One AI App Is Replacing Everything

Welcome back apprentices! 👋

Hey {{first_name|friend}},

This week, the biggest names in AI stopped playing nice and started playing for keeps. 

There were surprise product launches, a very awkward board resignation, a co-founder who clearly doesn't believe in delegation, and — somehow — a super-secret AI model that a group chat wasn't supposed to find but absolutely did. 

If you've ever felt like the tech world moves too fast to keep up with, you're not wrong. 

But here's the good news: we read everything, so you don't have to.

_{Read Time: 4 minutes}

👁️ Big Brother Has a Training Deadline. Meta is recording the screens, keystrokes, and mouse clicks of its U.S. employees — no opt-out, no exceptions — to feed its AI models real-world workflow data, and the timing couldn't be more tone-deaf: it kicks off a full month before 8,000 staffers get laid off on May 20. Think of it as robotics labs training robots to walk, except Meta's "robots" are software agents, and the unwilling demo subjects are the people about to lose their jobs. CTO Andrew Bosworth shut down internal pushback fast, framing the surveillance as every employee's chance to help Meta's models "get better simply by doing their daily work" — a sentence that hits very differently when your last day is already calendared.

🚀 Elon's $60B Cheat Code. SpaceX just struck a deal with AI coding darling Cursor — guaranteeing the startup $10B upfront and locking in an option to buy it outright for $60B before year's end. Why? Because Musk's own AI lab, xAI, has been getting lapped by Anthropic and OpenAI in the coding race, and apparently the fastest way to win is to just... buy the winner. SpaceX already poached two of Cursor's top engineers last month, with Musk openly saying the product "wasn't built right" — so this is less a love story and more a hostile hug.

🚪 OpenAI's Exit Interview Just Got Very Busy. In what can only be described as a very expensive game of musical chairs, OpenAI lost its CPO, its Sora lead, and its enterprise apps chief all in a single day — the clearest signal yet that Sam Altman's "we're a platform now, not a startup" pivot is actively reshaping who sticks around. The projects they shepherded? Sora got axed over cost, OpenAI for Science got dissolved into other teams, and the Prism app for scientists quietly got absorbed into Codex — a clean sweep of the "experimental" era. Whether these are departures because of the strategic shake-up or just coincidental timing, losing three senior leaders in 24 hours is definitely a statement.

Most AI tools ask you to trade your data for intelligence. Norton Neo doesn't. It's the first safe AI-native browser built by Norton, and it gives you powerful built-in AI without handing your privacy over to get it. Search, summarize, and write with AI built directly into your browser. Your data stays yours. Your context stays private.

Built-in VPN, anti-fingerprinting, and ad blocking come standard. No add-ons. No setup. No compromises.

Fast. Safe. Intelligent. That's Neo.

Download Norton Neo

If the past week in AI felt like drinking from a firehose, that's because it basically was. 

OpenAI is turning Codex into a full-blown superapp. 

Anthropic just took a swing at the design industry. 

Google's co-founder crawled out of semi-retirement to personally build a strike team aimed squarely at beating Claude at coding. 

Oh, and someone on Discord accidentally stumbled into Anthropic's most restricted AI model — the one that had the White House calling emergency meetings — just days after launch. 

The race to own the entire software stack is no longer a metaphor; it's a Tuesday. Every major player is expanding beyond their original lane at the same time, colliding in the same spaces, and the collateral disruption is landing on designers, researchers, analysts, and developers all at once. 

OpenAI is done pretending Codex is just a coding assistant. 

The latest update transforms it into a unified ChatGPT + Atlas + Codex environment with background computer use, parallel agents, an in-app browser, memory across sessions, and inline image generation via the new gpt-image-1.5. The headline feature — background computer use — lets Codex operate any Mac app independently, even ones without APIs, while multiple agents run simultaneously in the background. It's less "AI tool" and more "AI coworker who doesn't take lunch breaks."

And then there's the image side of the house. 

ChatGPT Images 2.0 doesn't just generate — it thinks first, searching the web for references, planning the output, and checking for errors before delivery. 

Sam Altman described it as the leap from GPT-3 to GPT-5 "all at once," which is either genuine excitement or the most effective marketing sentence of the year. Either way, it landed at No.1 on Arena AI's leaderboard by a clear margin, with 2K resolution, multilingual text rendering, and aspect ratios spanning ultrawide to portrait.

 For a company that spent much of the past year playing catch-up on images, this is a meaningful swing back.

Anthropic's Claude Design is the company's clearest statement yet about where it wants to live in the workflow: everywhere. 

The tool turns prompts, screenshots, and codebases into interactive prototypes, decks, and marketing assets using the new Opus 4.7 vision model. It builds a brand system from your existing work, applies it automatically going forward, and lets you refine output through chat, inline comments, sliders Claude generates itself, or direct edits. When you're done, it hands the work off to Claude Code as a build-ready bundle, or exports to Canva, PPTX, PDF, or HTML.

The Figma angle is hard to ignore. Anthropic CPO Mike Krieger — a Figma board member — resigned three days before launch. Figma built its empire on being the design tool developers actually use. Claude Design is now walking into that exact space, backed by a model that can read your codebase. Whether or not this is a direct competitive move, the timing speaks volumes.

Meanwhile, the company is also dealing with a less flattering story: Mythos, its unreleased cybersecurity model under "Project Glasswing," was reportedly accessed by a Discord group within days of its April 10 launch. 

The group allegedly used leaked credentials from the Mercor breach combined with guessed URL and naming patterns to locate the model online. 

Anthropic considered Mythos too powerful for public release. 

The fact that the first unauthorized users were curious hobbyists rather than nation-state actors is cold comfort — it exposes how thin the security perimeter around even the most restricted models can be as partner access scales.

Google had a strong end to 2025. 2026 has been quieter — and Sergey Brin, apparently, noticed. 

Brin personally assembled a DeepMind "strike team" after internal researchers told him Claude writes better code than Gemini. The team is led by pretraining veteran Sebastian Borgeaud and sits under CTO Koray Kavukcuoglu. In an internal memo, Brin framed the goal simply: build AI that trains the next AI, and coding is the capability that gets you there. 

Engineers are now required to use Google's internal agent tools on complex tasks, with adoption tracked on a company leaderboard called — and this is real — Jetski.

On the product side, Google shipped Deep Research and Deep Research Max, two research agents running on Gemini 3.1 Pro that generate structured reports from web sources, uploaded files, or MCP servers, complete with charts and infographics.

The Max tier benchmarks above Opus 4.6 and GPT 5.4 on retrieval and reasoning, and early enterprise integrations with PitchBook, S&P, and FactSet mean the tool already has a route into the workflows of analysts, consultants, and lawyers. 

The combination of Brin's internal urgency and a research product with genuine enterprise traction suggests Google isn't out of this race — it's just running it from a different angle.

If you use any combination of design tools, coding assistants, research platforms, or image generators professionally — or even semi-professionally — the ground just shifted under your feet, again. 

Claude Design is coming for Figma's audience. Codex is coming for the single-tool-per-task workflow. 

Google Deep Research is coming for the hours analysts spend trawling through reports. 

And ChatGPT Images 2.0 is making the "I'll just hire a designer for this quick mockup" calculation harder to justify. 

None of these tools are perfect replacements — yet — but the direction of travel is unmistakable: the entire creative and technical software stack is being rebuilt around AI orchestration, and the platforms that survive will be the ones that own the most steps between "idea" and "shipped."

Pick one tool from this week's launches — Codex's background agents, Claude Design, or Deep Research — and spend 20 minutes actually breaking it. 

Not using it. 

Breaking it. 

Find its edge, its failure mode, its weird outputs. 

That's the fastest way to understand what it actually replaces versus what it just looks like it replaces — and right now, that distinction is worth more than any benchmark.

Hey, you didn’t get all this info for nothing — share it! If you know someone who’s diving into AI, help them stay in the loop with this week’s updates.

Sharing is a win-win! Send this to a friend who’s all about tech, and you’ll win a little surprise 👀

Speak your prompts into ChatGPT or Claude and get detailed, paste-ready input that actually gives you useful output. Wispr Flow captures what you'd cut when typing. Free on Mac, Windows, and iPhone.

Try Wispr Flow free

A playground for your imagination (and low-key prompt skills).

Each send, we give you a customizable DALL·E prompt inspired by a real-world use case — something that could help you in your business or job if you wanted to use it that way. But it’s also just a fun creative experiment.

You tweak it, run it, and send us your favorite. We pick one winner to feature in the next issue.

Bonus: you’re secretly getting better at prompt design. 🤫

Last week, we challenged you to test GPT-4o’s visual generation skills with this prompt.

Here’s the WINNER:

Congrats to Martin for his creation!🥳

Want to be featured next? Keep those generations coming!

Deep in a mountain carved entirely from mirrored stone and fiber-optic veins lies a vast cylindrical chamber known as the Datafall Observatory. From the ceiling descends a colossal waterfall of pure information—millions of luminous strands cascading like liquid glass, each thread carrying shifting symbols, blueprints, and glowing particles. At the center of the chamber is a raised platform where the falling data converges into a single, ultra-detailed, photorealistic construct of [your creation], forming itself in real time as if the universe is rendering your idea from raw code. The camera frames the scene from a dramatic mid-distance angle, capturing razor-sharp reflections, high-contrast lighting, and the mesmerizing clarity of the data-stream illuminating the dark, vaulted space.

If AI outputs feel inconsistent, it’s usually not the model, it’s missing structure.
We documented the exact 6- Part System we use to get reliable results across ChatGPT, Claude, and Gemini.

It’s a short guide you can finish in under an hour, with plug-and-play prompts + exercises so you actually build the skill and fix the frustrating AI inconsistencies.

Subscriber Price: $10 (normally $19).

_FEEDBACK