Welcome to the new internet

The next big thing in AI browsers?

May 30, 2025

Welcome back, everyone! Anthropic CEO Dario Amodei recently said AI could vaporize half of all entry-level white-collar jobs and spike unemployment to 10-20% in the next five years. One recent report found nearly 50% of all white-collar employees are using AI at work—and in many cases, doing more with fewer people.

I'm curious: Is AI job displacement something you're concerned about?

Click one to tell me more

Login or Subscribe to participate in polls.

Opera’s new browser wants to surf the internet for you

Opera

Opera just announced Neon, a new AI browser that promises to go beyond chat. The company says the browser will complete tasks like researching, designing, and coding automatically.

Automates, not just augments: Neon is designed to be a quiet background worker, not just a front-end interface. It uses built-in AI agents to handle multi-step tasks across the web. Need to book travel? It fills in the fields, compares prices, and checks availability. Need to generate a site? It writes the copy and spins up the design.

So when can try it out? The product is still in early stages. It’ll be part of Opera’s premium tier, and some of the more ambitious automations depend on cloud-based AI, which could limit performance and privacy. There’s also no clear release date, just a promise that it’s coming "soon."

Keep in mind: The Neon news follows The Browser Company’s recent decision to halt work on its Arc browser in favor of its new AI-centric Dia browser. Why? Arc was too complicated to go mainstream, sometimes pretty slow, and vulnerable to security issues.

The bigger picture: Plenty of bright AI minds are chasing agents, but few are embedding them this deep into the OS of the internet. If Neon works, it could shift browsers from passive windows to active operators. That changes who controls the flow of attention—and who gets to monetize it.

Here’s why Meta is restructuring its AI teams

Meta is breaking up its AI org. The company is reportedly separating its AI product and AGI research efforts into two distinct teams to move faster and compete more effectively with OpenAI, Google, and ByteDance.

A tighter feedback loop: Under the new model, Meta will have an AI Products team building user-facing tools like Meta AI, AI Studio, and in-app features for Facebook, Instagram, and WhatsApp. Meanwhile, an AGI Foundations team will focus on advancing core model capabilities like reasoning, voice, and multimedia through the Llama family and other research efforts.

Why it matters: Slow and steady doesn’t win the AI race. Meta is betting that greater flexibility and faster deployment will give it the edge.

xAI is paying $300M to put Grok inside Telegram

Elon Musk’s xAI inked a deal with Telegram that will bring Grok directly into the messaging app for 900 million users. The integration (which rolls out first to Telegram Premium users) will let people use Grok without leaving their chats.

What Grok can do in Telegram:

  • Offer writing suggestions, chat summaries, and quick document overviews

  • Summarize links and pull key takeaways inline

  • Generate stickers and creative responses directly in conversations

Why it matters: This is part of a growing shift toward invisible AI—tools that live inside the apps you already use. Instead of jumping to a separate chatbot, Grok becomes a native part of the messaging experience. It’s how AI becomes ambient, ever-present, and harder to opt out of.

P.S. It was a busy news week for all things AI—keep reading to get the best of the rest in headlines that move the needle.

Structure your prompts like a product manager

PromptShuttle

PromptShuttle is a centralized workspace for building, testing, and managing AI prompts. Instead of juggling messy docs or scattered notes, it lets you create reusable templates, track versions, and collaborate on prompt design in one place.

How you can use it:

  • Build a shared prompt library for customer support, marketing, or product ops

  • Version control prompts across teams to ensure consistent AI behavior

  • Streamline onboarding by giving new team members proven, pre-tested prompts

Pricing: Free for individuals. Paid plans for teams start at $29/month.

Pinpoint where users drop off instantly

Prism

Prism analyzes session replays using AI to detect friction, auto-summarize journeys, and surface what’s broken without you lifting a finger.

How you can use it:

  • Get AI-generated summaries of where users struggle

  • Push insights directly into GitHub, Linear, or Cursor

  • Track drop-offs, latency, and conversion barriers in real time

Pricing: Free for early teams; paid plans available

Jobs, news, announcements, and big ideas

  • Amazon strikes deal with The New York Times to license content for AI. Side note: The NYT is still in the middle of a copyright infringement lawsuit against OpenAI and Microsoft.

  • Grammarly has raised $1 billion in non-dilutive financing from General Catalyst to expand its AI offerings—the goal? Become a comprehensive productivity platform.

  • Sakana’s new AI rewrites its own code to get smarter with every run.

  • Hugging Face is building open-source humanoid robots for release next year. The two new models are HopeJR (a full-size humanoid robot with 66 independent movements) and Reachy Mini (a desktop unit that can move its head, talk, listen, and test AI apps).

  • Black Forest Labs debuts FLUX.1 for next-gen image editing with AI context.

  • Perplexity Labs is live for Pro users to tackle more complex AI tasks.

  • ManusAI just dropped a tool that builds your entire slide deck on command.

AI didn’t start with ChatGPT. Follow along as we go through the history of AI.

That’s a wrap! See you next week.

—Matt (FutureTools.io)

P.S. This newsletter is 100% written by a human. Okay, maybe 96%.