Here’s what’s shaping AI right now - without the noise:

Google Labs has turned Stitch from a UI creation experiment into something much bigger: a full AI-native design platform built around an infinite canvas that works across images, code, and text in one place.

The headline feature here is not just prettier design generation. Stitch now includes a design agent that can understand how a project changes over time, plus an agent manager that lets users explore multiple directions in parallel. Google is also adding voice interaction, so creators can literally speak to the canvas and make updates conversationally instead of manually clicking through every step.

That makes Stitch feel less like a design helper and more like a creative operating system for building interfaces. Instead of bouncing between mockups, notes, and code, the workflow starts to collapse into one AI-assisted environment.

You don't have to read everything — just the right thing. 1440's daily newsletter distills the day's biggest stories from 100+ sources into one quick, 5-minute read. It's the fastest way to stay sharp, sound informed, and actually understand what's happening in the world. Join 4.5 million readers who start their day the smart way.

Join for free today!

Manus AI is preparing to take its always-on autonomous agent beyond Telegram by adding WhatsApp Business integration.

Clues inside the Manus web app suggest the setup will work much like WhatsApp Web: users scan a QR code, connect their account, and then link the official Manus bot. One important detail: this won’t work with a regular WhatsApp profile alone — users will need WhatsApp Business to complete the connection.

That matters because Manus isn’t just another chatbot. Its pitch is an always-on agent that can handle multi-step work like research, data processing, task execution, and content creation. And if this lands inside WhatsApp, the implications go well beyond messaging. Manus co-founder Tao Zhang has already said support for WhatsApp, LINE, Slack, and Discord is coming soon, which points to a much broader push into communication platforms people already use daily.

The bigger angle is hardware. Since Ray-Ban Meta smart glasses already support hands-free WhatsApp messaging, a WhatsApp-native Manus experience could eventually let users trigger a far more capable agent through voice alone — potentially making it more useful than the default Meta AI experience on the glasses.

At GTC 2026, Nvidia made its position clear: more compute still drives better AI. Jensen Huang’s message was straightforward — more intelligence creates more value, and more value creates more revenue.

But Nvidia also used the event to push a new idea: agentic scaling. After pretraining scaling, post-training scaling, and test-time scaling, the company is now arguing that the next major leap will come from AI systems talking to other AI systems. In this model, agents coordinate with subagents, workloads become multi-step and multi-model, and the demand for faster inference and larger context windows grows dramatically.

Nvidia says this shift will require systems that can serve tokens far faster and support much larger models, including trillion-parameter-class systems. The company is effectively making the case that the next scaling law is not about a single model getting smarter on its own, but about entire swarms of agents working together.

At the same time, a very different vision is gaining ground. Near co-founder Illia Polosukhin is championing user-owned AI — a model where user data stays separate from the model, agents operate more like a secure independent OS, and people keep tighter control over how AI works on their behalf. Near is also leaning into the agent wave with IronClaw, its more security-focused take on the OpenClaw movement, alongside a secure agent marketplace built for delegation and monetization.

So the bigger story coming out of GTC is not just that Nvidia wants bigger AI. It’s that the industry is starting to split into two camps: one chasing massive centralized compute, and another trying to build safer, user-controlled agent ecosystems from the ground up.

A few tools quietly worth exploring:

AI is evolving fast.

But the real edge isn’t knowing everything.
It’s knowing what actually matters.

See you next issue.