In partnership with
Did you know that we have LinkedIn, Instagram and X accounts
that you can follow?
Hi {{first_name|everyone}},Meta’s new Muse Wonder models are built to remember what you’re working on and keep going when a task gets messy, and the privacy side of that is worth a look. Next, there’s an AI planner from our friends that calls you out when you procrastinate, a few tools worth bookmarking, and an entry to Higgsfield that was made entirely with AI. Then we finish with an AI that solved several long-open math problems and a new way to stop shopping agents from buying the wrong thing, so let’s dig in.
🤿 DEEP DIVE
Muse Wonder Is Built To Finish The Job
Meet Muse Wonder, built to remember and keep going.
Meta seems pretty interested in what happens after an AI gives you an answer.
Because answering a question is one thing. Actually helping someone sort out an insurance claim, understand a lease, or build a working app is a different kind of job. And that's basically what Meta is going after with its new Muse Wonder models.
The company just introduced Muse Wonder and Muse Wonder Full, which it calls its most capable AI models yet, following a 14-month overhaul of its AI efforts.
And while Meta is making the usual big performance claims, I think the more interesting part is how it wants these models to behave when they're actually doing work.
Remember what matters, then keep going
There are two features that explain a lot of the thinking here.
The first is something Meta calls Active Memory.
Basically, the AI is designed to remember relevant information about the user across conversations and bring it back when it's useful. So rather than treating every conversation as a completely separate interaction, it can carry important context into whatever the user working on next.
The second is Active Focus, and this is probably the more interesting one.
Meta says the models are designed to keep working toward a goal even when things get difficult. If an approach doesn't work, they can try another. If something goes wrong, they can recover. And if they need clarification, they can ask instead of just guessing.
They’re also supposed to check with the user before following through on anything important.
Put those two ideas together and you get a pretty clear picture of what Meta is trying to build. An AI that remembers enough to understand the task and stays focused long enough to work through it.
Not just something that produces a convincing response and calls it a day.
But there's a privacy problem to solve too
There's another detail worth paying attention to.
Meta explicitly talks about these agents working across personal messages, calendars, and documents. Which makes sense for the kinds of tasks it's describing, but also means the AI could be handling some pretty sensitive information.
The company says it has trained the models to avoid sharing private information without permission, refuse assistance with harmful acts, and help security teams identify weaknesses in their own systems.
A fuller safety evaluation report is still coming, though, so the announcement doesn't give us the complete picture yet.
Who actually gets it?
Muse Wonder is available now through the Meta Model API and Muse Code, while Muse Wonder Full is available to Muse Code subscribers.
The models are also expected to power more experiences inside Muse, with subscribers getting access first.
So the bigger personal AI experience Meta is describing isn't fully rolled out yet.
But the direction is fairly clear. Meta isn't just talking about AI that knows more things. It's talking about AI that can remember what you're doing, work through problems, and actually follow a task to completion.
And honestly, that distinction is the whole story.
The quick take
Meta has introduced Muse Wonder and Muse Wonder Full, claiming state-of-the-art performance on personal and professional agentic tasks.
Active Memory lets the models carry relevant information across conversations.
Active Focus is designed to help them persist through difficult tasks, recover from errors, and ask for clarification when needed.
Availability is still limited. Developers and Muse Code users get access first, with broader Muse experiences coming later.
Safety is part of the pitch. Meta says the models are trained to protect private information and refuse harmful requests, with more evaluation details to follow.
🤝 FROM OUR FRIENDS
A quick share we thought you might find useful
Your AI planner that calls you out when
you procrastinate
You drag six tasks into your Monday morning and finish one by lunch. The rest got buried under email, Slack, and a meeting that could have been a message.
Vulpie AI is a desktop task planner for Mac and Windows with an AI fox companion that actually watches whether you're still working. Pick a task, estimate how long it'll take, and an overlay timer appears on top of every window. You can't close it until the task is done.
Set the fox to Savage mode and it'll tell you «you said 20 minutes, it's been 48» with zero interest in your excuses. After a week of seeing your «quick 10-minute» email sessions clock in at 25, you start entering honest estimates and your day lines up with what you planned.
Full task board with subtasks, recurring tasks, Google Calendar and Gmail sync, Notion integration, and morning briefings based on your calendar gaps.
Pro plan is $4.99/month billed yearly.
Also there’s a mobile version for Android that staright up blocks Instagram, YouTube and whatever else you waste time on, so now the fox controls your phone too.
⚒ TOOL SNAPSHOTS
Futuristic tools within AI, no-code, and productivity
Search across text, images, audio, video, and code in one place.
Why it’s useful: Handy for developers building multimodal search into apps without relying on cloud infrastructure, since the model is small enough to run directly on a device.
Turn rough ideas and screen recordings into videos worth sharing.
Why it’s useful: Makes it easier to create product demos, tutorials, and explainers without spending hours editing footage or putting together animations yourself.
Make better use of all the things you save and read.
Why it’s useful: Good for anyone collecting articles, videos, and research who wants to find connections and turn that knowledge into useful work instead of letting it sit forgotten.
🤖 AI MADE THIS
Interesting and inspiring creations made with AI
An AI researcher creates fake worlds to harvest real emotions in a film made entirely with AI
THE PROMPTER is an AI-made film about an AI researcher creating artificial worlds to harvest real human emotions, which makes the production itself part of the joke. Every image, voice and sound was AI-generated using tools including Seedance 2.5, Nano Banana Pro, GPT Image 2, Suno and ElevenLabs.
👀 ON OUR RADAR
One more thing we thought you'd like
Your buyer asked ChatGPT. It named your competitors.
Search didn't die. It moved.
Your buyers ask ChatGPT, Gemini, Claude and Perplexity what to buy, and act on the answer. If you're not in it, a competitor is.
ChatFeatured does the work to get you cited. Most customers see their first citations inside 7 days.
⚡ QUICK HITS
The latest in AI, tech, and productivity worth knowing
DeepMind’s AlphaProof Nexus solved nine of 353 formalized Erdős problems, including two that had been open for over 50 years. What’s interesting is that it produces proofs checked by mathematical software, rather than relying on humans to verify AI-generated explanations. Although an independent audit later put the newly solved count at eight.
American Express introduced a framework that checks whether an AI agent’s purchase matches what the customer actually authorized. So if you ask an agent to buy white Nike sneakers but it tries ordering Adidas instead, the system could block the transaction. The idea is to catch these mistakes before payment goes through.
ℹ️ ABOUT US
The Intelligent Worker helps you to be more productive at work with AI, automation, no-code, and other technologies.
We like real, practical, and tangible use-cases and hate hand-wavy, theoretical, and abstract concepts that don’t drive real-world outcomes.
Our mission is to empower individuals, boost their productivity, and future-proof their careers.
We read all your comments - please provide your feedback!
Did you like today's email?
What more do you want to see in this newsletter?
