Gemini comes for the mouse pointer

Google wants the pointer to do more

Google DeepMind is testing an AI-enabled mouse pointer that can understand the thing under your cursor and respond to short voice commands.

The demo uses Gemini to let people point at a PDF, image, table, code block, or map and ask for help without writing a long prompt. Google says the same ideas are moving into Chrome, a new Googlebook laptop experience, and Google Labs’ Disco.

This feels like the right direction because most AI tools still make the user do too much setup before anything useful happens. A cursor with context could make AI feel closer to the work on the screen, especially during small tasks at a desk with five tabs open.

The risk is accuracy. When the command is “move that” or “fix this,” the system has to know exactly what “that” means.

The real test is whether this saves time in normal work, or just looks clean in a shortened demo.

Broadband support gets an AI layer

Nokia wants telecom operators to use AI agents before a broadband problem turns into a customer complaint.

The company launched agentic AI features across its Altiplano, Corteca, and Broadband Easy platforms, drawing on experience from more than 600 million deployed broadband lines. Nokia says the tools can lift first-contact helpdesk resolution above 50%, qualify network incidents within 5 minutes, and cut return site visits by 50%.

That matters because broadband support is still full of slow calls, truck rolls, and blinking routers in quiet living rooms. I think Nokia is making a practical bet: operators will pay for AI that reduces labor pressure and keeps customers from leaving.

The bigger signal is control. Nokia is pitching an open setup where telecom providers can bring their own LLMs, interfaces, and data sources.

That sounds right for carriers, who want automation without handing over too much of the network stack. The question is how much autonomy operators will actually trust when real customers are offline.

Design is moving closer to production

Dessn is betting that AI design work should happen on a real codebase, not in a separate mockup file.

The startup raised $6 million from Connect Ventures, Betaworks, and N49P, and says teams at Color, Wispr, and Mercury are already using it. Its tool runs a company’s codebase in the cloud, so designers can prompt changes against production-like software and share links without asking a developer to set things up.

I think this is a smart wedge because Figma still leaves a gap between what a designer imagines and what a developer can ship. Dessn is aiming at the messy middle, where a product screen exists, the code is real, and someone wants to try a change before lunch.

That also narrows the market. It is less useful for blank-page ideation and more useful for teams with active products, enough code, and enough design volume to care.

The question is whether production fidelity becomes the new default, or just another paid layer in the design stack.

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.