This website uses cookies

Read our Privacy policy and Terms of use for more information.

Hi fam!

New models are getting cheaper and more capable, AI agents are starting to touch real-world systems, Google is thinking about putting AI infrastructure in space, and apparently we're now having serious conversations about how to keep AI from hacking things it wasn't supposed to touch.

This week, we're looking at what's actually happening in AI right now, one tiny trick you can use immediately, and a side hustle that doesn't involve selling 400 blank journals on Amazon.

The future of AI feels Like Magic, and it’s here!

The AI world has had a busy few days.

We've got a major new Claude model, AI potentially moving into orbit, companies asking for global AI standards, and one particularly uncomfortable reminder that giving AI access to real systems comes with real consequences.

Here are the stories worth knowing.

  • Anthropic launched Claude Opus 5.5 this week, positioning it as its newest high-end model for complex work, coding and long-running AI agents.

    The interesting part isn't just the performance.

    Anthropic says Opus 5.5 costs 40% less to run than Opus 5, with API pricing of $4 per million input tokens and $20 per million output tokens. The company also says the model went through external evaluations before release and includes stronger safeguards.

    Why this matters: AI models aren't just getting smarter. They're getting cheaper to use, which could make sophisticated AI agents much more practical for everyday work.

  • OpenAI has separately called for a US-led international coalition to develop common approaches to assessing AI capabilities and risks. Anthropic has also been pushing for stronger safeguards as its models become more capable.

    Why this matters: AI development is happening globally, but the rules governing it are still largely national. As models become more capable and interconnected, companies are increasingly arguing that AI safety can't be handled country by country.

  • The incident reportedly occurred in June and involved a public Medicare statistics portal that also contained non-public material.

    Why this matters: This is exactly why AI agents are a different proposition from chatbots. A chatbot can give you a bad answer. An agent with access to systems can potentially take an action.

    That makes permissions, monitoring and human review much more important as companies give AI more autonomy.

  • Google is taking one of those ideas that sounds like science fiction and turning it into an actual research project.

    Project Suncatcher is exploring whether large-scale AI computing could eventually happen in orbit. Google says satellites could take advantage of near-constant sunlight and potentially generate substantially more solar power than comparable systems on Earth.

    The company says a prototype satellite is scheduled for an initial orbital test to study how its Tensor Processing Units perform in space.

    Why this matters: AI's biggest physical problem is increasingly not just intelligence—it's energy, cooling and infrastructure. Google is asking whether the answer might eventually be above our heads.

The “One Level More Specific” Trick

AI has one very annoying habit.

It can take a perfectly reasonable idea and turn it into:

❝

“In today's rapidly evolving digital landscape…”

No.

Stop.

When AI gives you something generic, don't start rewriting the whole prompt.

Just tell it:

❝

“Make this one level more specific. Replace generic statements with concrete examples, actions, numbers, situations or details wherever possible. Keep the original meaning and tone.”

Example

Before:

❝

“AI can help businesses save time and improve productivity.”

After:

❝

“A small business can use AI to turn a 60-minute customer interview into a transcript, five social posts, a newsletter draft and three short-video ideas—without starting each piece from scratch.”

Much better.

And if it's still generic?

Ask for:

❝

“One more level of specificity.”

Sometimes that's all it takes to turn AI from corporate LinkedIn intern into something actually useful.

AI Social Media Makeover

Forget Amazon journals.

Forget AI colouring books.

Forget spending weeks creating a digital product that three people buy.

There’s a much simpler AI side hustle sitting right outside your front door:

Fixing terrible social media for local businesses.

And there are plenty of them.

A restaurant with great food but a terrible Instagram bio.
A hair salon whose last post was six months ago.
A dentist with fantastic reviews but blurry photos and no useful information online.

These businesses don't necessarily need a full-time social media manager.

They just need someone to make their online presence look better.

The offer

Don't sell:

“I provide AI social media services.”

Sell:

“I'll give your Instagram a complete makeover.”

For €100–€300, you could offer:

  • a rewritten bio

  • better profile positioning

  • 10–15 content ideas

  • 5 ready-to-use captions

  • Reel ideas

  • improved calls to action

  • visual recommendations

  • a simple content calendar

AI can help you do most of the research, brainstorming and writing.

Your value is knowing what actually needs fixing.

How to find clients

Open Instagram or Google Maps and look for local businesses with:

good products + bad social media.

That's the sweet spot.

Find 10 businesses where you can immediately see what could be improved.

Then send them a short message with three specific things you'd change about their profile.

Even better, create a quick example of how their bio or first few posts could look.

You're no longer sending:

❝

“Hi, I offer AI social media services.”

You're saying:

❝

“I looked at your Instagram and noticed three things I'd change. Here's what I'd do instead.”

Much easier conversation.

The math

You don't need hundreds of clients.

5 clients × €200 = €1,000

And once someone trusts you with their makeover, you can offer monthly content, Reels or full social media management.

The important part?

AI isn't the product.

The product is making a local business look better online—and saving the owner hours of work they don't want to do themselves.

Like Magic AI NFT 🏞️

Our master plan is to publish an NFT image in each newsletter and hand it out to our subscribers. The earlier you subscribe, the smaller the series are. It's a future collectible, a piece of digital art that captures the essence of this moment in time.

Thank you for being a valued subscriber. Together, let's embrace the magic of AI and creativity!

LMAI235-25092026

Was this email forwarded to you? Sign up here 👇

Reply

Avatar

or to participate