Google folds AI into every tab

Chrome puts AI next to every page

This changes how you move through tabs.

Google rolled out an AI Mode update in Chrome that opens webpages side-by-side with AI, so you can read, ask questions, and compare without leaving the page.
It also lets you pull in context from recent tabs, images, and PDFs through a plus menu, and it’s live in the U.S. starting April 2026.

I think this is Google tightening its grip on the browsing layer by keeping users inside a guided flow where search, reading, and questions all happen in one place for longer stretches of time.
The side-by-side view sounds simple, but it shifts power toward the assistant that interprets pages while you scroll, especially when it can reference multiple open tabs at once.

You can hear the quiet click of tabs opening less often. Publishers may see less direct traffic if answers get resolved in that pane, and comparison shopping becomes mediated by whatever the AI chooses to surface and emphasize.

It will stick. If this becomes default behavior, what does a “visit” to a website even mean anymor

Luma is trying to make films, not tools

This is a shift from software to output.

Luma launched Innovative Dreams, a production company with Wonder Project, starting with a Moses film on Prime Video starring Ben Kingsley.
The pitch is real-time filmmaking using AI agents that can adjust sets, lighting, and even faces during production instead of waiting for post.

I think this is the more serious move for AI video companies, because building tools is crowded but owning actual content gives them leverage and proof that the tech works under pressure.
The real-time angle matters since it cuts one of the most expensive parts of filmmaking, but it also puts AI directly in creative decisions on set.

You can picture a green screen and a laptop open nearby. If this works, studios may start backing smaller, faster productions instead of betting $100 million on a single film, especially when some founders are already pushing for making 50 films instead.

Will audiences notice or care how it was made?

NVIDIA wants your AI agent running locally

Running an AI agent now means running a small system.

NVIDIA published a guide to deploy a local, always-on AI agent using NemoClaw, OpenClaw, and a 120B parameter Nemotron model, all running on a DGX Spark machine.
The setup takes about 20–30 minutes plus an 87 GB model download, and connects the agent to tools like Telegram while keeping everything on-device.

This is NVIDIA pushing a clear idea: serious agents should run close to the user, with control over data, runtime, and permissions instead of relying on hosted APIs.
The security layer matters more than the model here, since the whole pitch depends on sandboxing an agent that can execute code and call tools without exposing the system.

You can hear the fan spin up. If this approach sticks, developers will start treating AI agents less like chat features and more like long-running services that sit on machines and do work over time.

How many people actually want to maintain that setup themselves?

Prohuman team

Free Guides

Explore our free guides and products to get into AI and master it.

All of them are free to access and would stay free for you.

Feeling generous?

You know someone who loves breakthroughs as much as you do.

Share The Prohuman it’s how smart people stay one update ahead.