Dr. Dennis Weekly AI Blog - March 31, 2026

Happy New Week, Everyone!

Ever walked into a Monday morning meeting and wished you could just send a hologram of yourself while you grab another coffee? Well, as of this week, that's not just a sci-fi fantasy anymore—it's an actual product update. Welcome to a week where AI didn't just get smarter; it started taking over our screens, our meetings, and even our national security debates.

The Trend We're Loving

The Era of "Agentic" AI is Officially Here

We've spent the last three years talking to AI. Now, AI is talking to our software. The biggest trend this week isn't a new chatbot; it's the explosion of Agentic AI—systems that can navigate your computer, click buttons, and execute multi-step workflows autonomously.

Why does this matter? Because the bottleneck in business isn't a lack of ideas; it's the friction of execution. When Anthropic announced this week that Claude can now control your computer directly from your smartphone, it signaled a massive shift. We are moving from AI as an "advisor" to AI as an "executor."

Key Implications:
• The Death of the "Swivel Chair" Workflow: You will no longer need to copy-paste data between five different SaaS apps. AI agents will do the clicking for you.
• Infrastructure is Ready: The Model Context Protocol (MCP)—the plumbing that lets AI talk to your tools—just crossed 97 million installs. This is now standard enterprise infrastructure.
• Security is the New Frontier: Giving AI the keys to your computer means we need entirely new frameworks for access control and oversight.

Further Reading:

• Anthropic gives Claude computer access — from a mobile device

• March 2026 AI Roundup: The Month That Changed AI Forever

Deep Dive

OpenAI Pulls the Plug on Sora

Remember Sora? The mind-bending AI video generator that OpenAI teased late last year, which prompted Disney to sign a billion-dollar licensing deal? Well, in a stunning reversal, OpenAI just shut it down.

It turns out that generating hyper-realistic, cinematic video requires an eye-watering amount of computing power. The economics simply didn't make sense at scale. OpenAI realized that the cost per generated minute was "economically irreconcilable" with what users were willing to pay. The Disney deal is dead, and OpenAI is pivoting its massive compute resources back to enterprise coding tools and agentic workflows.

This is a massive reality check for the generative AI space. It proves that just because a technology works brilliantly in a research lab doesn't mean it's a viable commercial product. The laws of physics—and server costs—still apply.

Anthropic's High-Stakes Standoff with the Pentagon

If you want to know who holds the power in the AI era, look at the courtroom. Anthropic, the company behind Claude, just went toe-to-toe with the U.S. Department of Defense. The Pentagon wanted to deploy Claude in classified systems without Anthropic's standard safety guardrails—specifically, the rules against mass domestic surveillance and fully autonomous weapons.

Anthropic's CEO Dario Amodei drew a hard line and refused. In retaliation, the government labeled Anthropic a "supply chain risk" and banned federal agencies from using Claude. But Anthropic fought back, and a federal judge just granted a preliminary injunction blocking the ban, calling the government's move an attempt to "cripple" the company.

This isn't just a legal drama; it's a defining moment for AI governance. We are watching private tech companies dictate terms of engagement to the world's most powerful military. Meanwhile, Anthropic's enterprise revenue is reportedly surging to a $20 billion annual run rate. Standing your ground, it seems, is good for business.

Zoom Wants to Attend Your Meetings For You

We've all suffered through meetings that could have been an email. Zoom's new solution? Send an AI avatar instead. This week, Zoom announced photorealistic AI avatars that can attend video calls on your behalf. You feed it your talking points, and a digital clone—complete with your facial expressions and lip movements—delivers them to your colleagues.

It's a fascinating, slightly dystopian feature that raises a glaring question: If a meeting is so unimportant that you can send a bot, why are we having the meeting at all? You're essentially swapping the burden of attending a meeting for the burden of reading a summary of a meeting your avatar attended.

Zoom is simultaneously shipping deepfake detection tools alongside these avatars, creating a bizarre ecosystem where AI is pretending to be you, while another AI tries to catch people pretending to be someone else. Welcome to the future of corporate communication.

Dr. Dennis' Picks

🔥 GPT-5.4 (OpenAI) – The new flagship model available in Standard, Thinking, and Pro variants, featuring native computer use and a 1-million-token context window. Try it here

 🔥 Gemini 3.1 Ultra (Google) – A massive leap in native multimodal reasoning, processing text, image, audio, and video simultaneously without transcription intermediaries. Try it here

 🔥 Mistral Small 4 (Mistral AI) – A wildly efficient, open-source 22B parameter model that beats closed models twice its size, perfect for enterprise deployment. Try it here

 🔥 Voxtral TTS (Mistral AI) – A lightning-fast, open-weights text-to-speech model that runs on edge devices and produces incredibly lifelike voice agents. Try it here

 🔥 Grok 4.20 (xAI) – The ultimate real-time research tool, using a multi-agent debate system to drastically cut hallucinations on breaking news and current events. Try it here

Takeaways

The AI hype cycle is maturing into the AI execution cycle. We are moving past parlor tricks and into hard economics, as seen with Sora's shutdown and the rise of autonomous agents.

Insight: "The bottleneck in business is no longer a lack of ideas; it's the friction of execution. Agentic AI removes the friction."

Action: Pick one repetitive, multi-step digital task you do weekly and test if an AI agent can automate it.

Next week: We'll dive into how the new state-level AI transparency laws will force a complete rewrite of your digital marketing strategy.

Forward this to colleagues who need to stay ahead of the AI revolution. Have insights to share? Reply – I read every message.