Superhuman AI logoSuperhuman AIZain Kahn

Fable 5.2 is here and "almost" AGI, says Wharton professor

Tuesday, November 10, 2026 · 6 min read

Welcome back, Superhuman. Anthropic may have finally topped Opus 5.5. The lab just released Fable 5.2, which a Wharton professor called “almost” AGI (AI smarter than human experts). Even better, there’s no cybersecurity delay. It’s available now, along with fresh usage and new prompting strategies.

Today: Google challenges Nvidia, how to turn your data into visualizations and interactive dashboards, and get the latest prompts and trending social posts.

TODAY IN AI

A Wharton professor asked Fable 5.2 to create the most impressive thing possible. See what it created, and read his full review of Anthropic’s latest model.

1. Anthropic debuts Fable and Mythos 5.2 as it nears IPO: The duo share the same underlying model, with Fable 5.2 adding extra protective guardrails. Anthropic says Fable 5.2 is best for your most ambitious projects, works up to 30% faster than before, and is a “step above” Opus 5.5. It’s live now across paid Claude plans, along with extra usage and fresh prompting strategies. Early reviews:

  • Wharton professor Ethan Mollick asked Fable 5.2 to build the most impressive thing it could for his blog. It delivered a thought-provoking 90-second video, complete with a custom soundtrack and twist ending. His verdict: “almost” AGI.

  • Entrepreneur Greg Isenberg says Fable 5.2 tells you when it isn’t sure of something, instead of guessing. He shared 8 other observations.

  • Tech creator Riley Brown used it to create a one-minute animated movie based on one prompt.

2. Google unveils two new AI chips as it ramps up competition with Nvidia: The tech giant introduced two new versions of its eighth-generation Tensor Processing Units (TPUs), the in-house chips that train and run Gemini. Both are expected to ship in 2027, and Bloomberg Intelligence projects TPU sales will climb from $12B this year to more than $135B by 2028. The announcement added even more pressure to Nvidia’s stock, which just hit a 7-month low amid the ongoing investigation into CEO Jensen Huang.

Stop burning time figuring out multi-agent orchestration. The new Startup Guide to AI Agents is a code-first blueprint thatgets your agents into production faster on Google Cloud using the Agent Development Kit (ADK). Inside, we break down how to:

  • Move from basic prompting to scalable context engineering

  • Optimize token costs to extend your startup's runway

  • Lock in production reliability with AgentOps 

TUTORIAL

How to turn your data into visualizations and interactive dashboards with Claude

You no longer need to learn SQL, Python, Tableau, or hire a dedicated developer or data analyst to build interactive charts and dashboards:

  • Connect a spreadsheet or Google Drive to Claude

  • Structure a strong four-part dashboard prompt

  • Add live filters, interactive charts, and hover tooltips

  • Expand, share, and export the finished dashboard

Want more AI skills you can put to work? Check out The AI Academy

FROM THE FRONTIER

AI overuse could be creating a “speed trap” at your company. Here’s how to avoid it.

Made with Midjourney

A “speed trap” is when AI churns out work faster than anyone can understand it. The productivity boost feels great at first. But some leaders say it leads to more confusion down the road.

At Every’s Thesis: 2027 conference, Notion CEO Ivan Zhao warned that a codebase can stop making sense within just one week of engineers looping in their own coding agents. A viral Reddit post claims this exact scenario played out at their company: their codebase tripled in size, but now no one understands it. It’s not just engineers, either.

Steve Huynh, a former Amazon principal engineer, argued that reliance on AI is threatening the company’s writing culture, potentially weakening decision-making.

How can you avoid speed traps?

  1. Keep learning in the loop: Anthropic recommends designing workflows so employees keep learning and retain the skills necessary to fact-check AI’s outputs.

  2. Always add your own perspective: Every CEO Dan Shipper argues anyone can get a default AI answer. The real value is the perspective you add on top of it, and that only comes from your understanding of the work itself.

  3. Make authors defend their work: Mathematician Terence Tao says if authors can’t give a clear, expert-level talk on an AI-assisted result, it shouldn’t be published. Consider holding your team’s memos, decks, and code to the same bar.

Getting AI to answer customers is easy now.

Getting it to answer with your best rep’s judgment is the hard part: knowing when to follow the process, bend a rule, or pass it up the hierarchy with full context.

Trusted by Amex, Kia, American Eagle, and more.

IN THE KNOW

What’s trending on socials & headlines today

Meme of the day

👀Caught on Camera: An internet sleuth says they’ve found photos of Jensen Huang and his mistress. With close to 7M views, the photos are quickly becoming internet infamous, although their authenticity is still unverified.

📝 Fable’s Talents: Before releasing its new models, Anthropic put Fable & Mythos 5.2 to work on a variety of ambitious projects. The thread quickly went viral. See everything they’re capable of.

🧠Jane’s Skill: Jane Street released an AI skill that collected 15K bookmarks — and it has nothing to do with finance. The skill turns your LLM into a Socratic Tutor that teaches you any topic using a proven 5-step formula.

✏ Test Prompt: Whenever a new model drops, Anthropic’s Alex Albert always runs this simple prompt through his projects. He recommends giving it a try.

🍪Burning Cookies: Do you ever feel like AI gaslights you? A viral skit captures that feeling through a simple conversation about burnt cookies (1M views).

PRODUCTIVITY
  • 🌟Section*:Build your agentic AI strategy with practical insights on deployment, governance, and ROI. Join the summit.

  • 📽Higgsfield Miracle: Use AI to edit any part of a video without changing the rest.

  • 🛠️ Floot: Publish and build real apps straight in your chatbot.

  • 💥Spira: Grow your brand, reach more people, and gain more followers with AI.

  • 🖥️ The Code: A 5-minute brief to help developers stay on the cutting edge.

PROMPT STATION

Assess revenue quality with operational context

Prompt: Act as a senior finance and operating analyst. Assess the quality of a company’s revenue by combining financial performance with the operational context behind that .............

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

IMAGE

Create a visual intro card for a new team member

Prompt: [upload reference photo] Create a polished visual intro card for a new team member using the uploaded photo of [NAME] as the identity reference. Preserve their recognizable ............

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

EXTRAS

Want more?

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.

What did you think of today's email?

Your feedback helps me create better emails for you!

Login or Subscribe to participate

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

More from Superhuman AI

All issues →
Nov 9, 2026 · 6 min read

ALSO: How to convert your product photos into high-converting video ads

Nov 8, 2026 · 6 min read

Keep up with the latest AI news, trends, and tools in just 3 minutes a day. Join 1,500,000+ professionals.

Nov 7, 2026 · 5 min read

Keep up with the latest AI news, trends, and tools in just 3 minutes a day. Join 1,500,000+ professionals.

Nov 6, 2026 · 6 min read

ALSO: How to build a working mobile app with Claude

Nov 5, 2026 · 6 min read

ALSO: How to build an on-demand data analyst for your team with Grok Bot