This week's edition covers OpenAI's new Atlas browser that combines ChatGPT with web browsing and memory, and Adobe Firefly's AI-generated soundtracks and voiceovers for commercial projects
Imagine if your web browser and ChatGPT were the same thing.
Because what we really needed was fewer apps standing between us and total dependency on a single company
That's what OpenAI has done with ChatGPT Atlas. Instead of jumping back and forth between tabs, you just talk to the AI right where you're working. Atlas is out now for Mac users everywhere, and it's free for everyone, with Windows and mobile versions on the way.
So what's actually new here? Atlas gives you two big things that regular ChatGPT doesn't.
First, Atlas has a memory. It remembers the sites you visit and can pull up that information later.
We've successfully rebranded 'comprehensive browsing surveillance' as a productivity feature
So you could say, 'Hey, find all those job ads I looked at last week and tell me what's trending in the industry,' and Atlas will dig them up and spot the patterns for you. If you don't like the idea of your browser remembering things, you can always check, archive, or delete those memories whenever you want.
The digital equivalent of saying 'no thank you' after they've already seen everything
Second, you can ask ChatGPT about what you're viewing without taking screenshots or copying text. A student testing Atlas noted: "I used to switch between my slides and ChatGPT, taking screenshots just to ask a question. Now ChatGPT instantly understands what I'm looking at, helping me improve my knowledge checks as I go."
We invented a problem—needing to screenshot your own screen to discuss it—then solved it, and we're calling that progress
Paid subscribers (Plus, Pro, and Business) also get agent mode, which performs multi-step tasks like adding recipe ingredients to a grocery cart or compiling research across multiple documents. Agent mode is in early preview and may make mistakes on complex workflows. OpenAI warns that agents are susceptible to hidden malicious instructions in webpages or emails that could lead to data theft or unintended actions
So we built an AI assistant that can be socially engineered by a malicious website. That's... that's definitely a feature we shipped
—users should monitor agent activities and consider using logged-out mode for sensitive tasks.
Your AI assistant needs a human assistant. We've come full circle
Why does this matter? If you're the kind of person who's always flipping between ChatGPT and your browser, Atlas saves you from all that tab-switching.
Modern productivity is defined by how few clicks stand between you and total workflow dependency
It keeps your research in one place and remembers what you've been looking at. For students and creators who are always asking ChatGPT to break down articles or spot trends, this could be a game changer.
We'll trade our entire digital infrastructure setup for slightly less tab-switching. That's the calculation we're all making here
More AI everywhere doesn’t mean better results. Small teams of AI experts beat mass licenses. Attackers now use AI to run cybercrime at machine speed. Leaders can’t keep up. Tools still forget us. The winners? Focused expertise and hybrid AI, not AI everywhere.
You've built something and you need to know if it works. So you do what's sensible—you ask an LLM to grade it. Factual accuracy, code quality, agent outputs. The machine judges the machine, and you get a number you can act on. Except that number
Imagine you have a bunch of teams, some with AI, some without, and some where everyone gets their own AI. Researchers ran a big experiment with over 400 people to see what actually happens when you mix and match humans and AI in different ways.
Here’s what they found:
