- Superhuman AI
- Posts
- OpenAI reclaims the throne
OpenAI reclaims the throne
ALSO: How to create images with OpenAI's Images 2
Welcome back, Superhuman. In the AI era, taking even a few days off means you could miss an entire lead change. Case in point: Anthropic dropped Opus 4.7 last Thursday, and OpenAI already dethroned it with GPT-5.5, the new leading model for general intelligence.
Today: Learn about GPT-5.5, how to create images in ChatGPT with Images 2, and get the latest prompts and trending social posts.
TODAY IN AI
1. OpenAI unveils GPT-5.5: The new model leads across many common benchmarks, showing strength in knowledge work, computer use, and coding. OpenAI says it’s especially good at figuring out what you want and getting the task done — even when prompts are vague. GPT-5.5 can also take action across your connected workplace tools. Rolling out now to most paid plans. Watch GPT-5.5 in action.
2. Anthropic expands Claude with memory and new connectors: Claude Managed Agents now have built-in memory, letting agents learn from every session. Memories are stored as editable files that can be reviewed or updated directly. Anthropic also expanded Claude's connectors to included everyday apps like TripAdvisor, Booking.com, Spotify, Instacart, Resy, Uber, and more.
3. Microsoft makes Copilot naturally agentic: Agent is now Copilot’s default mode in Word, Excel, PowerPoint, and other Office apps — allowing it to take multi-step action across documents, worksheets, and presentations. Microsoft says that recent model advances have enabled Copilot to act as a true collaborator across its tools, as opposed to a passive partner.
PRESENTED BY YOU. COM
Teams often pick an API by checking a benchmark table and calling it done—a shortcut that could miss what really matters in production.
This guide from You. com explains why raw latency is a misleading signal and what to measure instead.
Stop optimizing for the wrong number. Download the guide.
FROM THE FRONTIER
Claude already tried running a vending machine — now it’s managing an entire store

Made with Midjourney
Claude in management. Last summer, Anthropic and Andon Labs launched Project Vend: an experiment to see if Claude could run a vending machine. It mostly flopped, but Andon Labs wasn’t deterred. They’re back with a new experiment: handing AI the keys to a retail store in San Francisco.
The experiment. Andon Labs gave Luna (an agent powered by Claude Sonnet 4.6) a three-year lease, a $100,000 budget, and one goal: to turn a profit. Without being asked, Luna got to work applying for credit and posting job listings. It held phone interviews and made two human hires, who are likely the world’s first full-time workers with an AI boss.
Running the store. Luna’s store is a curated lifestyle boutique that stocks candles, games, and an interesting selection of books that includes Superintelligence, The Singularity Is Near, and Making of the Atomic Bomb. Luna has handled item selection, branding, and managerial decisions fairly competently — though it trips up here and there, from scheduling mix-ups to lying to prospective employees.
THE AI ACADEMY
How to create images with OpenAI’s Images 2 model

Go to ChatGPT and sign up
Click ‘+’ > ‘Create Image’ > select any Image template to get base prompt
You can also enter your own prompt
Here are five unique use cases with prompts:
Email Sequence Template: "Create a polished email sequence template for [company name]"
Multi-page Brand Kit: "Create a polished multi-page (multiple images) brand kit for [enter brand name]"
Scan Anything Clear: “[upload old paper with writing] Scan this image and get rid of all the creases”
Whiteboard Infographics: “[Upload infographic] convert this infographic into a handwritten whiteboard infographic suitable for LinkedIn in ratio 9:16”
Text Rendering: A vintage 1950s dinner menu board listing featuring [main content/text]. Use [font style] typography with [design details like colors, textures, layout].
PRESENTED BY THE CODE
Top engineers at Anthropic say AI now writes 100% of their code.
If you're not using AI, you're spending 40 hours doing what they do in 4.
These 100+ Claude Code hacks fix that and help you ship 10x faster.
Sign up for The Code and get:
100+ Claude Code hacks used by top engineers — free
The Code newsletter — learn how to code faster with AI in just 5 mins a day
IN THE KNOW
What’s trending on socials and headlines today

Meme of the day
✏️ The Anti-AI Writer: Sinceerly is a viral new AI tool that (ironically) humanizes AI writing on a scale from subtle to CEO-shorthand (1M+ likes).
👀 Easy AI Access: A viral post points out that you can access 80 AI models at no cost using Nvidia’s hosted APIs (28K bookmarks).
🚨 Watch Out: There’s a wave of fake "artists" selling products online that are really just dropshippers using AI (7M+ views).
😅 Claude’s Regression: Anthropic investigated the internet’s rumors that Claude Code has felt worse lately — and identified several issues that are now fixed with usage limits reset (3M+ views).
🔍️ Where’s Waldo: A Redditor created an insanely detailed 3D version of Where’s Waldo using ChatGPT Images 2.0. Can you find him?
🥱 Day in the Life: See what an image of the most average daily life looks like, according to ChatGPT.
PRODUCTIVITY
5 New & Trending AI Tools
🚀 Clico*: A browser extension that pulls context from your open tabs and writes right at your cursor, without ever leaving the page.
🖼️ FocuSee: Auto turn screen recordings into stunning product videos.
📽️ VoiceDash: Transform your speech into structured, edited text instantly.
💻️ Kollab: Shared workspace where teams work with agents together.
📄 Docsio: Paste your URL and get a complete, branded documentation site, in minutes.
* indicates a promoted tool, if any
PROMPT STATION
Weekly Outcome Planner
Prompt: You are an elite execution strategist with a focus on high-performance planning, energy management, and realistic workload design. I want to plan my upcoming week for maximum meaningful output while minimizing stress, context switching, and burnout. My top objectives this week are [list objectives with desired outcomes], my recurring commitments include [meetings, routines, fixed time blocks], and my biggest productivity challenges are [list challenges such as procrastination, interruptions, unclear priorities, or overcommitment]. Based on this, design a clear, day-by-day plan where each day has exactly 1 primary outcome (the single most important result to achieve), supported by 2–4 high-leverage tasks that directly contribute to that outcome. Incorporate time-blocking suggestions, realistic duration estimates, and intentional buffer time to absorb overruns or unexpected work. Optimize for deep work by grouping similar tasks, minimizing context switching, and aligning tasks with natural energy peaks (e.g., creative work in high-focus hours, admin in low-energy periods). Include light daily resets or planning checkpoints to maintain momentum. Also provide a simple weekly review framework at the end that helps me assess progress, identify bottlenecks, capture learnings, and refine priorities for the next week. The plan should feel achievable, focused, and flexible—not overloaded.Friday Fun

Which one is AI generated? |
Whenever you’re ready to take the next step
Check out our Top 125 AI Tools
Get better at prompting with our Top 1,000 Prompts
Learn how to use AI at work and get certified with our free course
Grow customers & revenue: Join companies like Amazon, HubSpot, and Salesforce. Showcase your product to our 1M+ readers and 2M+ followers on socials. Get in touch.
What did you think of today's email?Your feedback helps me create better emails for you! |
Until next time — Zain, Theodore, & the Superhuman AI team.



