Hi Fam! 👋
We’re making a little change around here.
From now on, Like Magic AI is moving to one newsletter a week — every Wednesday.
Because, honestly, less is more. We’d rather bring you one really good dose of AI each week than fill your inbox just for the sake of filling it.
So Wednesdays are now our thing. 🪄
Expect the AI news worth knowing, tools actually worth trying, useful tips, interesting experiments and a little AI inspiration — all in one place, once a week.
One email. More value. Less inbox noise.
See you Wednesday. ❤️
The future of AI feels Like Magic, and it’s here!
This week, AI is looking less and less like a chatbot you open when you have a question — and more like something that can actually work alongside you.
Agents are getting more autonomous, AI is moving into the tools we already use, and businesses are starting to hand over entire workflows instead of individual tasks.
Here are five things that caught our attention this week. 👇
OpenAI introduced Dots, a new type of AI assistant designed to work more
autonomously on ongoing tasks and projects.
Instead of simply answering a question, a Dot can work toward a goal, interact with tools and applications, and keep working in the background.
OpenAI is positioning Dots as customizable assistants for things like software development, research and content creation.
The bigger idea is simple:
You don't just chat with AI anymore. You delegate to it.
Why this matters: We're moving from Prompt → Answer to Goal → Action → Result.
Anthropic has released Claude Sonnet 5.5, its latest upgrade to the Sonnet family.
The company says the model is 30% faster and costs up to 30% less for most work.
That combination is becoming increasingly important.
AI companies have spent years competing on intelligence. Now another battle is becoming just as important:
How much intelligence can you get per dollar — and how quickly can you get it?
Why this matters: The best AI workflow isn't necessarily the one using the smartest model. It's the one using the right model for the job.
Meta is expanding Muse, its personal AI agent, into the small-business world.
Muse for Small Business can connect to tools including Canva, Shopify, Slack, Notion, Dropbox, Zoom, Stripe, Asana, Figma and Instagram/Facebook business accounts.
Instead of giving the AI a series of individual prompts, you can give it a goal and let it work across your connected tools.
Why this matters: This is where AI starts looking less like a chatbot and more like a junior operations team. For small businesses especially, one person can potentially accomplish work that previously required several different tools — or several different people.
Microsoft has redesigned Copilot around three areas: Home. Code. Autopilot.
The interesting one is Autopilot — a persistent, proactive AI agent designed to keep working toward tasks even when you're not actively interacting with it.
That's a significant change from the traditional chatbot model.
Why this matters: The major AI companies are increasingly converging on the same idea: AI isn't just a place where you ask questions. It's becoming a coworker that can stay with the task.
This one is especially interesting for anyone working with video. Runway has brought its AI tools directly into Adobe Premiere Pro and After Effects, so creators can use Runway without constantly jumping between applications.
The integration gives creators access to tools for generating images and video, transforming existing footage and working with AI-generated material directly inside their Adobe workflow.
Why this matters: AI video is moving from: “Let's try this cool AI video tool.”
to: “Let's use AI inside the software we already use.” And that's a much bigger deal for professional creators. You don't necessarily need to change your workflow. The AI comes to you.
The Context Dump
One of the easiest ways to get dramatically better results from AI is to stop giving it tiny pieces of information.
Instead:
Dump the context.
If you're asking AI to write something, don't just say:
“Write a LinkedIn post about our new product.”
Give it:
what the product does
who it's for
what problem it solves
the tone you want
an example you like
what you absolutely don't want
the key information that must be included
where the post will appear
Then ask it to write.
AI doesn't magically know what is inside your head.
Give it the material first.
A useful prompt structure:
Context: Here's everything you need to know.
Goal: Here's what I want to achieve.
Audience: Here's who this is for.
Constraints: Here's what you should and shouldn't do.
Output: Here's exactly what I want back.
More context doesn't always mean a better answer.
But relevant context almost always beats vague instructions.
The 10-Minute AI Assistant Setup
You don't need to build an AI agent to have your own little AI assistant. Pick one repetitive task you do every week and teach AI how you like it done.
Let's say you write client emails.
1. Give it 3–5 examples
Paste in emails you've written that sound like you.
2. Explain your rules
Tell it things like:
Keep it friendly and direct. Keep it short. No corporate language. Don't change my voice.
3. Give it the job
Try:
“From now on, when I give you a rough email, rewrite it in this style. Keep the meaning and ask me if anything important is missing.”
4. Test it.
Give it a real email and correct anything it gets wrong.
5. Save the prompt.
Next time, you don't start from zero. Paste the new email and let your AI assistant do the first draft.
Do this for emails, social posts, proposals, meeting notes or research and you've got a tiny assistant trained around the way you actually work.
Ten minutes of setup can save you dozens of repetitive decisions later.
From the Archive: SARA AI: The Personal AI Assistant That Might Just Replace You (In a Good Way)
Remember when having a personal assistant sounded like something reserved for CEOs?
In March 2025, we tested SARA AI, a personal AI assistant designed to help with everyday tasks, productivity and getting things done.
The idea was simple: instead of using AI only to answer questions, you give it a task and let it help you actually get the work done.
And that's interesting to revisit now — especially because so much of AI has moved toward agents and assistants that can take action rather than simply generate an answer.
Back then, it looked like an interesting idea. Today, it looks a lot more like where AI is heading.
Text generation | Image Generation | LMAI recommends |
|---|---|---|
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!
LMAI236-30092026
Was this email forwarded to you? Sign up here 👇
