those issues.

TOGETHER WITH WIZ

7 Security Best Practices for MCP

See what security teams are doing to protect MCP.

As MCP (Model Context Protocol) becomes the standard for connecting LLMs to tools and data, security teams are moving fast to keep these new services safe. 

This free cheat sheet outlines 7 best practices you can start using today, including:

 ✅ How to lock down MCP servers and supply chains
 ✅ Enforcing least-privilege access for tokens and tools
 ✅ Adding human-in-the-loop safeguards for critical actions

If you’re starting to see MCP show up in your environment, this is a great place to start.

Get your free MCP Security Cheat Sheet right here.

PRODUCTS

US Military gets its own chatbot

War plans are about to contain a lot more em dashes. The US Military now has its own Google Gemini-powered AI platform.

“The future of American warfare is here, and it’s spelled A-I,” Department of War Secretary Pete Hegseth said on Tuesday in an announcement video that totally wasn’t written by Gemini. The platform, which will only be used for unclassified processes, can be used for deep research, document formatting and video or image analysis, Hegseth said.

The site — available at GenAI.mil but only if you’re reading this from the Pentagon — is the fruit of a $200 million contract Google signed with the then-Department of Defense over the summer to grow the agency’s AI capabilities. It’s also a win for Google; however, xAI, OpenAI, Anthropic and Scale AI have signed military contracts for up to $200 million each this year. An earlier version of the site contained slots for ChatGPT, Claude and Grok, The Information reported.

Reading between the lines, the tool seems to be a custom Gemini interface with security measures baked in. Usage on the platform won’t be utilized to train Google’s public models, the company said. And while it’s hard to imagine DoW employees weren’t already using generative AI for at least some unclassified tasks, the site’s release can be read as a signal for a vibe shift in light of President Trump’s AI action plan from July, which specifically called for AI adoption within the Armed Forces. 

That being said, the military had already been shelling out for AI products before the chatbot’s release. Palmer Luckey’s Anduril contracts with the military to put AI “Directly into the Warfighter’s Helmet.” And the software company Palantir delivered AI-enabled TITAN trucks for the US Army earlier this year.

As silly as it may sound, the Department of War is actually a huge enterprise contract for Gemini. There are over 3 million active and civilian military personnel, making the US military one of the largest employers in the world. The Information reported that other chatbots were listed as coming soon on an early version of the military AI site, but Gemini may have a moat in winning the DoW’s business. In a press release, the agency said its site would reduce AI hallucinations by using Google Search.

TOGETHER WITH ORACLE NETSUITE

KPI Checklists: A Practical Step-by-Step Guide

Are you trying to devise, implement, or fix your current KPIs?

Using brief explanations and practical checklists, this book provides the tools to:

• Deliver meaningful measures that work

• Create reports that support decision-making

• Deploy the necessary tools to engage your organization.

Get your copy now

POLICY

Trump clears Nvidia's second-best chips for China

More advanced Nvidia chips are set to be sold in China. For now.

Earlier this week, President Trump cleared Nvidia to ship H200s, the country’s second-best chips, to China. In exchange, the United States would collect a 25% sales tax. 

After Trump’s Truth Social post went live, The Financial Times reported that China would limit access to Nvidia chips to bolster its domestic chipmaking industry. Then, Bloomberg reported that Trump decided to clear Nvidia chip sales after learning that the Chinese tech giant Huawei had chips that could perform similarly to the H200. 

It’s the latest in an ongoing back-and-forth between two countries with conflicting motives. The Trump administration has argued that selling no chips to China risks helping the nation’s domestic chip industry. At the same time, the US will continue to ban sales of the most advanced Blackwell chips, which are reserved for US firms. Such a policy could thread the needle between giving the US a leg up in AI and spurring demand for Chinese chips.

Beijing has roughly the opposite calculus: it wants to spur its domestic chip industry, but the country’s AI sector would likely benefit from access to the H200s, which are almost six times as powerful as the next most powerful Nvidia chips China has access to, per Reuters. 

The H200s will also reportedly travel from Taiwan to the US for a “national-security review” before returning to China to avoid a Constitutional ban on export taxes.

Trump’s decision to lift the ban on H200 sales in China is a win for Nvidia CEO Jensen Huang, who seems to be having his cake and eating it, too, by selling chips in both the US and China. However, Nvidia’s victory could be soured by Beijing limiting access to the chips or pushing Chinese firms to use the reportedly pretty powerful Huawei chips instead.

LINKS

• Sam Altman reportedly prioritizes ChatGPT’s popularity over moonshots

• Meta pursues a new frontier AI model, called Avocado, that may be proprietary

• OpenAI appoints former Slack CEO Denise Dresser as Chief Revenue Officer

• Identity startup Saviynt raises $700 million for AI security, access tools

• EU will probe Google for the use of online content in AI training

• Device startup Pebble introduces $75 smart ring

• ChatGPT nears 900 million weekly active users

• FLORA: A pro-grade creative platform for speed, precision, and collaboration. First 50 users get 50% off their first month of the Agency Standard Tier with code DEEPVIEW (monthly plan only). Claim yours now. (sponsored)

• ChatGPT and Instacart: Brainstorm meal ideas, make grocery lists and check out without leaving the ChatGPT interface. 

• Cosmic Agents: Autonomous assistants to build features, fix bugs and generate content.

• Questom: An agent for sales, handling inbound calls, emails, charts and texts, collecting order details and generating quotes.

• Snowflake: AI Applied Scientist, Code Intelligence 

• Snyk: Staff Developer Advocate

• Scale AI: Forward Deployed AI Engineer, Enterprise

• Luma AI: Research Scientist/Engineer – Data

GAMES

POLL RESULTS

Would you pay more for an ad-free AI chatbot?

• Yes, take my money (26%)

• No, I’ll tolerate ads (25%)

• I’d just switch to a competitor (49%)