AGI? Not Yet... But It’s Starting to Look Real Close 🤖

Plus: Meet o3 and o4-mini — smarter, tool-using models that can “think with images” and code in your terminal.

In partnership with

👋 Good morning, AI adventurer!

Today’s issue feels like a peek into the future — but it’s all happening now. OpenAI dropped its smartest agents yet (one even “thinks” in pictures 👀), Microsoft’s Copilot just learned how to use a computer like a real human, and Claude is becoming the ultimate research assistant.

Grab your ☕, and let’s dive in.

🤯 OpenAI’s New Brains Are Here

o3 and o4-mini debut with tools, vision… and a coding sidekick

OpenAI just turbocharged its brainpower.

The company launched o3, its smartest reasoning model yet, and o4-mini, a leaner model that still crushes benchmarks. These models don’t just talk — they act. Equipped with full tool access, from web search to code execution, they now operate like agents, not chatbots.

🧠 The big leap? They “think with images,” integrating visuals into reasoning. Imagine solving physics problems, debugging visual layouts, or analyzing images — all as part of a single flow of thought.

💻 Also released: Codex CLI, an open-source agent for your terminal that bridges natural language reasoning and hardcore code execution.

💬 Greg Brockman called it “a GPT-4 level qualitative step into the future.”

🔍 Why it matters: These aren't just smarter models — they’re tools-savvy, visually capable agents that generate new ideas and reason across domains. It’s a whole new step in OpenAI’s AGI playbook.

🖥️ Copilot Gets Physical

Microsoft teaches AI how to use your computer (literally)

Copilot Studio just leveled up.

Microsoft rolled out ‘computer use’ capabilities, allowing Copilot to interact with websites and desktop apps — clicking buttons, typing in fields, and selecting menus just like you would.

🔧 This means Copilot can now automate legacy software and clunky workflows that don’t have APIs. It adapts in real-time to UI changes and keeps running smoothly thanks to built-in reasoning.

🛡️ All this happens on Microsoft’s own infrastructure, and your data won’t be used for training.

🔍 Why it matters: This is a serious move into agentic AI. Unlike browser-bound bots, Copilot can now reach into your software stack — a win for enterprises trying to streamline operations without rebuilding legacy systems.

You’ve heard the hype. It’s time for results.

For all the buzz around agentic AI, most companies still aren't seeing results. But that's about to change. See real agentic workflows in action, hear success stories from our beta testers, and learn how to align your IT and business teams.

🔍 Claude Goes Research Mode

Anthropic’s assistant just got way smarter — and more connected

Claude’s no longer just a clever assistant — it’s becoming a full-fledged researcher.

Anthropic introduced autonomous research powers to Claude, allowing it to search the web and your own files to build richer, context-aware answers — with citations, of course.

🧾 There’s also a new Google Workspace integration, giving Claude secure access to emails, calendars, and Docs (no more copying and pasting). Enterprise users get advanced document cataloging using retrieval-augmented generation (RAG) for deep knowledge search.

🗺️ Launching in the US, Japan, and Brazil, the new tools are available to Max, Team, and Enterprise users.

🔍 Why it matters: Claude’s catching up fast. With a combo of SOTA models, smart research workflows, and workplace integration, it’s entering the AI assistant big leagues — and it’s bringing receipts.

🗞️ Other News

Here’s what else is bubbling in the AI universe:

🌀 OpenAI may acquire Windsurf (formerly Codeium) in a ~$3B deal — a big bet on owning the developer toolchain.
🧠 Microsoft drops BitNet b1.58 2B4T, a 1-bit model that runs efficiently on CPUs and matches larger models in performance.
🖌️ Tencent unveils FireEdit, a precise image editing AI powered by region-aware vision-language models.
🎙️ Claude’s getting voice mode this month — with voices called Airy, Mellow, and Buttery. Smooth.
📊 Metr's analysis of o3 and o4-mini suggests OpenAI may have fast-tracked safety evals to hit release targets.
🚨 Economist Tyler Cowen says o3 = AGI, asking if April 16 was the day the line was crossed.
🧩 Grok Studio, a new feature from OpenAI, turns chat into a creative canvas for editing docs and apps.

💡What else are we reading and seeing?

  • Future Chips Will Be Hotter Than Ever

  • The business of the AI labs

  • o3 is out and it is absolutely amazing!!

  • Inside Mark Zuckerberg's Failed Negotiations to End Antitrust Case

  • U.S. Tries to Crush China's AI Ambitions With Chips Crackdown

  • Elon Musk shut down internal Tesla analysis that showed Robotaxi would lose money

  • Lab-grown chicken ‘nuggets' hailed as ‘transformative step' for cultured meat

Find out why 1M+ professionals read Superhuman AI daily.

In 2 years you will be working for AI

Or an AI will be working for you

Here's how you can future-proof yourself:

  1. Join the Superhuman AI newsletter – read by 1M+ people at top companies

  2. Master AI tools, tutorials, and news in just 3 minutes a day

  3. Become 10X more productive using AI

Join 1,000,000+ pros at companies like Google, Meta, and Amazon that are using AI to get ahead.

🧠 That’s a Wrap!

We’re watching AI sprint into agent land — thinking with images, clicking your desktop, reading your calendar, and maybe… becoming AGI?

Stay curious, stay caffeinated, and we’ll see you tomorrow for more breakthroughs.

Thanks for reading!
The Not So Artificial Team

☕️ Got a friend who loves AI? Forward this and let them know where the good stuff lives.