Meta surprises everyone with an AI device for Muse

Welcome back, Superhuman. Zuck’s bringing Tamagotchis back in style, and this time, they can talk. Meta just surprised everyone by becoming the first company to debut an AI-powered device — beating OpenAI & Anthropic to the punch — as its unexpected turnaround story continues.

Today: ChatGPT Voice gets active, learn how to build an AI IT helpdesk in 15 minutes with StackAI, and get the latest prompts and trending social posts.

1. Meta unveils a Muse device, VR glasses, and more, at Connect 2026: Day one at Meta’s annual conference brought Muse Charm, a Tamagotchi-style, pocket-sized gadget powered by Meta’s Muse personal agent. You can see it in action here. The company also unveiled its Meta VR glasses that weigh one-fifth as much as its Quest 3 headset and can turn any flat surface into an interactive keyboard, plus an always-available agent for its line of smart glasses. Here’s a full rundown of everything Meta announced.

2. A ChatGPT Voice upgrade lets it take action through voice prompts: ChatGPT Voice now runs on GPT-6 and can access your email, calendar, and Slack. This upgrade lets you verbally prompt ChatGPT Work from your phone or computer to complete tasks across the internet. The agent can send emails, build websites, search for information, contact customer support, and more. Watch how the new update works.

3. Anthropic’s Claude may have made its first major scientific discovery: Anthropic claims Claude discovered a previously unknown enzyme system that shares similarities with CRISPR after 950 Claude agents worked for 21 hours. The lab says the enzyme system’s exact role is unknown and further research is needed to determine if it has potential benefits. For more details, here’s what CEO Dario Amodei had to say.

Most AI tools hand you an agent and an empty chat box, then call that automation. It isn't.

With Bolter, you start from a blueprint built by operators like you, or simply describe the process in plain words, like you'd brief a new hire. Bolter runs it for you.

Why you should use Bolter:

Get early access (plus 1 month at no cost for Superhuman readers).

Sample Prompt: “I’m getting a “VPN connection failed” error on my work laptop after the latest update. Walk me through the troubleshooting steps in order. Use our internal IT documentation where relevant, cite the policy or guide you’re relying on, and tell me when I should escalate the issue to IT.”

Try out StackAI's IT Helpdesk Agent.

📱Someone Take His Phone: When someone accused Meta’s chief AI officer Alexandr Wang of having a social team write his posts, he went on an unusually unfiltered posting spree (Exhibit A, Exhibit B) — the kind you probably wouldn’t expect from a senior exec at a $1T+ company. (1M+ views)

📝 Opus in Action: The internet is clamoring over Opus 5.5, especially for its animation and writing skills. Explore 8 projects people have already built using the new model.

✏ a16z University: Education needs a major overhaul in the AI era. A prominent VC firm thinks it has the answer: an AI-first school for gifted youngsters (1.5M views).

🤦 Wasted Usage: A Redditor’s weekly usage limit was about to reset, so he told Claude to “waste the rest” on something ridiculous. The LLM didn’t disappoint.

🏎 AGI Race: Keeping up with the AI industry is tough. This AI-generated Mad Max spoof does a surprisingly good job of summarizing the key players, events, and motivations in 60 seconds, and has racked up over 3M views.

Your next prediction is just one API call away.

TabPFN-3.5 turns messy tabular data into state-of-the-art predictions in seconds. Test more ideas, compare what works, and get state-of-the-art models into production faster.

Try TabPFN today.

"Right now is maybe the most exciting time to build because the barrier to entry is so low", says Tulsee Doshi, Senior Director and Head of Product for Gemini models at Google DeepMind. We sat down live with Doshi to discuss the progress on Gemini 4, how the frontier is changing, and more.

Some of our biggest takeaways:

1. Gemini 4 is in the pipeline. Google has publicly confirmed it's working on Gemini 4, and Doshi says the team is "really excited about what's coming in the next 10 months", giving us some insights into the product experiences currently in the works for Gemini 4.

2. We’re entering the “Pareto” frontier. Raw performance still matters, but the price and speed at which you get intelligence are becoming increasingly important. Google's answer is to offer models at different points on that curve, and Gemini 3.8 Flash "really sits at a sweet spot" for teams watching their token budgets.

3. The "Google is slow" myth doesn't hold up. Doshi notes that Gemini’s last three models launched within 3 weeks of each other, and every iteration came from developer and enterprise feedback — an approach Demis Hassabis calls "relentless shipping."

And more: Doshi also walks us through how she personally uses Gemini, including two products she calls “game changers.”

👉 Click here for the full conversation

👉 Click here to open the full prompt in ChatGPT or Claude

👉 Click here to open the full prompt in ChatGPT (don’t forget to switch to Images!)

Grow customers & revenue: Join companies like Amazon, HubSpot, and Salesforce. Showcase your product to our 2 million+ readers and followers on socials. Get in touch.

Until next time — Zain, Theodore, & the Superhuman AI team.